Merge c8b3d54cf8 into 7f287b705e

Signed-off-by: bigcat88 <bigcat88@icloud.com>

.ci/windows_amd_base_files/README_VERY_IMPORTANT.txt

@@ -1,5 +1,4 @@
-As of the time of writing this you need this driver for best results:
-https://www.amd.com/en/resources/support-articles/release-notes/RN-AMDGPU-WINDOWS-PYTORCH-7-1-1.html
+As of the time of writing this you need a recent driver. Updating to the latest driver is recommended.
 
 HOW TO RUN:
 
@@ -7,9 +6,9 @@ If you have a AMD gpu:
 
 run_amd_gpu.bat
 
-If you have memory issues you can try disabling the smart memory management by running comfyui with:
+If you have memory issues you can try enabling the new dynamic memory management by running comfyui with:
 
-run_amd_gpu_disable_smart_memory.bat
+run_amd_gpu_enable_dynamic_vram.bat
 
 IF YOU GET A RED ERROR IN THE UI MAKE SURE YOU HAVE A MODEL/CHECKPOINT IN: ComfyUI\models\checkpoints
 

.coderabbit.yaml

@@ -4,12 +4,12 @@ early_access: false
 tone_instructions: "Only comment on issues introduced by this PR's changes. Do not flag pre-existing problems in moved, re-indented, or reformatted code."
 
 reviews:
-  profile: "chill"
-  request_changes_workflow: false
+  profile: "assertive"
+  request_changes_workflow: true
   high_level_summary: false
   poem: false
   review_status: false
-  review_details: false
+  review_details: true
   commit_status: true
   collapse_walkthrough: true
   changed_files_summary: false
@@ -39,6 +39,14 @@ reviews:
     - path: "**"
       instructions: |
         IMPORTANT: Only comment on issues directly introduced by this PR's code changes.
+        Treat AGENTS.md as mandatory repository policy, not optional style guidance.
+        Flag PR changes that violate AGENTS.md even when the code is otherwise functional.
+        In particular, enforce architecture boundaries, dtype/device/memory rules,
+        interface contracts, import style, no unnecessary try/except blocks, no inline
+        imports, no outbound internet paths in core ComfyUI, and narrow scoped fixes.
+        Prefer direct findings over suggestions when a rule is violated. Only ignore
+        AGENTS.md when it clearly conflicts with a newer explicit maintainer instruction
+        in the PR.
         Do NOT flag pre-existing issues in code that was merely moved, re-indented,
         de-indented, or reformatted without logic changes. If code appears in the diff
         only due to whitespace or structural reformatting (e.g., removing a `with:` block),
@@ -123,5 +131,10 @@ chat:
 
 knowledge_base:
   opt_out: false
+  code_guidelines:
+    enabled: true
+    filePatterns:
+      - files: "AGENTS.md"
+        applyTo: "**"
   learnings:
     scope: "auto"

.github/workflows/backport_release.yaml

@@ -0,0 +1,519 @@
+name: Backport Release
+
+on:
+  workflow_dispatch:
+    inputs:
+      commit:
+        description: 'Full 40-char SHA of the tip commit of the backport source branch (the PR head commit that passed tests). The branch is resolved from this SHA and must be unique.'
+        required: true
+        type: string
+
+permissions:
+  contents: read
+  pull-requests: read
+  checks: read
+
+jobs:
+  backport-release:
+    name: Create backport release
+    runs-on: ubuntu-latest
+    environment: backport release
+
+    steps:
+      - name: Generate GitHub App token
+        id: app-token
+        uses: actions/create-github-app-token@bcd2ba49218906704ab6c1aa796996da409d3eb1
+        with:
+          app-id: ${{ secrets.FEN_RELEASE_APP_ID }}
+          private-key: ${{ secrets.FEN_RELEASE_PRIVATE_KEY }}
+
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd
+        with:
+          token: ${{ steps.app-token.outputs.token }}
+          fetch-depth: 0
+          fetch-tags: true
+
+      - name: Configure git
+        run: |
+          git config user.name  "fen-release[bot]"
+          git config user.email "fen-release[bot]@users.noreply.github.com"
+
+      - name: Resolve source branch from commit SHA
+        id: resolve
+        env:
+          SOURCE_COMMIT:  ${{ inputs.commit }}
+          DEFAULT_BRANCH: ${{ github.event.repository.default_branch }}
+        run: |
+          set -euo pipefail
+
+          # Require a full 40-char lowercase-hex SHA. Short SHAs are ambiguous
+          # and we will be comparing this value against API responses (PR head
+          # SHA, ref tips) that always return the full form.
+          if [[ ! "${SOURCE_COMMIT}" =~ ^[0-9a-f]{40}$ ]]; then
+            echo "::error::Input commit '${SOURCE_COMMIT}' is not a full 40-char lowercase hex SHA."
+            exit 1
+          fi
+
+          # Fetch all remote branches so we can search for which one(s) point
+          # at this SHA. `actions/checkout` with fetch-depth: 0 fetches full
+          # history of the checked-out ref but does not necessarily populate
+          # every refs/remotes/origin/*, so do it explicitly.
+          git fetch --prune origin '+refs/heads/*:refs/remotes/origin/*'
+
+          # Verify the commit actually exists in this repo's object DB.
+          if ! git cat-file -e "${SOURCE_COMMIT}^{commit}" 2>/dev/null; then
+            echo "::error::Commit ${SOURCE_COMMIT} was not found in the repository."
+            exit 1
+          fi
+
+          # Find every remote branch whose tip == SOURCE_COMMIT. Exactly one
+          # branch must point at it. If zero, the commit isn't anyone's tip
+          # (likely stale, force-pushed past, or never the PR head). If more
+          # than one, the (branch -> SHA) mapping is ambiguous and we refuse
+          # to guess — the operator must give us a unique branch to release.
+          mapfile -t matching_branches < <(
+            git for-each-ref \
+              --format='%(refname:strip=3)' \
+              --points-at="${SOURCE_COMMIT}" \
+              refs/remotes/origin/ \
+              | grep -vx 'HEAD' || true
+          )
+
+          if [[ "${#matching_branches[@]}" -eq 0 ]]; then
+            echo "::error::No branch on origin has ${SOURCE_COMMIT} as its tip."
+            echo "::error::Either the branch was updated after you copied this SHA, or this commit was never the head of a branch."
+            exit 1
+          fi
+
+          if [[ "${#matching_branches[@]}" -gt 1 ]]; then
+            echo "::error::More than one branch on origin has ${SOURCE_COMMIT} as its tip; cannot pick one:"
+            for b in "${matching_branches[@]}"; do
+              echo "::error::  - ${b}"
+            done
+            echo "::error::Refusing to proceed with an ambiguous source branch."
+            exit 1
+          fi
+
+          source_branch="${matching_branches[0]}"
+
+          if [[ "${source_branch}" == "${DEFAULT_BRANCH}" ]]; then
+            echo "::error::Source branch must not be the default branch ('${DEFAULT_BRANCH}')."
+            exit 1
+          fi
+
+          echo "Resolved commit ${SOURCE_COMMIT} to branch '${source_branch}'."
+          echo "source_branch=${source_branch}" >> "$GITHUB_OUTPUT"
+
+      - name: Determine latest stable release
+        id: latest
+        env:
+          GH_TOKEN: ${{ steps.app-token.outputs.token }}
+        run: |
+          set -euo pipefail
+
+          # List all tags matching vMAJOR.MINOR.PATCH and pick the highest by numeric
+          # comparison of each component. We DO NOT use `sort -V` because it treats
+          # v0.19.99 as higher than v0.20.1.
+          latest_tag="$(
+            git tag --list 'v[0-9]*.[0-9]*.[0-9]*' \
+              | grep -E '^v[0-9]+\.[0-9]+\.[0-9]+$' \
+              | awk -F'[v.]' '{ printf "%010d %010d %010d %s\n", $2, $3, $4, $0 }' \
+              | sort -k1,1n -k2,2n -k3,3n \
+              | tail -n1 \
+              | awk '{print $4}'
+          )"
+
+          if [[ -z "${latest_tag}" ]]; then
+            echo "::error::No stable release tags (vMAJOR.MINOR.PATCH) were found."
+            exit 1
+          fi
+
+          # Parse components
+          ver="${latest_tag#v}"
+          major="${ver%%.*}"
+          rest="${ver#*.}"
+          minor="${rest%%.*}"
+          patch="${rest#*.}"
+
+          new_patch=$((patch + 1))
+          new_version="v${major}.${minor}.${new_patch}"
+          release_branch="release/v${major}.${minor}"
+
+          latest_sha="$(git rev-list -n 1 "refs/tags/${latest_tag}")"
+
+          echo "latest_tag=${latest_tag}"             >> "$GITHUB_OUTPUT"
+          echo "latest_sha=${latest_sha}"             >> "$GITHUB_OUTPUT"
+          echo "major=${major}"                       >> "$GITHUB_OUTPUT"
+          echo "minor=${minor}"                       >> "$GITHUB_OUTPUT"
+          echo "patch=${patch}"                       >> "$GITHUB_OUTPUT"
+          echo "new_version=${new_version}"           >> "$GITHUB_OUTPUT"
+          echo "new_version_no_v=${major}.${minor}.${new_patch}" >> "$GITHUB_OUTPUT"
+          echo "release_branch=${release_branch}"     >> "$GITHUB_OUTPUT"
+
+          echo "Latest stable release: ${latest_tag} (${latest_sha})"
+          echo "New version will be:   ${new_version}"
+          echo "Release branch:        ${release_branch}"
+
+      - name: Validate source branch is cut directly from the latest stable release
+        env:
+          SOURCE_BRANCH:   ${{ steps.resolve.outputs.source_branch }}
+          SOURCE_COMMIT:   ${{ inputs.commit }}
+          LATEST_TAG_SHA:  ${{ steps.latest.outputs.latest_sha }}
+          LATEST_TAG:      ${{ steps.latest.outputs.latest_tag }}
+        run: |
+          set -euo pipefail
+
+          # Use the user-provided SHA directly rather than re-resolving the branch
+          # tip — the resolve step already proved the branch tip equals SOURCE_COMMIT,
+          # and pinning to the SHA here makes the rest of the job TOCTOU-safe against
+          # someone pushing to the branch mid-run.
+          source_sha="${SOURCE_COMMIT}"
+
+          # Walking first-parent from the source tip must reach LATEST_TAG_SHA.
+          # We capture rev-list into a variable and grep against a here-string
+          # rather than piping `rev-list | grep -q`: under `set -o pipefail`,
+          # `grep -q` would exit on first match and SIGPIPE the still-streaming
+          # `rev-list`, propagating exit 141 as a spurious "not found".
+          first_parent_chain="$(git rev-list --first-parent "${source_sha}")"
+          if ! grep -Fxq "${LATEST_TAG_SHA}" <<< "${first_parent_chain}"; then
+            echo "::error::Source branch '${SOURCE_BRANCH}' is not cut from '${LATEST_TAG}'."
+            echo "::error::Its first-parent history does not include ${LATEST_TAG_SHA}."
+            exit 1
+          fi
+
+          # Additionally, every commit added on top of the tag (the set we are
+          # about to publish) must itself be a descendant of the tag along
+          # first-parent — i.e. no sibling commits from master sneak in via a
+          # non-first-parent path. Enforce by requiring that the symmetric
+          # difference is empty in one direction: commits in source that are
+          # NOT first-parent-reachable from source starting at the tag.
+          # We do this by intersecting:
+          #   A = commits reachable from source but not from tag (full DAG)
+          #   B = commits on the first-parent chain from source down to tag
+          # and requiring A == B.
+          all_added="$(git rev-list "${LATEST_TAG_SHA}..${source_sha}" | sort)"
+          first_parent_added="$(
+            git rev-list --first-parent "${LATEST_TAG_SHA}..${source_sha}" | sort
+          )"
+
+          if [[ "${all_added}" != "${first_parent_added}" ]]; then
+            echo "::error::Source branch '${SOURCE_BRANCH}' contains commits not on its first-parent chain from '${LATEST_TAG}'."
+            echo "::error::This usually means the branch was cut from master (not from the tag) or contains a merge from master."
+            echo "Commits reachable but not on first-parent chain:"
+            comm -23 <(printf '%s\n' "${all_added}") <(printf '%s\n' "${first_parent_added}") \
+              | while read -r sha; do
+                  echo "  $(git log -1 --format='%h %s' "${sha}")"
+                done
+            exit 1
+          fi
+
+          added_count="$(printf '%s\n' "${all_added}" | grep -c . || true)"
+          echo "Source branch is cut directly from ${LATEST_TAG} with ${added_count} commit(s) on top."
+
+      - name: Validate PR exists, is open, named correctly, has latest commit, and checks pass
+        env:
+          GH_TOKEN:      ${{ steps.app-token.outputs.token }}
+          SOURCE_BRANCH: ${{ steps.resolve.outputs.source_branch }}
+          SOURCE_COMMIT: ${{ inputs.commit }}
+          NEW_VERSION:   ${{ steps.latest.outputs.new_version }}
+          REPO:          ${{ github.repository }}
+        run: |
+          set -euo pipefail
+
+          expected_title="ComfyUI backport release ${NEW_VERSION}"
+
+          # Find open PRs from this branch into master. The --state open filter
+          # is load-bearing: a closed/merged PR with passing checks must not be
+          # accepted as authorization for a new release.
+          pr_json="$(
+            gh pr list \
+              --repo "${REPO}" \
+              --state open \
+              --head "${SOURCE_BRANCH}" \
+              --base master \
+              --json number,title,headRefOid,state \
+              --limit 10
+          )"
+
+          pr_count="$(echo "${pr_json}" | jq 'length')"
+          if [[ "${pr_count}" -eq 0 ]]; then
+            echo "::error::No open PR found from '${SOURCE_BRANCH}' into 'master'. The PR must exist and be open."
+            exit 1
+          fi
+
+          # Pick the PR matching the expected title
+          pr_number="$(echo "${pr_json}" | jq -r --arg t "${expected_title}" '
+            map(select(.title == $t)) | .[0].number // empty
+          ')"
+          pr_head_sha="$(echo "${pr_json}" | jq -r --arg t "${expected_title}" '
+            map(select(.title == $t)) | .[0].headRefOid // empty
+          ')"
+
+          if [[ -z "${pr_number}" ]]; then
+            echo "::error::No open PR from '${SOURCE_BRANCH}' into 'master' is titled '${expected_title}'."
+            echo "Found PRs:"
+            echo "${pr_json}" | jq -r '.[] | "  #\(.number): \(.title)"'
+            exit 1
+          fi
+
+          # The PR's current head commit must equal the SHA the operator gave us.
+          # This is what closes the door on releasing stale code: if anyone has
+          # pushed to the branch since the operator validated tests passed, the
+          # PR head will have advanced past SOURCE_COMMIT and we abort. (The
+          # resolve step already proved the branch tip == SOURCE_COMMIT; this
+          # ties that same SHA to the PR that authorizes the release.)
+          if [[ "${pr_head_sha}" != "${SOURCE_COMMIT}" ]]; then
+            echo "::error::PR #${pr_number} head commit is ${pr_head_sha}, but the operator-provided commit is ${SOURCE_COMMIT}."
+            echo "::error::The PR has new commits since this release was authorized. Re-run with the new head SHA after verifying its checks."
+            exit 1
+          fi
+
+          echo "Found open PR #${pr_number} titled '${expected_title}' at head ${pr_head_sha} (matches operator-provided commit)."
+
+          # Verify all check runs on the head commit have completed successfully.
+          # A check is considered passing if conclusion is success, neutral, or skipped.
+          checks_json="$(
+            gh api \
+              --paginate \
+              "repos/${REPO}/commits/${pr_head_sha}/check-runs" \
+              --jq '.check_runs[] | {name: .name, status: .status, conclusion: .conclusion}'
+          )"
+
+          if [[ -z "${checks_json}" ]]; then
+            echo "::error::No check runs found on PR head commit ${pr_head_sha}."
+            exit 1
+          fi
+
+          echo "Check runs on ${pr_head_sha}:"
+          echo "${checks_json}" | jq -s '.'
+
+          failing="$(echo "${checks_json}" | jq -s '
+            map(select(
+              .status != "completed"
+              or (.conclusion as $c
+                  | ["success","neutral","skipped"]
+                  | index($c) | not)
+            ))
+          ')"
+
+          failing_count="$(echo "${failing}" | jq 'length')"
+          if [[ "${failing_count}" -gt 0 ]]; then
+            echo "::error::One or more checks have not passed on PR head commit ${pr_head_sha}:"
+            echo "${failing}" | jq -r '.[] | "  - \(.name): status=\(.status) conclusion=\(.conclusion)"'
+            exit 1
+          fi
+
+          echo "All checks have passed on ${pr_head_sha}."
+
+      - name: Prepare release branch
+        id: prepare
+        env:
+          GH_TOKEN:        ${{ steps.app-token.outputs.token }}
+          REPO:            ${{ github.repository }}
+          RELEASE_BRANCH:  ${{ steps.latest.outputs.release_branch }}
+          LATEST_TAG:      ${{ steps.latest.outputs.latest_tag }}
+          LATEST_TAG_SHA:  ${{ steps.latest.outputs.latest_sha }}
+          PATCH:           ${{ steps.latest.outputs.patch }}
+        run: |
+          set -euo pipefail
+
+          # Try to fetch the release branch. If patch == 0, it shouldn't exist yet
+          # and we'll create it from the latest stable tag. If patch > 0, it must
+          # already exist and its tip must equal the latest stable tag commit (i.e.
+          # the previous patch release).
+          if git ls-remote --exit-code --heads origin "${RELEASE_BRANCH}" >/dev/null 2>&1; then
+            echo "Release branch '${RELEASE_BRANCH}' already exists on origin."
+            git fetch origin "refs/heads/${RELEASE_BRANCH}:refs/remotes/origin/${RELEASE_BRANCH}"
+            git checkout -B "${RELEASE_BRANCH}" "refs/remotes/origin/${RELEASE_BRANCH}"
+
+            current_tip="$(git rev-parse HEAD)"
+            if [[ "${current_tip}" != "${LATEST_TAG_SHA}" ]]; then
+              echo "::error::Release branch '${RELEASE_BRANCH}' tip (${current_tip}) is not at the latest stable release '${LATEST_TAG}' (${LATEST_TAG_SHA})."
+              echo "::error::Refusing to release on top of a divergent branch."
+              exit 1
+            fi
+            echo "branch_existed=true" >> "$GITHUB_OUTPUT"
+          else
+            if [[ "${PATCH}" != "0" ]]; then
+              echo "::error::Release branch '${RELEASE_BRANCH}' does not exist on origin, but the latest stable release '${LATEST_TAG}' has patch=${PATCH} (>0). This is inconsistent."
+              exit 1
+            fi
+            echo "Release branch '${RELEASE_BRANCH}' does not exist. Creating from ${LATEST_TAG}."
+            git checkout -B "${RELEASE_BRANCH}" "refs/tags/${LATEST_TAG}"
+            echo "branch_existed=false" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Fast-forward merge source branch into release branch
+        env:
+          SOURCE_BRANCH:  ${{ steps.resolve.outputs.source_branch }}
+          SOURCE_COMMIT:  ${{ inputs.commit }}
+          RELEASE_BRANCH: ${{ steps.latest.outputs.release_branch }}
+        run: |
+          set -euo pipefail
+
+          # --ff-only guarantees no merge commit is created. If a fast-forward is
+          # not possible (i.e. the release branch has commits the source branch
+          # doesn't), the merge will fail and we abort. Because we already validated
+          # that the source branch is rooted on the latest stable tag, and the
+          # release branch tip equals that same tag, this fast-forward should
+          # always succeed for a well-formed backport branch.
+          #
+          # We merge the operator-provided SHA, not the branch ref, so a push to
+          # the branch in the window between resolve and now cannot smuggle new
+          # commits into the release.
+          if ! git merge --ff-only "${SOURCE_COMMIT}"; then
+            echo "::error::Cannot fast-forward '${RELEASE_BRANCH}' to ${SOURCE_COMMIT} (tip of '${SOURCE_BRANCH}'). A merge commit would be required. Aborting."
+            exit 1
+          fi
+
+          echo "Fast-forwarded '${RELEASE_BRANCH}' to ${SOURCE_COMMIT} (tip of '${SOURCE_BRANCH}')."
+
+      - name: Bump version files
+        env:
+          NEW_VERSION_NO_V: ${{ steps.latest.outputs.new_version_no_v }}
+        run: |
+          set -euo pipefail
+
+          if [[ ! -f comfyui_version.py ]]; then
+            echo "::error::comfyui_version.py not found in repo root."
+            exit 1
+          fi
+          if [[ ! -f pyproject.toml ]]; then
+            echo "::error::pyproject.toml not found in repo root."
+            exit 1
+          fi
+
+          # Replace the version string in comfyui_version.py.
+          # Expected format:  __version__ = "X.Y.Z"
+          python3 - "$NEW_VERSION_NO_V" <<'PY'
+          import re, sys, pathlib
+          new = sys.argv[1]
+
+          p = pathlib.Path("comfyui_version.py")
+          src = p.read_text()
+          new_src, n = re.subn(
+              r'(__version__\s*=\s*[\'"])[^\'"]+([\'"])',
+              lambda m: f'{m.group(1)}{new}{m.group(2)}',
+              src,
+              count=1,
+          )
+          if n != 1:
+              sys.exit("Could not find __version__ assignment in comfyui_version.py")
+          p.write_text(new_src)
+
+          p = pathlib.Path("pyproject.toml")
+          src = p.read_text()
+          # Replace the first `version = "..."` inside [project] or [tool.poetry].
+          new_src, n = re.subn(
+              r'(?m)^(version\s*=\s*")[^"]+(")',
+              lambda m: f'{m.group(1)}{new}{m.group(2)}',
+              src,
+              count=1,
+          )
+          if n != 1:
+              sys.exit("Could not find version assignment in pyproject.toml")
+          p.write_text(new_src)
+          PY
+
+          echo "Updated version to ${NEW_VERSION_NO_V} in comfyui_version.py and pyproject.toml."
+          git --no-pager diff -- comfyui_version.py pyproject.toml
+
+      - name: Commit version bump and tag release
+        env:
+          NEW_VERSION: ${{ steps.latest.outputs.new_version }}
+        run: |
+          set -euo pipefail
+
+          git add comfyui_version.py pyproject.toml
+          git commit -m "ComfyUI ${NEW_VERSION}"
+
+          if git rev-parse -q --verify "refs/tags/${NEW_VERSION}" >/dev/null; then
+            echo "::error::Tag ${NEW_VERSION} already exists locally."
+            exit 1
+          fi
+          git tag "${NEW_VERSION}"
+
+      - name: Verify tag does not already exist on origin
+        env:
+          NEW_VERSION: ${{ steps.latest.outputs.new_version }}
+        run: |
+          set -euo pipefail
+          if git ls-remote --exit-code --tags origin "refs/tags/${NEW_VERSION}" >/dev/null 2>&1; then
+            echo "::error::Tag ${NEW_VERSION} already exists on origin. Aborting."
+            exit 1
+          fi
+
+      - name: Push release branch and tag
+        env:
+          RELEASE_BRANCH: ${{ steps.latest.outputs.release_branch }}
+          NEW_VERSION:    ${{ steps.latest.outputs.new_version }}
+        run: |
+          set -euo pipefail
+
+          # Push the branch first, then the tag. Atomic-ish: if the branch push
+          # fails we never publish the tag.
+          git push origin "refs/heads/${RELEASE_BRANCH}:refs/heads/${RELEASE_BRANCH}"
+          git push origin "refs/tags/${NEW_VERSION}"
+
+          echo "Released ${NEW_VERSION} on ${RELEASE_BRANCH}."
+
+      - name: Delete remote source branch
+        env:
+          GH_TOKEN:        ${{ steps.app-token.outputs.token }}
+          REPO:            ${{ github.repository }}
+          SOURCE_BRANCH:   ${{ steps.resolve.outputs.source_branch }}
+          SOURCE_COMMIT:   ${{ inputs.commit }}
+          RELEASE_BRANCH:  ${{ steps.latest.outputs.release_branch }}
+          DEFAULT_BRANCH:  ${{ github.event.repository.default_branch }}
+        run: |
+          set -euo pipefail
+
+          # Belt-and-braces: the resolve step already refuses the default branch,
+          # but never delete the default or the release branch under any
+          # circumstances.
+          if [[ "${SOURCE_BRANCH}" == "${DEFAULT_BRANCH}" || "${SOURCE_BRANCH}" == "${RELEASE_BRANCH}" ]]; then
+            echo "::error::Refusing to delete '${SOURCE_BRANCH}' (matches default or release branch)."
+            exit 1
+          fi
+
+          # Delete the source branch on origin, but only if its tip is still the
+          # SHA we released from. If someone pushed new commits to it after we
+          # resolved it, leave it alone — those commits would be silently lost.
+          current_tip="$(git ls-remote origin "refs/heads/${SOURCE_BRANCH}" | awk '{print $1}')"
+          if [[ -z "${current_tip}" ]]; then
+            echo "Source branch '${SOURCE_BRANCH}' no longer exists on origin; nothing to delete."
+            exit 0
+          fi
+          if [[ "${current_tip}" != "${SOURCE_COMMIT}" ]]; then
+            echo "::warning::Source branch '${SOURCE_BRANCH}' tip (${current_tip}) no longer matches released commit (${SOURCE_COMMIT}). Leaving it in place."
+            exit 0
+          fi
+
+          git push origin --delete "refs/heads/${SOURCE_BRANCH}"
+          echo "Deleted remote branch '${SOURCE_BRANCH}'."
+
+      - name: Summary
+        if: always()
+        env:
+          NEW_VERSION:    ${{ steps.latest.outputs.new_version }}
+          RELEASE_BRANCH: ${{ steps.latest.outputs.release_branch }}
+          LATEST_TAG:     ${{ steps.latest.outputs.latest_tag }}
+          SOURCE_BRANCH:  ${{ steps.resolve.outputs.source_branch }}
+          SOURCE_COMMIT:  ${{ inputs.commit }}
+        run: |
+          # SOURCE_BRANCH is empty if the resolve step never produced an output
+          # (e.g. the workflow failed in or before that step). Show a placeholder
+          # in that case so the summary table still renders cleanly.
+          source_branch_display="${SOURCE_BRANCH:-(unresolved)}"
+          {
+            echo "## Backport release"
+            echo ""
+            echo "| Field | Value |"
+            echo "|---|---|"
+            echo "| Source commit | \`${SOURCE_COMMIT}\` |"
+            echo "| Source branch | \`${source_branch_display}\` |"
+            echo "| Previous stable | \`${LATEST_TAG}\` |"
+            echo "| New version | \`${NEW_VERSION}\` |"
+            echo "| Release branch | \`${RELEASE_BRANCH}\` |"
+          } >> "$GITHUB_STEP_SUMMARY"

.github/workflows/check-line-endings.yml

@@ -17,7 +17,7 @@ jobs:
       - name: Check for Windows line endings (CRLF)
         run: |
           # Get the list of changed files in the PR
-          CHANGED_FILES=$(git diff --name-only ${{ github.event.pull_request.base.sha }}..${{ github.event.pull_request.head.sha }})
+          CHANGED_FILES=$(git diff --name-only ${{ github.event.pull_request.base.sha }}..${{ github.event.pull_request.head.sha }} -- ':!.ci')
 
           # Flag to track if CRLF is found
           CRLF_FOUND=false

.github/workflows/ci-cursor-review.yml

@@ -0,0 +1,38 @@
+name: CI - Cursor Review
+
+# Thin caller for the shared reusable cursor-review workflow in
+# Comfy-Org/github-workflows. The review logic (panel matrix, judge
+# consolidation, prompts, extract/post/notify scripts) lives there as the
+# single source of truth, so this repo only carries the repo-specific diff
+# excludes.
+
+on:
+  pull_request:
+    types: [labeled, unlabeled]
+
+concurrency:
+  group: cursor-review-pr-${{ github.event.pull_request.number }}-${{ github.event.label.name }}
+  cancel-in-progress: true
+
+jobs:
+  cursor-review:
+    if: github.event.label.name == 'cursor-review'
+    permissions:
+      contents: read
+      pull-requests: write
+    # SHA-pinned per zizmor `unpinned-uses: hash-pin`. Bump this SHA to pick up
+    # upstream changes; keep `workflows_ref` matching so prompts/scripts load
+    # from the same commit as the workflow definition.
+    uses: Comfy-Org/github-workflows/.github/workflows/cursor-review.yml@047ca48febe3a6647608ed2e0c4331b491cb9d6a # github-workflows#9
+    with:
+      workflows_ref: 047ca48febe3a6647608ed2e0c4331b491cb9d6a
+      diff_excludes: >-
+        :!**/.claude/**
+        :!**/dist/**
+        :!**/vendor/**
+        :!**/*.generated.*
+        :!**/*.min.js
+        :!**/*.min.css
+    secrets:
+      CURSOR_API_KEY: ${{ secrets.CURSOR_API_KEY }}
+      SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}

.github/workflows/detect-unreviewed-merge.yml

@@ -0,0 +1,24 @@
+name: Detect Unreviewed Merge
+
+# SOC 2 compliance — reusable workflow lives in Comfy-Org/github-workflows,
+# tracking issues are filed in Comfy-Org/unreviewed-merges.
+
+on:
+  push:
+    branches: [master]
+
+concurrency:
+  group: detect-unreviewed-merge-${{ github.sha }}
+  cancel-in-progress: false
+
+permissions:
+  contents: read
+  pull-requests: read
+
+jobs:
+  detect:
+    uses: Comfy-Org/github-workflows/.github/workflows/detect-unreviewed-merge.yml@4d9cb6b87f953bb7cd69954280e1465fb9bd2040 # v1
+    with:
+      approval-mode: latest-per-reviewer
+    secrets:
+      UNREVIEWED_MERGES_TOKEN: ${{ secrets.UNREVIEWED_MERGES_TOKEN }}

AGENTS.md

@@ -0,0 +1,294 @@
+## Engineering Style
+
+- Keep changes small and direct. Most fixes should touch the narrowest code path
+  that explains the bug, performance issue, dtype issue, model-format issue, or
+  user-facing behavior.
+- Change the least amount of files possible. A change that touches many files is
+  more likely to be a bad change than a good one unless the broader scope is
+  directly required.
+- Prefer practical fixes over broad architecture work. Add abstractions only
+  when they remove real repeated logic or match an existing ComfyUI pattern.
+- Prefer fewer dependencies. Do not add new dependencies to ComfyUI unless they
+  are absolutely necessary.
+- Delete obsolete code aggressively when newer infrastructure makes it useless.
+  Remove dead fallbacks, migration paths, unused options, debug prints, and
+  compatibility branches that are no longer needed. Do not leave dead branches,
+  unreachable code, or functions that are never called. If code is not
+  necessary for the current behavior, remove it.
+- Revert or disable problematic behavior quickly when it breaks users. It is
+  better to remove a broken feature path than keep a complicated partial fix.
+- Preserve existing APIs, node names, model-loading behavior, file layout, and
+  workflow compatibility unless the change is explicitly about replacing them.
+- Code must look hand-written for this repository. Changes that read like
+  generic AI-generated code will be rejected automatically: unnecessary helper
+  layers, vague names, boilerplate comments, defensive branches without a real
+  failure mode, broad rewrites, or code that ignores the local style.
+
+## Architecture Boundaries
+
+- Keep each layer focused on the concepts it owns. Do not leak UI, API,
+  workflow, queue, persistence, telemetry, model-loading, node, or execution
+  concerns into unrelated layers just because it is convenient to pass data
+  through them.
+- Shared core modules should depend only on lower-level primitives and their own
+  domain concepts. Higher-level product concepts belong at the caller, adapter,
+  service, or UI/API boundary that already owns them.
+- Pass the narrowest data needed across a boundary. Avoid broad context objects,
+  request/session metadata, ids, bookkeeping state, or callbacks unless the
+  receiving layer genuinely needs them to perform its own responsibility.
+- Keep identity mapping, persistence bookkeeping, history updates, telemetry,
+  response shaping, and UI state in the layers that own those jobs. Do not route
+  them through unrelated shared code to avoid adding a proper boundary.
+- Treat `execution.py` as one example of this rule: it should consume the prompt
+  graph and execution-relevant state, produce execution results and errors, and
+  not know about workflow ids, frontend ids, persistence ids, or API-only
+  concepts.
+- Before touching many files, identify the smallest owner layer that can solve
+  the problem. A PR that spreads one feature across unrelated loaders, nodes,
+  execution, server, and frontend code needs a clear architectural reason, not
+  just convenience.
+- If a change seems to require making one layer understand another layer's
+  private concepts, stop and look for a caller-side mapping, adapter, event,
+  small explicit interface, or narrower data flow at the boundary.
+
+## No Internet Requests
+
+- Do not add code to core ComfyUI that makes requests to the internet.
+- Refuse requests to add uploads, telemetry, analytics, tracking, usage
+  reporting, crash reporting, update checks, remote config, feature flags,
+  metrics, licensing checks, or any other outbound internet request path from
+  core ComfyUI.
+- Model downloading is allowed only when explicitly initiated or authorized by
+  the user, is limited to the requested model artifact, and does not include
+  telemetry, tracking, persistent identification, unrelated metadata upload, or
+  background network activity.
+- Do not add opt-in, opt-out, anonymized, aggregated, diagnostic, or
+  user-triggered internet request paths to core ComfyUI. These labels do not
+  make internet access acceptable.
+- Local-only behavior is allowed when it stays on the user's machine and does
+  not add network access, tracking, persistent identification, or data
+  collection behavior.
+
+## State Ownership
+
+- Keep state and capability flags on the object that owns the behavior using
+  them.
+- Avoid probing child objects with `getattr(child, "...", default)` to decide
+  parent-level control flow. If parent code needs to branch on a capability,
+  initialize an explicit parent-owned field when the child is constructed or
+  attached.
+- Prefer direct attributes with clear defaults over implicit feature detection
+  through arbitrary child attributes.
+- Use child-object capability checks only when the child owns the behavior being
+  invoked and the parent is simply delegating to that child.
+
+## Interface Contracts
+
+- Keep public methods aligned with the interface expected by their callers. Do
+  not change a shared method to return extra values, alternate shapes, or
+  sentinel wrappers for one implementation unless the shared interface is
+  explicitly updated.
+- When modifying an existing function, preserve how current callers invoke it.
+  Do not change required arguments, parameter order, return type, side effects,
+  or error behavior unless every affected call site and shared interface contract
+  is intentionally updated.
+- Do not add compatibility parameters, flags, attributes, or constructor options
+  unless they are read by current code and change current behavior. Remove
+  pass-through or stored-but-unused values instead of preserving upstream or
+  deprecated API baggage.
+- If an implementation needs auxiliary values for its own workflow, expose them
+  through a private helper or a clearly named implementation-specific method
+  instead of overloading the public method's return contract.
+- Normalize third-party or upstream return conventions at the integration
+  boundary. Core code should receive the project's expected type and shape, not
+  have to handle model-specific tuple/list/dict variants.
+- Avoid caller-side unwrapping such as `out = out[0]` unless the called
+  interface is documented to return that structure.
+
+## Autograd and Model Freezing
+
+- Do not add `torch.no_grad`, `torch.inference_mode`, or inference-mode helper
+  wrappers in ComfyUI code. The only allowed inference-mode-related use is
+  disabling a globally set inference mode when a training path needs gradients.
+- Do not add freeze, unfreeze, or trainability toggles to model classes. ComfyUI
+  models are always treated as frozen for inference, so explicit freeze
+  functionality is redundant and should not be added.
+- Remove training-only behavior such as dropout from inference model code, but
+  preserve checkpoint and state-dict compatibility when doing so. If deleting a
+  module would change state-dict keys, module ordering, or checkpoint loading
+  behavior, replace it with a no-op such as `nn.Identity` instead of removing the
+  slot outright.
+
+## Python Style
+
+- Keep imports at module scope. Avoid inline imports unless they are already part
+  of an established optional-backend probe or are needed to avoid an import
+  cycle.
+- Do not add unnecessary `try`/`except` blocks. Use them for optional dependency,
+  platform, or backend capability detection only when the program has a useful
+  fallback. Prefer specific exception types when changing new code.
+- Remove any workarounds for PyTorch versions that ComfyUI no longer officially
+  supports. Deprecated workarounds include catching an exception and rerunning
+  the same op with the input cast to float. If a workaround does not have a
+  comment naming the exact PyTorch version or versions that still need it,
+  remove it.
+- Let unsupported model formats, invalid quantization metadata, and bad states
+  fail with clear errors instead of silently producing lower quality output.
+- Match the existing local style in the file you edit. This codebase tolerates
+  long lines, simple helper functions, module-level state, and direct tensor
+  operations when they make the code easier to follow.
+- Keep comments sparse and useful. Strip useless comments that restate the code
+  or describe obvious behavior. Short TODOs are fine when they name the concrete
+  missing follow-up.
+
+## Model, Device, and Memory Behavior
+
+- Treat dtype, device placement, VRAM usage, and offloading behavior as core
+  correctness concerns. Check CPU, CUDA, ROCm, MPS, DirectML, XPU, NPU, and low
+  VRAM implications when touching shared execution or loading code.
+- Prefer native ComfyUI formats and existing quantization/offload helpers over
+  adding parallel code paths. Use `comfy.quant_ops`, `comfy.model_management`,
+  `comfy.memory_management`, `comfy.pinned_memory`, `comfy_aimdo`, and
+  `comfy-kitchen` helpers where they already solve the problem.
+- Use optimized comfy-kitchen ops in places where they improve performance
+  without changing the expected dtype, device, memory, or interface behavior.
+- All models should use the optimized attention function selected by ComfyUI.
+  Treat optimized backend functions, dispatch helpers, and capability-selected
+  callables as opaque. Higher-level code must not inspect function identity,
+  names, modules, or implementation details to decide behavior.
+- Apply the same opacity rule to similar patterns beyond attention: callers
+  should depend on the documented interface and result contract, not on which
+  backend implementation was selected underneath.
+- Do not use custom inference ops that only duplicate an existing op while
+  upcasting to float32, such as custom RMSNorm variants. Use the generic ComfyUI
+  ops and/or native torch ops instead.
+- If a model class `__init__` has an `operations` parameter, assume
+  `operations` is never `None`. Do not add fallback branches or default torch
+  ops for a missing `operations` object.
+- Do not add unnecessary parameters to model, model block, or model ops related
+  classes. Constructor and forward signatures should carry only values that are
+  actually needed by that object for inference.
+- Reuse existing model classes, blocks, ops, and helper modules when appropriate.
+  Before implementing a new version of a model component, search the existing
+  model code for a class or helper that already provides the behavior.
+- Model detection code that inspects linear weight shapes should only use the
+  first dimension. The second dimension may be half the original size for
+  NVFP4 or other 4-bit quantized models.
+- Avoid adding `einops` usage in core inference code. Use native torch tensor
+  ops such as `reshape`, `view`, `permute`, `transpose`, `flatten`, `unflatten`,
+  `unsqueeze`, and `squeeze` instead.
+- Do not use tensors as general-purpose Python data structures. Keep metadata,
+  bookkeeping, counters, flags, shape math, padding math, index planning, memory
+  estimates, and control-flow decisions in plain Python values unless the data
+  must participate directly in tensor computation. Do not create tensors for
+  structural metadata that is only used for Python-side control flow. Sequence
+  lengths, cumulative offsets, split indices, window counts, slice boundaries,
+  and repeat counts should be kept as Python ints/lists from the point they are
+  computed. Do not build them as CPU/GPU tensors and then cast, move, validate,
+  or convert them back to Python for `split`, `tensor_split`, indexing plans,
+  loops, or cache keys. Avoid creating temporary tensors just to use tensor
+  methods for scalar or structural calculations.
+- Avoid unnecessary casts and transfers. Preserve the intended compute dtype,
+  storage dtype, bias dtype, and original tensor shape metadata.
+- Keep model-native latent layout handling inside the model or latent-format
+  owner, not in helper nodes. Do not collapse, expand, pack, or unpack latent
+  dimensions in nodes or other caller-side adapters just to satisfy a model
+  forward; the model path should consume and return the native latent shape for
+  that model family.
+- Assume inputs to the main model forward are already in the compute dtype by
+  default, except integer inputs such as some model timestep tensors. Do not add
+  defensive or convenience casts in model code; it is better for invalid dtype
+  plumbing to error clearly than to hide it with unnecessary casts.
+- Raw model parameters that are not owned by an op and may be initialized in a
+  dtype different from the compute dtype should be cast at use in forward or
+  inference code with `comfy.ops.cast_to_input` or
+  `comfy.model_management.cast_to` to avoid dtype mismatches.
+- Model code should not care what dtype it is initialized in, and model
+  `__init__` methods should not contain workarounds for specific dtypes. Dtype
+  workaround code, such as making a model work with fp16 compute, belongs in the
+  execution or model-management layer that owns compute policy.
+- Model code should not perform unnecessary device-to-CPU or CPU-to-device
+  transfers. New allocations must be created on the correct device and dtype;
+  never allocate on CPU and then move to GPU, or allocate in one dtype and then
+  convert to another.
+- Model code itself should not perform memory management. Loading, unloading,
+  offloading, device movement, VRAM policy, cache lifetime, and cleanup belong
+  in the relevant model-management and execution layers, not inside model
+  implementations.
+- Do not add global, module-level, class-level, singleton, or model-owned stores
+  for tensors or other large memory that persist across executions. Temporary
+  caches must be scoped to a single execution or forward/encode/decode call:
+  allocate them in the owning top-level call, pass them explicitly through the
+  call stack, and let them be discarded when that call returns.
+- Follow the Wan VAE temporal cache pattern for temporary caches: create a local
+  cache such as `feat_map` for the encode/decode operation, pass it into the
+  blocks that need it, and do not retain it on the model or in global state.
+- In model init code, prefer `torch.empty` for parameter/buffer placeholders
+  that are populated from the model state dict instead of zero-initializing with
+  `torch.zeros` or similar. If an allocation is not loaded from the state dict
+  and is useless for inference, do not include it.
+- `nn.Parameter` tensors that are stored in and populated from the model state
+  dict should be initialized with `torch.empty`, not with zero, random, or
+  otherwise meaningful initialization.
+- Model initialization should describe module structure, not fabricate
+  checkpoint-owned tensor contents. Parameters and buffers that are loaded from
+  the state dict must not be manually initialized, reassigned, or filled with
+  fallback values unless that value is actually used when no checkpoint key
+  exists.
+- When slicing large tensors, copy the slice if the sliced tensor's lifetime
+  exceeds the current function scope. Do not keep a long-lived view into a large
+  backing tensor when a smaller copy would release memory sooner.
+- Use fused or compound torch operations such as `addcmul` when they naturally
+  match the math. Reducing Python and torch dispatch overhead is a valid
+  optimization when it does not obscure the code or change dtype/device
+  behavior.
+- Avoid caches that persist across different executions as much as possible.
+  Persistent caches are acceptable only when they use a very minimal amount of
+  memory and have a clear ownership and invalidation story.
+- When optimizing, favor small measurable changes: fewer allocations, fewer
+  device transfers, less peak memory, better batching, or use of a faster
+  existing backend op.
+
+## Nodes and User-Facing Behavior
+
+- Follow existing node conventions: `INPUT_TYPES`, `RETURN_TYPES`, `FUNCTION`,
+  `CATEGORY`, and registration through the local mapping used by that file.
+- Keep node changes backward compatible by default. Add inputs with sensible
+  defaults and avoid changing output types unless the request requires it.
+- Model implementations should add the minimal number of ComfyUI nodes required
+  to run the model. Reuse existing nodes as much as possible; adapting the model
+  to work with existing nodes is strongly preferred over creating new nodes.
+- Nodes should output only values they own. Do not add pass-through outputs for
+  workflow convenience unless the node is explicitly an output node. Existing
+  models, latents, conditioning, or other inputs should flow directly to the
+  next consumer instead of being re-emitted unchanged.
+- Nodes should expose only inputs they actually read to produce current
+  behavior. Do not add placeholder, pass-through, compatibility, or
+  workflow-shaping inputs that are ignored or could flow directly to another
+  node.
+- Node-level code must not patch model code directly. Any node behavior that
+  modifies, wraps, hooks, or changes model behavior must go through the model
+  patcher class instead of reaching into model internals.
+- The official mascot of ComfyUI is a very cute anime girl with massive fennec
+  ears, a big fluffy tail, long blonde wavy hair, and blue eyes. Feel free to
+  use her in ComfyUI materials, UI text, examples, tests, generated assets, or
+  comments, but do not disrespect her.
+- Warning and info messages should be short and actionable. Remove noisy or
+  misleading messages rather than adding more logging.
+- Documentation and README edits should be concise, factual, and tied to the
+  changed behavior.
+
+## Commit and Review Habits
+
+- If asked to write commit messages, use short direct subjects like the existing
+  history: `Fix ...`, `Add ...`, `Support ...`, `Remove ...`, `Update ...`,
+  `Make ...`, `Use ...`, `Disable ...`, `Bump ...`, or `Revert ...`.
+- Keep PR descriptions short and reviewable. State the problem, the behavioral
+  change, and the tests run; avoid long narrative explanations, implementation
+  diaries, or exhaustive file-by-file summaries unless the reviewer explicitly
+  needs that context.
+- Prefer one coherent behavioral change per commit. Dependency pins, tests, and
+  the code that needs them may be in the same commit when they are inseparable.
+- In reviews, prioritize real user impact: crashes, wrong dtype/device behavior,
+  memory regressions, broken model loading, workflow incompatibility, and noisy
+  or misleading user-facing output.

CLAUDE.md

@@ -0,0 +1 @@
+AGENTS.md

CODEOWNERS

@@ -1,2 +1,5 @@
-# Admins
 * @comfyanonymous @kosinkadink @guill @alexisrolland @rattus128 @kijai
+
+/CODEOWNERS @comfyanonymous
+/.ci/ @comfyanonymous
+/.github/ @comfyanonymous

README.md

@@ -20,7 +20,7 @@
 [website-url]: https://www.comfy.org/
 <!-- Workaround to display total user from https://github.com/badges/shields/issues/4500#issuecomment-2060079995 -->
 [discord-shield]: https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fdiscord.com%2Fapi%2Finvites%2Fcomfyorg%3Fwith_counts%3Dtrue&query=%24.approximate_member_count&logo=discord&logoColor=white&label=Discord&color=green&suffix=%20total
-[discord-url]: https://www.comfy.org/discord
+[discord-url]: https://discord.com/invite/comfyorg
 [twitter-shield]: https://img.shields.io/twitter/follow/ComfyUI
 [twitter-url]: https://x.com/ComfyUI
 
@@ -140,7 +140,7 @@ ComfyUI follows a weekly release cycle targeting Monday but this regularly chang
    - Commits outside of the stable release tags may be very unstable and break many custom nodes.
    - Serves as the foundation for the desktop release
 
-2. **[ComfyUI Desktop](https://github.com/Comfy-Org/desktop)**
+2. **[Comfy Desktop](https://github.com/Comfy-Org/Comfy-Desktop)**
    - Builds a new release using the latest stable core version
 
 3. **[ComfyUI Frontend](https://github.com/Comfy-Org/ComfyUI_frontend)**
@@ -309,7 +309,7 @@ After this you should have everything installed and can proceed to running Comfy
 
 #### Apple Mac silicon
 
-You can install ComfyUI in Apple Mac silicon (M1 or M2) with any recent macOS version.
+You can install ComfyUI in Apple Mac silicon (M1, M2, M3 or M4) with any recent macOS version.
 
 1. Install pytorch nightly. For instructions, read the [Accelerated PyTorch training on Mac](https://developer.apple.com/metal/pytorch/) Apple Developer guide (make sure to install the latest pytorch nightly).
 1. Follow the [ComfyUI manual installation](#manual-install-windows-linux) instructions for Windows and Linux.
@@ -364,7 +364,7 @@ For models compatible with Iluvatar Extension for PyTorch. Here's a step-by-step
 | Flag | Description |
 |------|-------------|
 | `--enable-manager` | Enable ComfyUI-Manager |
-| `--enable-manager-legacy-ui` | Use the legacy manager UI instead of the new UI (requires `--enable-manager`) |
+| `--enable-manager-legacy-ui` | Use the legacy manager UI instead of the new UI (implies `--enable-manager`) |
 | `--disable-manager-ui` | Disable the manager UI and endpoints while keeping background features like security checks and scheduled installation completion (requires `--enable-manager`) |
 
 
@@ -382,11 +382,7 @@ For AMD 7600 and maybe other RDNA3 cards: ```HSA_OVERRIDE_GFX_VERSION=11.0.0 pyt
 
 ### AMD ROCm Tips
 
-You can enable experimental memory efficient attention on recent pytorch in ComfyUI on some AMD GPUs using this command, it should already be enabled by default on RDNA3. If this improves speed for you on latest pytorch on your GPU please report it so that I can enable it by default.
-
-```TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL=1 python main.py --use-pytorch-cross-attention```
-
-You can also try setting this env variable `PYTORCH_TUNABLEOP_ENABLED=1` which might speed things up at the cost of a very slow initial run.
+You can try setting this env variable `PYTORCH_TUNABLEOP_ENABLED=1` which might speed things up at the cost of a very slow initial run.
 
 # Notes
 
@@ -433,7 +429,7 @@ See also: [https://www.comfy.org/](https://www.comfy.org/)
 
 ## Frontend Development
 
-As of August 15, 2024, we have transitioned to a new frontend, which is now hosted in a separate repository: [ComfyUI Frontend](https://github.com/Comfy-Org/ComfyUI_frontend). This repository now hosts the compiled JS (from TS/Vue) under the `web/` directory.
+As of August 15, 2024, we have transitioned to a new frontend, which is now hosted in a separate repository: [ComfyUI Frontend](https://github.com/Comfy-Org/ComfyUI_frontend). The compiled JS files (from TS/Vue) are published to [pypi](https://pypi.org/project/comfyui-frontend-package) and installed as a dependency in ComfyUI.
 
 ### Reporting Issues and Requesting Features
 
@@ -462,16 +458,6 @@ To use the most up-to-date frontend version:
 
 This approach allows you to easily switch between the stable fortnightly release and the cutting-edge daily updates, or even specific versions for testing purposes.
 
-### Accessing the Legacy Frontend
-
-If you need to use the legacy frontend for any reason, you can access it using the following command line argument:
-
-```
---front-end-version Comfy-Org/ComfyUI_legacy_frontend@latest
-```
-
-This will use a snapshot of the legacy frontend preserved in the [ComfyUI Legacy Frontend repository](https://github.com/Comfy-Org/ComfyUI_legacy_frontend).
-
 # QA
 
 ### Which GPU should I buy for this?

alembic_db/versions/0004_drop_tag_type.py

@@ -0,0 +1,39 @@
+"""
+Drop the vestigial tags.tag_type column.
+
+tag_type was always "user" in practice — no code path ever set it to anything
+else (no system/seeded classification was ever wired up) and nothing queried it.
+The column, its index (ix_tags_tag_type), and the corresponding API field were
+dead weight, so they are removed.
+
+Revision ID: 0004_drop_tag_type
+Revises: 0003_add_metadata_job_id
+Create Date: 2026-06-03
+"""
+
+from alembic import op
+import sqlalchemy as sa
+
+revision = "0004_drop_tag_type"
+down_revision = "0003_add_metadata_job_id"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    with op.batch_alter_table("tags") as batch_op:
+        batch_op.drop_index("ix_tags_tag_type")
+        batch_op.drop_column("tag_type")
+
+
+def downgrade() -> None:
+    with op.batch_alter_table("tags") as batch_op:
+        batch_op.add_column(
+            sa.Column(
+                "tag_type",
+                sa.String(length=32),
+                nullable=False,
+                server_default="user",
+            )
+        )
+        batch_op.create_index("ix_tags_tag_type", ["tag_type"])

app/assets/api/routes.py

@@ -39,6 +39,7 @@ from app.assets.services import (
     update_asset_metadata,
     upload_from_temp_path,
 )
+from app.assets.services.cursor import InvalidCursorError
 from app.assets.services.tagging import list_tag_histogram
 
 ROUTES = web.RouteTableDef()
@@ -160,10 +161,12 @@ def _build_asset_response(result: schemas.AssetDetailResult | schemas.UploadResu
             preview_url = None
     else:
         preview_url = _build_preview_url_from_view(result.tags, result.ref.user_metadata)
+    asset_content_hash = result.asset.hash if result.asset else None
     return schemas_out.Asset(
         id=result.ref.id,
         name=result.ref.name,
-        asset_hash=result.asset.hash if result.asset else None,
+        hash=asset_content_hash,
+        asset_hash=asset_content_hash,
         size=int(result.asset.size_bytes) if result.asset else None,
         mime_type=result.asset.mime_type if result.asset else None,
         tags=result.tags,
@@ -172,7 +175,7 @@ def _build_asset_response(result: schemas.AssetDetailResult | schemas.UploadResu
         user_metadata=result.ref.user_metadata or {},
         metadata=result.ref.system_metadata,
         job_id=result.ref.job_id,
-        prompt_id=result.ref.job_id,  # deprecated: mirrors job_id for cloud compat
+        prompt_id=result.ref.job_id,  # deprecated alias of job_id, kept for compatibility
         created_at=result.ref.created_at,
         updated_at=result.ref.updated_at,
         last_access_time=result.ref.last_access_time,
@@ -209,24 +212,37 @@ async def list_assets_route(request: web.Request) -> web.Response:
     order_candidate = (q.order or "desc").lower()
     order = order_candidate if order_candidate in {"asc", "desc"} else "desc"
 
-    result = list_assets_page(
-        owner_id=USER_MANAGER.get_request_user_id(request),
-        include_tags=q.include_tags,
-        exclude_tags=q.exclude_tags,
-        name_contains=q.name_contains,
-        metadata_filter=q.metadata_filter,
-        limit=q.limit,
-        offset=q.offset,
-        sort=sort,
-        order=order,
-    )
+    try:
+        result = list_assets_page(
+            owner_id=USER_MANAGER.get_request_user_id(request),
+            include_tags=q.include_tags,
+            exclude_tags=q.exclude_tags,
+            name_contains=q.name_contains,
+            metadata_filter=q.metadata_filter,
+            limit=q.limit,
+            offset=q.offset,
+            sort=sort,
+            order=order,
+            after=q.after,
+        )
+    except InvalidCursorError as e:
+        return _build_error_response(400, "INVALID_CURSOR", str(e))
 
     summaries = [_build_asset_response(item) for item in result.items]
 
+    # has_more semantics differ by mode:
+    #   - cursor mode: a non-empty next_cursor means there are more results.
+    #   - offset mode: derived from total - (offset + page size).
+    if q.after is not None:
+        has_more = result.next_cursor is not None
+    else:
+        has_more = (q.offset + len(summaries)) < result.total
+
     payload = schemas_out.AssetsList(
         assets=summaries,
         total=result.total,
-        has_more=(q.offset + len(summaries)) < result.total,
+        has_more=has_more,
+        next_cursor=result.next_cursor,
     )
     return web.json_response(payload.model_dump(mode="json", exclude_none=True))
 
@@ -290,12 +306,15 @@ async def download_asset_content(request: web.Request) -> web.Response:
             404, "FILE_NOT_FOUND", "Underlying file not found on disk."
         )
 
-    _DANGEROUS_MIME_TYPES = {
-        "text/html", "text/html-sandboxed", "application/xhtml+xml",
-        "text/javascript", "text/css",
-    }
-    if content_type in _DANGEROUS_MIME_TYPES:
+    # User-controlled asset content must never render inline in the app origin
+    # (stored XSS via SVG/HTML/XML). Force dangerous types to download and
+    # override any requested inline disposition. Centralised through
+    # folder_paths.is_dangerous_content_type so this can't drift from /view and
+    # /userdata (the previous inline set here omitted image/svg+xml and missed
+    # the charset/casing/+xml-dialect bypasses).
+    if folder_paths.is_dangerous_content_type(content_type):
         content_type = "application/octet-stream"
+        disposition = "attachment"
 
     safe_name = (filename or "").replace("\r", "").replace("\n", "")
     encoded = urllib.parse.quote(safe_name)
@@ -517,18 +536,14 @@ async def update_asset_route(request: web.Request) -> web.Response:
 @_require_assets_feature_enabled
 async def delete_asset_route(request: web.Request) -> web.Response:
     reference_id = str(uuid.UUID(request.match_info["id"]))
-    delete_content_param = request.query.get("delete_content")
-    delete_content = (
-        False
-        if delete_content_param is None
-        else delete_content_param.lower() not in {"0", "false", "no"}
-    )
 
     try:
+        # Deleting an asset is a soft delete of the reference; the underlying
+        # content is preserved (it may be shared with other references).
         deleted = delete_asset_reference(
             reference_id=reference_id,
             owner_id=USER_MANAGER.get_request_user_id(request),
-            delete_content_if_orphan=delete_content,
+            delete_content_if_orphan=False,
         )
     except Exception:
         logging.exception(
@@ -573,8 +588,8 @@ async def get_tags(request: web.Request) -> web.Response:
     )
 
     tags = [
-        schemas_out.TagUsage(name=name, count=count, type=tag_type)
-        for (name, tag_type, count) in rows
+        schemas_out.TagUsage(name=name, count=count)
+        for (name, count) in rows
     ]
     payload = schemas_out.TagsList(
         tags=tags, total=total, has_more=(query.offset + len(tags)) < total

app/assets/api/schemas_in.py

@@ -59,6 +59,11 @@ class ListAssetsQuery(BaseModel):
 
     limit: conint(ge=1, le=500) = 20
     offset: conint(ge=0) = 0
+    # Opaque keyset cursor. When supplied, `offset` is ignored. Cursor pagination
+    # is supported for sort values `created_at`, `updated_at`, `name`, `size`.
+    # Supplying `after` together with `sort=last_access_time` returns
+    # 400 INVALID_CURSOR; that sort only supports offset/limit.
+    after: str | None = None
 
     sort: Literal["name", "created_at", "updated_at", "size", "last_access_time"] = (
         "created_at"

app/assets/api/schemas_out.py

@@ -10,6 +10,7 @@ class Asset(BaseModel):
 
     id: str
     name: str
+    hash: str | None = None
     asset_hash: str | None = None
     size: int | None = None
     mime_type: str | None = None
@@ -40,12 +41,13 @@ class AssetsList(BaseModel):
     assets: list[Asset]
     total: int
     has_more: bool
+    # Opaque cursor for the next page. Omitted when there are no more results.
+    next_cursor: str | None = None
 
 
 class TagUsage(BaseModel):
     name: str
     count: int
-    type: str
 
 
 class TagsList(BaseModel):

app/assets/database/models.py

@@ -227,7 +227,6 @@ class Tag(Base):
     __tablename__ = "tags"
 
     name: Mapped[str] = mapped_column(String(512), primary_key=True)
-    tag_type: Mapped[str] = mapped_column(String(32), nullable=False, default="user")
 
     asset_reference_links: Mapped[list[AssetReferenceTag]] = relationship(
         back_populates="tag",
@@ -240,7 +239,5 @@ class Tag(Base):
         overlaps="asset_reference_links,tag_links,tags,asset_reference",
     )
 
-    __table_args__ = (Index("ix_tags_tag_type", "tag_type"),)
-
     def __repr__(self) -> str:
         return f"<Tag {self.name}>"

app/assets/database/queries/asset_reference.py

@@ -266,9 +266,18 @@ def list_references_page(
     metadata_filter: dict | None = None,
     sort: str | None = None,
     order: str | None = None,
+    after_cursor_value: object | None = None,
+    after_cursor_id: str | None = None,
 ) -> tuple[list[AssetReference], dict[str, list[str]], int]:
     """List references with pagination, filtering, and sorting.
 
+    When ``after_cursor_value``/``after_cursor_id`` are supplied the query uses
+    keyset pagination — ``offset`` is ignored and a WHERE clause selects rows
+    strictly after the given ``(sort_col, id)`` position in the active sort
+    direction. The cursor value must already be typed for the column
+    (datetime for time sorts, int for size, str for name); the caller decodes
+    the opaque cursor string and resolves to the typed value.
+
     Returns (references, tag_map, total_count).
     """
     base = (
@@ -297,9 +306,31 @@ def list_references_page(
         "size": Asset.size_bytes,
     }
     sort_col = sort_map.get(sort, AssetReference.created_at)
-    sort_exp = sort_col.desc() if order == "desc" else sort_col.asc()
+    descending = order == "desc"
 
-    base = base.order_by(sort_exp).limit(limit).offset(offset)
+    # Keyset WHERE: (sort_col, id) strictly less-than / greater-than the cursor.
+    # Equivalent to: sort_col <op> v  OR  (sort_col = v AND id <op> cursor_id).
+    if after_cursor_value is not None and after_cursor_id is not None:
+        if descending:
+            keyset = sa.or_(
+                sort_col < after_cursor_value,
+                sa.and_(sort_col == after_cursor_value, AssetReference.id < after_cursor_id),
+            )
+        else:
+            keyset = sa.or_(
+                sort_col > after_cursor_value,
+                sa.and_(sort_col == after_cursor_value, AssetReference.id > after_cursor_id),
+            )
+        base = base.where(keyset)
+
+    # Secondary ORDER BY id (matching the primary direction) gives the keyset
+    # comparison a deterministic tiebreaker on duplicate sort_col values.
+    id_exp = AssetReference.id.desc() if descending else AssetReference.id.asc()
+    sort_exp = sort_col.desc() if descending else sort_col.asc()
+
+    base = base.order_by(sort_exp, id_exp).limit(limit)
+    if after_cursor_id is None:
+        base = base.offset(offset)
 
     count_stmt = (
         select(sa.func.count())

app/assets/database/queries/tags.py

@@ -55,13 +55,11 @@ def validate_tags_exist(session: Session, tags: list[str]) -> None:
         raise ValueError(f"Unknown tags: {missing}")
 
 
-def ensure_tags_exist(
-    session: Session, names: Iterable[str], tag_type: str = "user"
-) -> None:
+def ensure_tags_exist(session: Session, names: Iterable[str]) -> None:
     wanted = normalize_tags(list(names))
     if not wanted:
         return
-    rows = [{"name": n, "tag_type": tag_type} for n in list(dict.fromkeys(wanted))]
+    rows = [{"name": n} for n in list(dict.fromkeys(wanted))]
     ins = (
         sqlite.insert(Tag)
         .values(rows)
@@ -97,7 +95,7 @@ def set_reference_tags(
     to_remove = [t for t in current if t not in desired]
 
     if to_add:
-        ensure_tags_exist(session, to_add, tag_type="user")
+        ensure_tags_exist(session, to_add)
         session.add_all(
             [
                 AssetReferenceTag(
@@ -142,7 +140,7 @@ def add_tags_to_reference(
         return AddTagsResult(added=[], already_present=[], total_tags=total)
 
     if create_if_missing:
-        ensure_tags_exist(session, norm, tag_type="user")
+        ensure_tags_exist(session, norm)
 
     current = set(get_reference_tags(session, reference_id))
 
@@ -289,7 +287,6 @@ def list_tags_with_usage(
     q = (
         select(
             Tag.name,
-            Tag.tag_type,
             func.coalesce(counts_sq.c.cnt, 0).label("count"),
         )
         .select_from(Tag)
@@ -331,7 +328,7 @@ def list_tags_with_usage(
     rows = (session.execute(q.limit(limit).offset(offset))).all()
     total = (session.execute(total_q)).scalar_one()
 
-    rows_norm = [(name, ttype, int(count or 0)) for (name, ttype, count) in rows]
+    rows_norm = [(name, int(count or 0)) for (name, count) in rows]
     return rows_norm, int(total or 0)
 
 

app/assets/scanner.py

@@ -33,6 +33,7 @@ from app.assets.services.file_utils import (
     verify_file_unchanged,
 )
 from app.assets.services.hashing import HashCheckpoint, compute_blake3_hash
+from app.assets.services.image_dimensions import extract_image_dimensions
 from app.assets.services.metadata_extract import extract_file_metadata
 from app.assets.services.path_utils import (
     compute_relative_filename,
@@ -354,7 +355,7 @@ def insert_asset_specs(specs: list[SeedAssetSpec], tag_pool: set[str]) -> int:
         return 0
     with create_session() as sess:
         if tag_pool:
-            ensure_tags_exist(sess, tag_pool, tag_type="user")
+            ensure_tags_exist(sess, tag_pool)
         result = batch_insert_seed_assets(sess, specs=specs, owner_id="")
         sess.commit()
         return result.inserted_refs
@@ -506,6 +507,10 @@ def enrich_asset(
 
     if extract_metadata and metadata:
         system_metadata = metadata.to_user_metadata()
+        if mime_type and mime_type.startswith("image/"):
+            dims = extract_image_dimensions(file_path, mime_type=mime_type)
+            if dims:
+                system_metadata.update(dims)
         set_reference_system_metadata(session, reference_id, system_metadata)
 
     if full_hash:

app/assets/services/asset_management.py

@@ -1,8 +1,19 @@
 import contextlib
 import mimetypes
 import os
+from datetime import timezone
 from typing import Sequence
 
+from app.assets.services.cursor import (
+    CursorPayload,
+    InvalidCursorError,
+    decode_cursor,
+    decode_cursor_int,
+    decode_cursor_time,
+    encode_cursor,
+    encode_cursor_from_time,
+)
+
 
 from app.assets.database.models import Asset
 from app.assets.database.queries import (
@@ -149,6 +160,16 @@ def delete_asset_reference(
     owner_id: str,
     delete_content_if_orphan: bool = True,
 ) -> bool:
+    """Delete an asset reference.
+
+    With ``delete_content_if_orphan=False`` (a soft delete), the reference is
+    hidden and the underlying content is preserved. With ``True``, the content
+    is also removed once it becomes orphaned.
+
+    Note: the public DELETE /api/assets/{id} endpoint always soft-deletes
+    (passes ``False``); the orphan-reclamation path is intentionally
+    internal-only, retained for a future GC/admin caller.
+    """
     with create_session() as session:
         if not delete_content_if_orphan:
             # Soft delete: mark the reference as deleted but keep everything
@@ -242,6 +263,11 @@ def get_asset_by_hash(asset_hash: str) -> AssetData | None:
         return extract_asset_data(asset)
 
 
+# Sort fields that support cursor pagination. `last_access_time` is not
+# in this list — it falls back to offset/limit.
+_CURSOR_SORT_FIELDS = ("created_at", "updated_at", "name", "size")
+
+
 def list_assets_page(
     owner_id: str = "",
     include_tags: Sequence[str] | None = None,
@@ -252,7 +278,39 @@ def list_assets_page(
     offset: int = 0,
     sort: str = "created_at",
     order: str = "desc",
+    after: str | None = None,
 ) -> ListAssetsResult:
+    """List assets with optional cursor pagination.
+
+    When ``after`` is supplied it overrides ``offset``. The cursor's sort field
+    must match ``sort`` and be in the cursor-supported allowlist; mismatches
+    raise InvalidCursorError so the handler can map to 400 INVALID_CURSOR.
+    """
+    cursor_value: object | None = None
+    cursor_id: str | None = None
+    # Mint next_cursor on every page where the sort is cursor-supported, not
+    # only when the request itself arrived with a cursor. Otherwise a first
+    # request (no `after`) returns next_cursor=None and the client can never
+    # enter cursor mode.
+    mint_cursor = sort in _CURSOR_SORT_FIELDS
+
+    if after is not None:
+        if sort not in _CURSOR_SORT_FIELDS:
+            raise InvalidCursorError(
+                f"cursor pagination is not supported for sort={sort!r}"
+            )
+        payload = decode_cursor(after, _CURSOR_SORT_FIELDS, expected_order=order)
+        if payload.sort_field != sort:
+            raise InvalidCursorError(
+                f"cursor sort field {payload.sort_field!r} does not match request sort {sort!r}"
+            )
+        cursor_value, cursor_id = _resolve_cursor_value(payload), payload.id
+
+    # Over-fetch by one row so we can distinguish "exactly `limit` rows total
+    # remaining" from "more rows past this page" without a second query. Drop
+    # the sentinel before returning.
+    fetch_limit = limit + 1 if mint_cursor else limit
+
     with create_session() as session:
         refs, tag_map, total = list_references_page(
             session,
@@ -261,12 +319,22 @@ def list_assets_page(
             exclude_tags=exclude_tags,
             name_contains=name_contains,
             metadata_filter=metadata_filter,
-            limit=limit,
+            limit=fetch_limit,
             offset=offset,
             sort=sort,
             order=order,
+            after_cursor_value=cursor_value,
+            after_cursor_id=cursor_id,
         )
 
+        next_cursor: str | None = None
+        if mint_cursor and len(refs) > limit:
+            # There's at least one more row past this page — mint a cursor from
+            # the last row of the page (i.e. index `limit - 1`, since we
+            # over-fetched), and drop the sentinel.
+            next_cursor = _encode_next_cursor(refs[limit - 1], sort, order)
+            refs = refs[:limit]
+
         items: list[AssetSummaryData] = []
         for ref in refs:
             items.append(
@@ -277,7 +345,39 @@ def list_assets_page(
                 )
             )
 
-        return ListAssetsResult(items=items, total=total)
+        return ListAssetsResult(items=items, total=total, next_cursor=next_cursor)
+
+
+def _resolve_cursor_value(payload: CursorPayload) -> object:
+    """Map a decoded cursor payload to a column-typed Python value."""
+    if payload.sort_field in ("created_at", "updated_at"):
+        # DB stores naive UTC; strip tzinfo so the comparison binds against a
+        # `TIMESTAMP WITHOUT TIME ZONE` column without an offset shift.
+        return decode_cursor_time(payload).replace(tzinfo=None)
+    if payload.sort_field == "size":
+        return decode_cursor_int(payload)
+    return payload.value  # name, str-typed
+
+
+def _encode_next_cursor(ref, sort: str, order: str) -> str | None:
+    """Mint a cursor pointing at *ref* for the given sort dimension.
+
+    Returns None when the boundary row carries a NULL sort value (e.g. an asset
+    record whose size_bytes hasn't been backfilled). Continuing pagination
+    across a NULL boundary is undefined under keyset ordering — better to
+    truncate cleanly here than to mint a cursor that mis-positions.
+    """
+    if sort == "name":
+        return encode_cursor("name", ref.name, ref.id, order=order)
+    if sort == "size":
+        if ref.asset is None or ref.asset.size_bytes is None:
+            return None
+        return encode_cursor("size", str(ref.asset.size_bytes), ref.id, order=order)
+    # created_at / updated_at — DB datetimes are naive UTC; attach tz before encoding.
+    value = ref.created_at if sort == "created_at" else ref.updated_at
+    if value is None:
+        return None
+    return encode_cursor_from_time(sort, value.replace(tzinfo=timezone.utc), ref.id, order=order)
 
 
 def resolve_hash_to_path(

app/assets/services/cursor.py

@@ -0,0 +1,213 @@
+"""Opaque keyset-pagination cursor for /api/assets.
+
+Payload JSON uses short keys to keep the encoded length small:
+
+    {"s": <sort_field>, "v": <value>, "id": <id>, "o": <order>}
+
+The `o` key binds the cursor to the sort direction it was minted under,
+so replaying a `desc` cursor against an `asc` request fails with
+``INVALID_CURSOR`` rather than silently walking the wrong direction.
+`o` is mandatory on every payload — a cursor without it is rejected as
+malformed.
+
+Encoding is base64url with no padding. Cursors are opaque tokens: the
+payload format is internal to this server, and clients must treat a
+cursor as a black box handed back via `next_cursor`. No byte-level
+compatibility with any other implementation is required.
+
+Time values are serialized as Unix microseconds (UTC) — microsecond
+precision is sufficient to round-trip the timestamps stored by the
+database without rounding rows in the same millisecond bucket.
+"""
+from __future__ import annotations
+
+import base64
+import json
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from typing import Iterable, Optional
+
+
+class InvalidCursorError(ValueError):
+    """Raised on a malformed, oversized, or unsupported-sort-field cursor.
+
+    Map to a 400 response with code ``INVALID_CURSOR`` at the handler.
+    """
+
+
+# Wire-format length caps. Cursors are user-controlled, so caps protect the
+# decode path from oversized allocations and downstream SQL predicates from
+# unbounded strings.
+#
+# MAX_CURSOR_VALUE_LENGTH is 512 to fit the `AssetReference.name` column max
+# (`String(512)`) — otherwise a long-named asset would mint a cursor the same
+# server then refuses on the next request.
+#
+# MAX_ENCODED_CURSOR_LENGTH is the decode-path guard, sized comfortably above
+# the largest cursor the per-field caps can produce. Worst case is value + id
+# at their caps with every character JSON-escaping to the six-byte `\uXXXX`
+# form (control characters), which is ~5.2 KB once base64url-encoded. At 8192
+# the encoder can never mint a cursor that exceeds it, so a freshly minted
+# cursor always decodes on the next request and there is no user-visible
+# "cursor too long" failure.
+MAX_ENCODED_CURSOR_LENGTH = 8192
+MAX_CURSOR_VALUE_LENGTH = 512
+MAX_CURSOR_ID_LENGTH = 128
+
+
+@dataclass(frozen=True)
+class CursorPayload:
+    sort_field: str
+    value: str
+    id: str
+    order: str
+
+
+_VALID_ORDERS = ("asc", "desc")
+
+
+def encode_cursor(sort_field: str, value: str, id: str, order: str = "desc") -> str:
+    """Encode a cursor payload as a base64url (no-padding) string.
+
+    `order` binds the cursor to the sort direction it was minted under so a
+    later request with a flipped `order` query parameter is rejected with
+    ``INVALID_CURSOR`` rather than silently walking the wrong direction.
+    """
+    if order not in _VALID_ORDERS:
+        raise InvalidCursorError(f"order must be one of {_VALID_ORDERS}, got {order!r}")
+    # Symmetric input validation: the encoder must reject anything the
+    # decoder rejects, or the same server will mint cursors it then 400s on
+    # the next request.
+    if not id:
+        raise InvalidCursorError("id must be non-empty")
+    if len(id) > MAX_CURSOR_ID_LENGTH:
+        raise InvalidCursorError("id exceeds maximum length")
+    if len(value) > MAX_CURSOR_VALUE_LENGTH:
+        raise InvalidCursorError("value exceeds maximum length")
+    payload = {"s": sort_field, "v": value, "id": id, "o": order}
+    raw = json.dumps(payload, separators=(",", ":"), ensure_ascii=False)
+    # No mint-time length guard is needed: the per-field caps above bound the
+    # encoded length well below MAX_ENCODED_CURSOR_LENGTH (see its definition),
+    # so the encoder can never produce a cursor the decode path would reject.
+    return base64.urlsafe_b64encode(raw.encode("utf-8")).rstrip(b"=").decode("ascii")
+
+
+def encode_cursor_from_time(sort_field: str, t: datetime, id: str, order: str = "desc") -> str:
+    """Encode a time-typed cursor at Unix microsecond precision.
+
+    Accepts an aware datetime (any timezone) and normalizes to UTC. Naive
+    datetimes are rejected so callers can't accidentally encode the local
+    wall-clock value of a UTC-stored timestamp.
+    """
+    if t.tzinfo is None:
+        raise ValueError("encode_cursor_from_time requires an aware datetime")
+    micros = _datetime_to_unix_micros(t.astimezone(timezone.utc))
+    return encode_cursor(sort_field, str(micros), id, order=order)
+
+
+def decode_cursor(
+    cursor: str,
+    allowed_sort_fields: Iterable[str],
+    expected_order: str | None = None,
+) -> CursorPayload:
+    """Parse an opaque cursor.
+
+    ``allowed_sort_fields`` is the endpoint's accepted sort-field list — a
+    cursor carrying a field outside this set is rejected so a cursor minted
+    for one column can't be replayed against another (e.g. a ``created_at``
+    timestamp string compared against a ``name`` column).
+
+    ``expected_order`` (``"asc"``/``"desc"``), when supplied, must match the
+    payload's ``o`` field. ``o`` is required on every payload; a cursor
+    missing it is rejected as malformed.
+
+    Passing no allowed fields rejects every cursor.
+    """
+    if len(cursor) > MAX_ENCODED_CURSOR_LENGTH:
+        raise InvalidCursorError("cursor exceeds maximum length")
+
+    try:
+        # urlsafe_b64decode requires correct padding; we strip on encode, so
+        # restore the trailing '=' pad here.
+        padding = "=" * (-len(cursor) % 4)
+        raw = base64.urlsafe_b64decode(cursor + padding)
+    except (ValueError, base64.binascii.Error) as e:
+        raise InvalidCursorError(f"encoding: {e}") from e
+
+    try:
+        decoded = json.loads(raw)
+    except (json.JSONDecodeError, UnicodeDecodeError) as e:
+        raise InvalidCursorError(f"payload: {e}") from e
+
+    if not isinstance(decoded, dict):
+        raise InvalidCursorError("payload: expected object")
+
+    sort_field = decoded.get("s")
+    value = decoded.get("v")
+    id = decoded.get("id")
+    order = decoded.get("o")
+
+    if not isinstance(sort_field, str) or not isinstance(value, str) or not isinstance(id, str):
+        raise InvalidCursorError("payload: missing or non-string s/v/id")
+
+    if id == "":
+        raise InvalidCursorError("missing id")
+    if len(id) > MAX_CURSOR_ID_LENGTH:
+        raise InvalidCursorError("id exceeds maximum length")
+    if len(value) > MAX_CURSOR_VALUE_LENGTH:
+        raise InvalidCursorError("value exceeds maximum length")
+
+    if sort_field not in allowed_sort_fields:
+        raise InvalidCursorError(f"unsupported sort field {sort_field!r}")
+
+    if not isinstance(order, str):
+        raise InvalidCursorError("missing or non-string o")
+    if order not in _VALID_ORDERS:
+        raise InvalidCursorError(f"unsupported order {order!r}")
+    if expected_order is not None and order != expected_order:
+        raise InvalidCursorError(
+            f"cursor order {order!r} does not match request order {expected_order!r}"
+        )
+
+    return CursorPayload(sort_field=sort_field, value=value, id=id, order=order)
+
+
+def decode_cursor_time(payload: Optional[CursorPayload]) -> datetime:
+    """Parse a time-typed cursor value as Unix microseconds, returning UTC."""
+    if payload is None:
+        raise InvalidCursorError("nil cursor payload")
+    try:
+        micros = int(payload.value)
+    except ValueError as e:
+        raise InvalidCursorError(f"value is not a valid timestamp: {e}") from e
+    try:
+        return _unix_micros_to_datetime(micros)
+    except (OverflowError, OSError, ValueError) as e:
+        # Crafted out-of-range microseconds (e.g. > datetime.MAX_YEAR) blow up
+        # in fromtimestamp / datetime construction. Map to 400, not 500.
+        raise InvalidCursorError(f"value is out of representable range: {e}") from e
+
+
+def decode_cursor_int(payload: Optional[CursorPayload]) -> int:
+    """Parse a cursor value as a base-10 integer."""
+    if payload is None:
+        raise InvalidCursorError("nil cursor payload")
+    try:
+        return int(payload.value)
+    except ValueError as e:
+        raise InvalidCursorError(f"value is not a valid integer: {e}") from e
+
+
+_EPOCH = datetime(1970, 1, 1, tzinfo=timezone.utc)
+
+
+def _datetime_to_unix_micros(t: datetime) -> int:
+    """Convert an aware UTC datetime to Unix microseconds (integer math)."""
+    delta = t - _EPOCH
+    return (delta.days * 86_400 + delta.seconds) * 1_000_000 + delta.microseconds
+
+
+def _unix_micros_to_datetime(micros: int) -> datetime:
+    """Convert Unix microseconds to a UTC datetime, preserving precision."""
+    seconds, micro_remainder = divmod(micros, 1_000_000)
+    return datetime.fromtimestamp(seconds, tz=timezone.utc).replace(microsecond=micro_remainder)

app/assets/services/image_dimensions.py

@@ -0,0 +1,63 @@
+"""Image dimension extraction for asset ingest.
+
+Reads only the image header via Pillow to capture width/height cheaply,
+without a full pixel decode. Returns a metadata dict suitable for merging
+into ``AssetReference.system_metadata``.
+"""
+from __future__ import annotations
+
+import logging
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+def extract_image_dimensions(
+    file_path: str, mime_type: str | None = None
+) -> dict[str, Any] | None:
+    """Extract image dimensions for the file at ``file_path``.
+
+    Args:
+        file_path: Absolute path to a file on disk.
+        mime_type: Optional MIME type hint. When provided and not prefixed
+            with ``image/``, extraction is skipped without touching the file.
+
+    Returns:
+        ``{"kind": "image", "width": W, "height": H}`` when the file is a
+        recognizable image with positive dimensions, otherwise ``None``.
+
+    The dict shape is intended to be merged into ``system_metadata`` so the
+    asset response surfaces ``metadata.kind`` plus dimension fields for image
+    assets. Forward-compatible: future media kinds (e.g. ``"video"`` with
+    duration/fps) can extend this shape without schema changes.
+    """
+    if mime_type is not None and not mime_type.startswith("image/"):
+        return None
+
+    try:
+        from PIL import Image, UnidentifiedImageError
+    except ImportError:
+        logger.debug(
+            "Pillow not available; skipping image dimension extraction for %s",
+            file_path,
+        )
+        return None
+
+    try:
+        with Image.open(file_path) as img:
+            width, height = img.size
+    except (OSError, UnidentifiedImageError, ValueError) as exc:
+        logger.debug(
+            "Failed to read image dimensions from %s: %s", file_path, exc
+        )
+        return None
+
+    if (
+        not isinstance(width, int)
+        or not isinstance(height, int)
+        or width <= 0
+        or height <= 0
+    ):
+        return None
+
+    return {"kind": "image", "width": width, "height": height}

app/assets/services/ingest.py

@@ -17,9 +17,11 @@ from app.assets.database.queries import (
     get_reference_by_file_path,
     get_reference_tags,
     get_or_create_reference,
+    list_references_by_asset_id,
     reference_exists,
     remove_missing_tag_for_asset_id,
     set_reference_metadata,
+    set_reference_system_metadata,
     set_reference_tags,
     update_asset_hash_and_mime,
     upsert_asset,
@@ -29,6 +31,7 @@ from app.assets.database.queries import (
 from app.assets.helpers import get_utc_now, normalize_tags
 from app.assets.services.bulk_ingest import batch_insert_seed_assets
 from app.assets.services.file_utils import get_size_and_mtime_ns
+from app.assets.services.image_dimensions import extract_image_dimensions
 from app.assets.services.path_utils import (
     compute_relative_filename,
     get_name_and_tags_from_asset_path,
@@ -118,6 +121,14 @@ def _ingest_file_from_path(
                 user_metadata=user_metadata,
             )
 
+            _maybe_store_image_dimensions(
+                session,
+                reference_id=reference_id,
+                file_path=locator,
+                mime_type=mime_type,
+                current_system_metadata=ref.system_metadata,
+            )
+
         try:
             remove_missing_tag_for_asset_id(session, asset_id=asset.id)
         except Exception:
@@ -288,6 +299,13 @@ def _register_existing_asset(
                 user_metadata=new_meta,
             )
 
+        _backfill_image_dimensions_from_siblings(
+            session,
+            asset_id=asset.id,
+            new_reference_id=ref.id,
+            current_system_metadata=ref.system_metadata,
+        )
+
         if tags is not None:
             set_reference_tags(
                 session,
@@ -334,6 +352,87 @@ def _update_metadata_with_filename(
         )
 
 
+_IMAGE_DIMENSION_KEYS = ("kind", "width", "height")
+
+
+def _maybe_store_image_dimensions(
+    session: Session,
+    reference_id: str,
+    file_path: str,
+    mime_type: str | None,
+    current_system_metadata: dict | None,
+) -> None:
+    """Populate ``kind``/``width``/``height`` on system_metadata for image refs.
+
+    Non-image MIME types are a no-op. Pre-existing keys (e.g. enricher-written
+    safetensors metadata, download provenance) are preserved by merge.
+    """
+    if not mime_type or not mime_type.startswith("image/"):
+        return
+
+    dims = extract_image_dimensions(file_path, mime_type=mime_type)
+    if not dims:
+        return
+
+    current = current_system_metadata or {}
+    merged = dict(current)
+    merged.update(dims)
+    if merged != current:
+        set_reference_system_metadata(
+            session,
+            reference_id=reference_id,
+            system_metadata=merged,
+        )
+
+
+def _backfill_image_dimensions_from_siblings(
+    session: Session,
+    asset_id: str,
+    new_reference_id: str,
+    current_system_metadata: dict | None,
+) -> None:
+    """Copy image dimension keys from any sibling reference of the same asset.
+
+    The from-hash path doesn't read the file bytes, so dimensions can't be
+    extracted there directly. When another reference of the same asset already
+    carries image dimensions, copy them onto the new reference so consumers
+    see consistent metadata regardless of how the asset was registered.
+
+    Best-effort: missing siblings, non-image siblings, or absent dimension
+    keys leave the target reference unchanged.
+    """
+    current = current_system_metadata or {}
+    if current.get("kind") == "image" and "width" in current and "height" in current:
+        return
+
+    for sibling in list_references_by_asset_id(session, asset_id):
+        if sibling.id == new_reference_id:
+            continue
+        meta = sibling.system_metadata or {}
+        if meta.get("kind") != "image":
+            continue
+        width = meta.get("width")
+        height = meta.get("height")
+        if (
+            type(width) is not int
+            or type(height) is not int
+            or width <= 0
+            or height <= 0
+        ):
+            continue
+        merged = dict(current)
+        merged["kind"] = "image"
+        merged["width"] = width
+        merged["height"] = height
+        if merged != current:
+            set_reference_system_metadata(
+                session,
+                reference_id=new_reference_id,
+                system_metadata=merged,
+            )
+        return
+
+
 def _sanitize_filename(name: str | None, fallback: str) -> str:
     n = os.path.basename((name or "").strip() or fallback)
     return n if n else fallback

app/assets/services/metadata_extract.py

@@ -4,7 +4,6 @@ Tier 1: Filesystem metadata (zero parsing)
 Tier 2: Safetensors header metadata (fast JSON read only)
 """
 
-from __future__ import annotations
 
 import json
 import logging

app/assets/services/schemas.py

@@ -56,7 +56,6 @@ class IngestResult:
 
 class TagUsage(NamedTuple):
     name: str
-    tag_type: str
     count: int
 
 
@@ -71,6 +70,7 @@ class AssetSummaryData:
 class ListAssetsResult:
     items: list[AssetSummaryData]
     total: int
+    next_cursor: str | None = None
 
 
 @dataclass(frozen=True)

app/assets/services/tagging.py

@@ -75,7 +75,7 @@ def list_tags(
             owner_id=owner_id,
         )
 
-    return [TagUsage(name, tag_type, count) for name, tag_type, count in rows], total
+    return [TagUsage(name, count) for name, count in rows], total
 
 
 def list_tag_histogram(

app/custom_node_manager.py

@@ -1,5 +1,3 @@
-from __future__ import annotations
-
 import os
 import folder_paths
 import glob

app/frontend_management.py

@@ -1,4 +1,3 @@
-from __future__ import annotations
 import argparse
 import logging
 import os
@@ -62,6 +61,8 @@ def get_comfy_package_versions():
 def check_comfy_packages_versions():
     """Warn for every comfy* package whose installed version is below requirements.txt."""
     from packaging.version import InvalidVersion, parse as parse_pep440
+    outdated_packages = []
+
     for pkg in get_comfy_package_versions():
         installed_str = pkg["installed"]
         required_str = pkg["required"]
@@ -73,19 +74,26 @@ def check_comfy_packages_versions():
             logging.error(f"Failed to check {pkg['name']} version: {e}")
             continue
         if outdated:
-            app.logger.log_startup_warning(
-                f"""
+            outdated_packages.append((pkg["name"], installed_str, required_str))
+        else:
+            logging.info("{} version: {}".format(pkg["name"], installed_str))
+
+    if outdated_packages:
+        package_warnings = "\n".join(
+            f"Installed {name} version {installed} is lower than the recommended version {required}."
+            for name, installed, required in outdated_packages
+        )
+        app.logger.log_startup_warning(
+            f"""
 ________________________________________________________________________
 WARNING WARNING WARNING WARNING WARNING
 
-Installed {pkg["name"]} version {installed_str} is lower than the recommended version {required_str}.
+{package_warnings}
 
 {get_missing_requirements_message()}
 ________________________________________________________________________
 """.strip()
-            )
-        else:
-            logging.info("{} version: {}".format(pkg["name"], installed_str))
+        )
 
 
 REQUEST_TIMEOUT = 10  # seconds

app/logger.py

@@ -5,6 +5,40 @@ import logging
 import sys
 import threading
 
+ANSI_NAMED_COLORS = {
+    'black':   '\033[30m',
+    'red':     '\033[31m',
+    'green':   '\033[32m',
+    'yellow':  '\033[33m',
+    'blue':    '\033[34m',
+    'magenta': '\033[35m',
+    'cyan':    '\033[36m',
+    'white':   '\033[37m',
+}
+
+ANSI_LEVEL_COLORS = {
+    'DEBUG':    ANSI_NAMED_COLORS['cyan'],
+    'INFO':     ANSI_NAMED_COLORS['green'],
+    'WARNING':  ANSI_NAMED_COLORS['yellow'],
+    'ERROR':    ANSI_NAMED_COLORS['red'],
+    'CRITICAL': ANSI_NAMED_COLORS['magenta'],
+}
+
+ANSI_RESET = '\033[0m'
+ANSI_BOLD  = '\033[1m'
+
+
+class ColoredFormatter(logging.Formatter):
+    def format(self, record):
+        color = ANSI_LEVEL_COLORS.get(record.levelname, '')
+        bold  = ANSI_BOLD if record.levelno >= logging.WARNING else ''
+        level_tag = f"{bold}{color}[{record.levelname}]{ANSI_RESET} "
+        message = super().format(record)
+        line_color = ANSI_NAMED_COLORS.get(getattr(record, 'color', ''), '')
+        if line_color:
+            return f"{level_tag}{line_color}{message}{ANSI_RESET}"
+        return level_tag + message
+
 logs = None
 stdout_interceptor = None
 stderr_interceptor = None
@@ -68,8 +102,10 @@ def setup_logger(log_level: str = 'INFO', capacity: int = 300, use_stdout: bool
     logger = logging.getLogger()
     logger.setLevel(log_level)
 
+    formatter = ColoredFormatter("%(message)s")
+
     stream_handler = logging.StreamHandler()
-    stream_handler.setFormatter(logging.Formatter("%(message)s"))
+    stream_handler.setFormatter(formatter)
 
     if use_stdout:
         # Only errors and critical to stderr
@@ -77,7 +113,7 @@ def setup_logger(log_level: str = 'INFO', capacity: int = 300, use_stdout: bool
 
         # Lesser to stdout
         stdout_handler = logging.StreamHandler(sys.stdout)
-        stdout_handler.setFormatter(logging.Formatter("%(message)s"))
+        stdout_handler.setFormatter(formatter)
         stdout_handler.addFilter(lambda record: record.levelno < logging.ERROR)
         logger.addHandler(stdout_handler)
 

app/model_downloader/allowlist.py

@@ -0,0 +1,51 @@
+"""URL allowlist for server-side model fetches.
+
+Mirrors the frontend's ``isModelDownloadable`` allowlist so the two flows
+agree on which URLs are eligible for download. Server-side allowlisting is
+the primary SSRF defense for this subsystem — workflow JSON is untrusted
+input (anyone can hand-craft one), so we never let the server fetch URLs
+outside this list.
+"""
+
+from urllib.parse import urlparse
+
+# Frontend parity: ``missingModelDownload-*.js`` exports the same triple
+# (Civitai / HuggingFace / localhost). Keyed by exact hostname → allowed
+# schemes, and matched against the *parsed* host (not a raw string prefix),
+# so URL-userinfo tricks can't slip past — see ``is_url_allowed``.
+_ALLOWED_HOSTS = {
+    "huggingface.co": {"https"},
+    "civitai.com": {"https"},
+    "localhost": {"http"},
+    "127.0.0.1": {"http"},
+}
+
+# Frontend parity: same set as ``a = [...]`` in the bundle.
+_ALLOWED_MODEL_EXTENSIONS = (
+    ".safetensors",
+    ".sft",
+    ".ckpt",
+    ".pth",
+    ".pt",
+)
+
+
+def is_url_allowed(url: str) -> bool:
+    """Check whether ``url`` is permitted as a server-side download source.
+
+    True only when the parsed host + scheme are allowlisted AND the path ends
+    in a model extension. Matching on ``parsed.hostname`` (not a string prefix)
+    defeats userinfo tricks like ``http://127.0.0.1:80@169.254.169.254/x.safetensors``,
+    whose real host is ``169.254.169.254``; the extension check rejects non-model
+    URLs on allowed hosts (e.g. ``huggingface.co/api/...``).
+    """
+    if not isinstance(url, str) or not url:
+        return False
+    try:
+        parsed = urlparse(url)
+    except ValueError:
+        return False
+    host = parsed.hostname
+    if host is None or parsed.scheme not in _ALLOWED_HOSTS.get(host, ()):
+        return False
+    return any(parsed.path.endswith(ext) for ext in _ALLOWED_MODEL_EXTENSIONS)

app/model_downloader/api/routes.py

@@ -0,0 +1,359 @@
+"""Aiohttp routes for the server-side model download subsystem.
+
+Endpoint surface (all under ``/api/``, all kebab-case):
+
+  - ``POST /api/models-availability-status``     — bulk status + metadata query.
+  - ``POST /api/download-models``                — start a batch of downloads.
+  - ``POST /api/cancel-model-download-session``  — cancel a single in-flight one.
+  - ``GET  /api/hf-auth-token-status``           — current HF login state.
+  - ``POST /api/hf-auth-login-start``            — begin the HF OAuth flow.
+  - ``POST /api/hf-auth-logout``                 — drop the stored HF token.
+
+The contract is intentionally narrow: only model_ids of the form
+``<directory>/<filename>`` (validated via ``app.model_downloader.paths``)
+are accepted, and only URLs on the same allowlist the frontend already
+uses (HuggingFace, Civitai, localhost) can be fetched. Both are required
+to keep the server out of the SSRF business for this feature.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+import logging
+from typing import Any, Literal, Optional
+
+from aiohttp import web
+from pydantic import BaseModel, ValidationError
+
+from app.model_downloader.allowlist import is_url_allowed
+from app.model_downloader.download_server import (
+    DOWNLOAD_SERVER,
+    DownloadSession,
+)
+from app.model_downloader.downloader import schedule_batch
+from app.model_downloader.gated_detection import probe_url
+from app.model_downloader.hf_auth.auth_store import HF_AUTH_STORE
+from app.model_downloader.hf_auth.eligibility import is_hf_auth_eligible
+from app.model_downloader.hf_auth.oauth import (
+    OAuthCallbackError,
+    OAuthInProgressError,
+    start_login_flow,
+)
+from app.model_downloader.paths import (
+    InvalidModelId,
+    parse_model_id,
+    resolve_existing,
+)
+from app.model_downloader.api import schemas_in, schemas_out
+
+ROUTES = web.RouteTableDef()
+
+
+def register_routes(app: web.Application) -> None:
+    """Wire the model-downloader routes into the running aiohttp app.
+
+    Called once from ``server.py`` during ``PromptServer`` startup.
+    """
+    app.add_routes(ROUTES)
+
+
+# ----- response helpers (same envelope as app/assets/api/routes.py) -----
+
+
+ErrorCode = Literal[
+    "INVALID_JSON",
+    "INVALID_BODY",
+    "EMPTY_REQUEST",
+    "INVALID_MODEL_ID",
+    "URL_NOT_ALLOWED",
+    "ALREADY_AVAILABLE",
+    "ALREADY_DOWNLOADING",
+    "MODEL_NOT_DOWNLOADABLE",
+    "NOT_DOWNLOADING",
+    "HF_AUTH_NOT_ELIGIBLE",
+    "HF_AUTH_IN_PROGRESS",
+    "HF_AUTH_START_FAILED",
+]
+
+
+def _error(status: int, code: ErrorCode, message: str, details: dict | None = None) -> web.Response:
+    return web.json_response(
+        {"error": {"code": code, "message": message, "details": details or {}}},
+        status=status,
+    )
+
+
+def _validation_error(code: ErrorCode, ve: ValidationError) -> web.Response:
+    return _error(400, code, "Validation failed.", {"errors": json.loads(ve.json())})
+
+
+def _ok(payload: BaseModel, status: int = 200) -> web.Response:
+    return web.json_response(
+        payload.model_dump(mode="json", exclude_none=False),
+        status=status,
+    )
+
+
+async def _parse_body(request: web.Request, model: type[BaseModel]) -> Any:
+    """Parse a JSON body into a pydantic model or raise a 400 response."""
+    try:
+        raw = await request.json()
+    except json.JSONDecodeError:
+        return _error(400, "INVALID_JSON", "Request body must be valid JSON.")
+    try:
+        return model.model_validate(raw)
+    except ValidationError as ve:
+        return _validation_error("INVALID_BODY", ve)
+
+
+# ----- 1. availability status (unified: state + metadata per id) -----
+
+
+@ROUTES.post("/api/models-availability-status")
+async def models_availability_status(request: web.Request) -> web.Response:
+    """Return per-id ``{state, progress, file_size, is_hf_downloadable}``.
+
+    State (``available`` / ``missing`` / ``downloading``) is cheap to
+    recompute per call. ``file_size`` and ``is_gated`` are cached
+    server-side per URL. ``is_hf_downloadable`` is recomputed every
+    call from the current token state — that's what makes login + license
+    acceptance show up in the UI within one poll cycle without any
+    frontend cache plumbing.
+    """
+    parsed = await _parse_body(request, schemas_in.AvailabilityStatusRequest)
+    if isinstance(parsed, web.Response):
+        return parsed
+
+    items = list(parsed.models.items())
+
+    # Run all probes concurrently; each is internally cached per URL.
+    probes = await asyncio.gather(*(probe_url(url) for _, url in items))
+
+    response_models: dict[str, schemas_out.ModelStatusEntry] = {}
+    for (model_id, _url), probe in zip(items, probes):
+        try:
+            parse_model_id(model_id)
+        except InvalidModelId:
+            # Ill-formed identifier: report as missing without 400-ing the
+            # whole batch — the workflow author probably typo'd.
+            response_models[model_id] = schemas_out.ModelStatusEntry(
+                state="missing",
+                file_size=probe.file_size,
+                is_hf_downloadable=probe.is_hf_downloadable,
+            )
+            continue
+
+        active = DOWNLOAD_SERVER.get(model_id)
+        if active is not None:
+            response_models[model_id] = schemas_out.ModelStatusEntry(
+                state="downloading",
+                progress=schemas_out.DownloadProgress(
+                    bytes_downloaded=active.bytes_downloaded,
+                    total_bytes=active.total_bytes,
+                    progress=active.progress,
+                ),
+                file_size=probe.file_size,
+                is_hf_downloadable=probe.is_hf_downloadable,
+            )
+            continue
+
+        state: schemas_out.ModelState = (
+            "available" if resolve_existing(model_id) is not None else "missing"
+        )
+        response_models[model_id] = schemas_out.ModelStatusEntry(
+            state=state,
+            file_size=probe.file_size,
+            is_hf_downloadable=probe.is_hf_downloadable,
+        )
+
+    return _ok(schemas_out.AvailabilityStatusResponse(
+        models=response_models,
+        hf_auth=schemas_out.HfAuthStatus(
+            token_available=HF_AUTH_STORE.has_token(),
+            eligible=is_hf_auth_eligible(),
+        ),
+    ))
+
+
+# ----- 2. start downloads -----
+
+
+@ROUTES.post("/api/download-models")
+async def download_models(request: web.Request) -> web.Response:
+    parsed = await _parse_body(request, schemas_in.DownloadModelsRequest)
+    if isinstance(parsed, web.Response):
+        return parsed
+
+    if not parsed.models:
+        return _error(400, "EMPTY_REQUEST", "No models supplied.")
+
+    # ----- precondition pass: validate everything BEFORE registering anything -----
+    # Atomic semantics: if any model fails any precondition (invalid id,
+    # not allow-listed URL, already on disk, already downloading, or gated),
+    # the entire request fails and no state is changed.
+    requested = list(parsed.models.items())
+
+    for model_id, url in requested:
+        try:
+            parse_model_id(model_id)
+        except InvalidModelId as e:
+            return _error(400, "INVALID_MODEL_ID", str(e),
+                          {"model_id": model_id})
+
+        if not is_url_allowed(url):
+            return _error(
+                400, "URL_NOT_ALLOWED",
+                "Server-side downloads only accept HuggingFace, Civitai, "
+                "or localhost URLs ending in a known model extension.",
+                {"model_id": model_id, "url": url},
+            )
+
+        if resolve_existing(model_id) is not None:
+            return _error(409, "ALREADY_AVAILABLE",
+                          f"Model already exists on disk: {model_id}",
+                          {"model_id": model_id})
+
+        if DOWNLOAD_SERVER.is_downloading(model_id):
+            return _error(409, "ALREADY_DOWNLOADING",
+                          f"A download for {model_id} is already in progress.",
+                          {"model_id": model_id})
+
+    # Reachability check last — it's the only one that talks to the
+    # network. Concurrent probes. For HF URLs ``is_hf_downloadable``
+    # reflects current token access; for non-HF URLs it's None, and we
+    # treat that as "no info, proceed".
+    probes = await asyncio.gather(*(probe_url(url) for _, url in requested))
+    for (model_id, url), probe in zip(requested, probes):
+        if probe.is_hf_downloadable is False:
+            return _error(
+                400, "MODEL_NOT_DOWNLOADABLE",
+                f"Model {model_id} is gated on HuggingFace and the current "
+                f"server token (if any) does not grant access.",
+                {"model_id": model_id, "url": url},
+            )
+
+    # ----- registration pass: try_register is atomic per model_id -----
+    # Defensive: another request might have raced past our pre-check
+    # between the loop above and here. try_register handles that.
+    sessions: list[DownloadSession] = []
+    for model_id, url in requested:
+        session = DOWNLOAD_SERVER.try_register(model_id, url)
+        if session is None:
+            # Race: someone else got in. Roll back what we registered.
+            for s in sessions:
+                DOWNLOAD_SERVER.cancel(s.model_id)
+            return _error(409, "ALREADY_DOWNLOADING",
+                          f"A download for {model_id} is already in progress (race).",
+                          {"model_id": model_id})
+        sessions.append(session)
+
+    DOWNLOAD_SERVER.sweep_orphan_tmp_files()
+    schedule_batch(sessions)
+    logging.info(
+        "[model_downloader] scheduled %d downloads: %s",
+        len(sessions), [s.model_id for s in sessions],
+    )
+
+    return _ok(schemas_out.DownloadModelsResponse(
+        accepted=True,
+        scheduled=[s.model_id for s in sessions],
+    ), status=202)
+
+
+# ----- 3. cancel a session -----
+
+
+@ROUTES.post("/api/cancel-model-download-session")
+async def cancel_model_download_session(request: web.Request) -> web.Response:
+    parsed = await _parse_body(request, schemas_in.CancelDownloadSessionRequest)
+    if isinstance(parsed, web.Response):
+        return parsed
+
+    try:
+        parse_model_id(parsed.model_id)
+    except InvalidModelId as e:
+        return _error(400, "INVALID_MODEL_ID", str(e), {"model_id": parsed.model_id})
+
+    cancelled = DOWNLOAD_SERVER.cancel(parsed.model_id)
+    if not cancelled:
+        return _error(404, "NOT_DOWNLOADING",
+                      f"No active download for {parsed.model_id}.",
+                      {"model_id": parsed.model_id})
+
+    return _ok(schemas_out.CancelDownloadSessionResponse(cancelled=True))
+
+
+# ----- 4. HuggingFace OAuth status / login start / logout -----
+
+
+@ROUTES.get("/api/hf-auth-token-status")
+async def hf_auth_token_status(request: web.Request) -> web.Response:
+    """Return whether the server holds a usable HF token + its username.
+
+    Used by the settings UI and (out-of-band) by the frontend on
+    login completion. ``token_available`` is true even if the cached
+    access_token is expired — as long as a refresh_token exists, the
+    user is "logged in" from their perspective.
+    """
+    token_present = HF_AUTH_STORE.has_token()
+    username: Optional[str] = None
+    if token_present:
+        # Resolve the username via whoami. Done in a worker thread because
+        # huggingface_hub's whoami is synchronous + blocks on a network call.
+        tok = await HF_AUTH_STORE.get_valid_token()
+        if tok is not None:
+            try:
+                username = await asyncio.to_thread(_whoami_username, tok.access_token)
+            except Exception as e:
+                logging.debug("[hf_auth] whoami failed: %s", e)
+    return _ok(schemas_out.HfAuthTokenStatusResponse(
+        token_available=token_present,
+        username=username,
+    ))
+
+
+def _whoami_username(token: str) -> Optional[str]:
+    """Sync helper: ask HF for the user name attached to a token."""
+    from huggingface_hub import HfApi
+    info = HfApi().whoami(token=token)
+    if isinstance(info, dict):
+        return info.get("name") or info.get("fullname")
+    return None
+
+
+@ROUTES.post("/api/hf-auth-login-start")
+async def hf_auth_login_start(request: web.Request) -> web.Response:
+    """Begin one OAuth attempt: bind the callback port, return the URL.
+
+    Rejected outright if this deployment isn't eligible (we don't
+    surface the option on multi-tenant / public-IP installs).
+    """
+    if not is_hf_auth_eligible():
+        return _error(
+            403, "HF_AUTH_NOT_ELIGIBLE",
+            "This server is not eligible for interactive HuggingFace login. "
+            "It must be bound to a loopback address and not running in "
+            "--multi-user mode.",
+        )
+    try:
+        url = await start_login_flow()
+    except OAuthInProgressError:
+        return _error(
+            409, "HF_AUTH_IN_PROGRESS",
+            "Another HuggingFace login attempt is in progress. Try again "
+            "after it completes or times out.",
+        )
+    except OAuthCallbackError as e:
+        return _error(
+            503, "HF_AUTH_START_FAILED",
+            f"Could not start the HuggingFace login flow: {e}",
+        )
+    return _ok(schemas_out.HfAuthLoginStartResponse(authorize_url=url))
+
+
+@ROUTES.post("/api/hf-auth-logout")
+async def hf_auth_logout(request: web.Request) -> web.Response:
+    """Drop the in-memory + on-disk HF token."""
+    HF_AUTH_STORE.clear()
+    return _ok(schemas_out.HfAuthLogoutResponse(logged_out=True))

app/model_downloader/api/schemas_in.py

@@ -0,0 +1,41 @@
+"""Request schemas for the model-downloader API.
+
+Each endpoint accepts a small JSON body. Pydantic enforces the shape at
+the boundary; route handlers operate only on validated values past that.
+"""
+
+from __future__ import annotations
+
+from pydantic import BaseModel, Field
+
+
+class AvailabilityStatusRequest(BaseModel):
+    """``POST /api/models-availability-status``.
+
+    Sent by the frontend on each poll. Each entry is ``{model_id: url}``;
+    the URL is the one declared in ``properties.models[i].url`` in the
+    workflow JSON and lets the server compute per-id metadata
+    (``file_size`` + ``is_hf_downloadable``) on the same request.
+    """
+    models: dict[str, str] = Field(default_factory=dict)
+
+
+class DownloadModelsRequest(BaseModel):
+    """``POST /api/download-models``.
+
+    Same shape as the metadata request — the URL for each model_id.
+    Returns immediately after validation and scheduling.
+    """
+    models: dict[str, str] = Field(default_factory=dict)
+
+
+class CancelDownloadSessionRequest(BaseModel):
+    """``POST /api/cancel-model-download-session``."""
+    model_id: str
+
+
+__all__ = [
+    "AvailabilityStatusRequest",
+    "DownloadModelsRequest",
+    "CancelDownloadSessionRequest",
+]

app/model_downloader/api/schemas_out.py

@@ -0,0 +1,81 @@
+"""Response schemas for the model-downloader API."""
+
+from __future__ import annotations
+
+from typing import Literal, Optional
+
+from pydantic import BaseModel
+
+
+ModelState = Literal["available", "missing", "downloading"]
+
+
+class DownloadProgress(BaseModel):
+    """Embedded in a model entry when its state is ``downloading``."""
+    bytes_downloaded: int
+    total_bytes: Optional[int] = None
+    progress: Optional[float] = None  # fraction in [0,1]; null until total known
+
+
+class ModelStatusEntry(BaseModel):
+    """Everything the UI needs to render one row, in one shot.
+
+    ``state`` reflects what the server has on disk + in-flight; ``file_size``
+    and ``is_hf_downloadable`` come from probes (intrinsic; cached).
+    The HF fields are populated for every poll (cached on the server),
+    so license-acceptance flips show up within one poll interval without
+    any frontend cache invalidation.
+    """
+    state: ModelState
+    progress: Optional[DownloadProgress] = None
+    file_size: Optional[int] = None
+    # HF-only: True iff the server can fetch this URL with current auth
+    # state. False iff gated and lacking access. None for non-HF URLs.
+    is_hf_downloadable: Optional[bool] = None
+
+
+class HfAuthStatus(BaseModel):
+    """Snapshot of HF login state, embedded in availability response."""
+    token_available: bool
+    eligible: bool
+
+
+class AvailabilityStatusResponse(BaseModel):
+    models: dict[str, ModelStatusEntry]
+    hf_auth: HfAuthStatus
+
+
+class DownloadModelsResponse(BaseModel):
+    accepted: bool
+    scheduled: list[str]
+
+
+class CancelDownloadSessionResponse(BaseModel):
+    cancelled: bool
+
+
+class HfAuthTokenStatusResponse(BaseModel):
+    token_available: bool
+    username: Optional[str] = None
+
+
+class HfAuthLoginStartResponse(BaseModel):
+    authorize_url: str
+
+
+class HfAuthLogoutResponse(BaseModel):
+    logged_out: bool
+
+
+__all__ = [
+    "ModelState",
+    "DownloadProgress",
+    "ModelStatusEntry",
+    "HfAuthStatus",
+    "AvailabilityStatusResponse",
+    "DownloadModelsResponse",
+    "CancelDownloadSessionResponse",
+    "HfAuthTokenStatusResponse",
+    "HfAuthLoginStartResponse",
+    "HfAuthLogoutResponse",
+]

app/model_downloader/download_server.py

@@ -0,0 +1,179 @@
+"""Process-wide registry of in-flight model downloads.
+
+A single instance, ``DOWNLOAD_SERVER``, tracks every currently-running
+server-side model fetch. Designed to be safe with multiple concurrent
+clients hitting the API: each model_id has at most one active session,
+and the API rejects requests that conflict with in-flight downloads.
+
+Cancellation is cooperative — the download loop checks ``is_active`` on
+its own session between chunks and raises ``DownloadCancelled`` when the
+session has been removed from the registry. This avoids the complications
+of ``Task.cancel()`` from outside the loop while still giving deterministic
+rollback semantics (the worker is responsible for deleting its own
+``.tmp`` on the cancel path).
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+import threading
+from dataclasses import dataclass, field
+from typing import Optional
+
+from app.model_downloader.paths import iter_all_tmp_paths
+
+
+class DownloadCancelled(Exception):
+    """Raised by the streaming loop when its session has been removed
+    from the registry (cancellation request) and the worker should roll
+    back its ``.tmp`` file."""
+
+
+@dataclass
+class DownloadSession:
+    """One in-flight download.
+
+    ``progress`` is a fraction in ``[0.0, 1.0]``; ``None`` until the first
+    byte arrives and we know whether the response carries a
+    ``Content-Length``. ``total_bytes`` mirrors that header when present.
+    """
+    model_id: str
+    url: str
+    progress: Optional[float] = None
+    bytes_downloaded: int = 0
+    total_bytes: Optional[int] = None
+    # Sequence number used solely as identity for the cancellation check —
+    # so that "cancel + restart" doesn't get confused by stale workers.
+    epoch: int = field(default_factory=lambda: 0)
+
+
+class DownloadServer:
+    """Singleton registry of active downloads.
+
+    All mutation goes through this object so concurrent route handlers
+    see a consistent view. The ``_lock`` is a plain threading lock
+    because the registry is consulted from both the asyncio event-loop
+    thread (route handlers) and from any worker coroutines spawned to
+    perform downloads.
+    """
+
+    def __init__(self) -> None:
+        self._lock = threading.Lock()
+        self._sessions: dict[str, DownloadSession] = {}
+        self._epoch_counter = 0
+        self._orphan_sweep_done = False
+
+    # ----- lifecycle -----
+
+    def sweep_orphan_tmp_files(self) -> None:
+        """Idempotently sweep ``*.tmp`` files left by crashed downloads.
+
+        Deferred off the import path so module load doesn't block on
+        filesystem I/O against potentially-slow mounts. Each route handler
+        that might create a new ``.tmp`` runs this exactly once.
+        """
+        with self._lock:
+            if self._orphan_sweep_done:
+                return
+            self._orphan_sweep_done = True
+        for path in iter_all_tmp_paths():
+            try:
+                os.remove(path)
+                logging.info("[model_downloader] removed orphan tmp file: %s", path)
+            except OSError as e:
+                logging.warning("[model_downloader] could not remove %s: %s", path, e)
+
+    # ----- queries -----
+
+    def is_downloading(self, model_id: str) -> bool:
+        with self._lock:
+            return model_id in self._sessions
+
+    def get(self, model_id: str) -> Optional[DownloadSession]:
+        with self._lock:
+            return self._sessions.get(model_id)
+
+    def snapshot(self) -> dict[str, DownloadSession]:
+        """Return a shallow copy of the current sessions map."""
+        with self._lock:
+            return dict(self._sessions)
+
+    # ----- mutations -----
+
+    def try_register(self, model_id: str, url: str) -> Optional[DownloadSession]:
+        """Atomically register a new session iff none exists for ``model_id``.
+
+        Returns the new session on success, ``None`` if a session is already
+        in flight. Callers must check the return value — the caller is the
+        sole owner of the session it gets back.
+        """
+        with self._lock:
+            if model_id in self._sessions:
+                return None
+            self._epoch_counter += 1
+            session = DownloadSession(
+                model_id=model_id,
+                url=url,
+                epoch=self._epoch_counter,
+            )
+            self._sessions[model_id] = session
+            return session
+
+    def update_progress(
+        self,
+        session: DownloadSession,
+        bytes_downloaded: int,
+        total_bytes: Optional[int],
+    ) -> None:
+        """Update progress on a session. No-op if the session has been
+        removed (cancelled) — caller should check ``is_active`` separately."""
+        with self._lock:
+            current = self._sessions.get(session.model_id)
+            if current is None or current.epoch != session.epoch:
+                return
+            current.bytes_downloaded = bytes_downloaded
+            current.total_bytes = total_bytes
+            if total_bytes and total_bytes > 0:
+                current.progress = min(1.0, bytes_downloaded / total_bytes)
+
+    def is_active(self, session: DownloadSession) -> bool:
+        """True iff this exact session is still the registered one for
+        its model_id. False after cancellation, after completion, or if
+        another session has replaced it."""
+        with self._lock:
+            current = self._sessions.get(session.model_id)
+            return current is not None and current.epoch == session.epoch
+
+    def finish(self, session: DownloadSession) -> None:
+        """Remove a completed (or cancelled) session from the registry.
+
+        Safe to call multiple times. Only removes if the epoch matches —
+        we never accidentally evict a *newer* session for the same model_id.
+        """
+        with self._lock:
+            current = self._sessions.get(session.model_id)
+            if current is not None and current.epoch == session.epoch:
+                del self._sessions[session.model_id]
+
+    def reset_for_tests(self) -> None:
+        """Clear all sessions and reset the epoch counter. Test-only."""
+        with self._lock:
+            self._sessions.clear()
+            self._epoch_counter = 0
+
+    def cancel(self, model_id: str) -> bool:
+        """Remove the session registered for ``model_id``.
+
+        Returns True if there was an active session to cancel. The worker
+        will discover the cancellation on its next ``is_active`` check
+        and roll back its ``.tmp`` file.
+        """
+        with self._lock:
+            if model_id in self._sessions:
+                del self._sessions[model_id]
+                return True
+            return False
+
+
+DOWNLOAD_SERVER = DownloadServer()

app/model_downloader/downloader.py

@@ -0,0 +1,216 @@
+"""Streaming download worker with progress reporting and cancellation.
+
+Each download writes to ``<final_path>.tmp`` and atomically renames into
+place on success. Between chunks the worker checks the registry for
+cancellation (via ``DownloadServer.is_active``) and rolls back its
+``.tmp`` on cancel or on any error.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+import os
+from typing import Optional
+
+import aiohttp
+
+from app.model_downloader.download_server import (
+    DOWNLOAD_SERVER,
+    DownloadCancelled,
+    DownloadSession,
+)
+from app.model_downloader.hf_auth.auth_store import HF_AUTH_STORE
+from app.model_downloader.hf_url import is_hf_url
+from app.model_downloader.http_client import get_session, parse_content_length
+from app.model_downloader.paths import resolve_destination
+
+
+CHUNK_SIZE = 64 * 1024  # 64 KiB — same scale as other ComfyUI download paths.
+REQUEST_TIMEOUT = aiohttp.ClientTimeout(total=None, sock_connect=30, sock_read=120)
+
+
+async def stream_to_disk(session: DownloadSession) -> str:
+    """Run a single download to completion or cancellation.
+
+    Returns the final on-disk path on success. Removes the ``.tmp`` and
+    raises on cancellation or failure. The session is finished
+    (removed from the registry) exactly once, here — callers do not
+    need to call ``DOWNLOAD_SERVER.finish`` themselves.
+    """
+    final_path, tmp_path = resolve_destination(session.model_id, session.epoch)
+    os.makedirs(os.path.dirname(final_path), exist_ok=True)
+
+    # Wipe any stale .tmp from a previous failed attempt before we start —
+    # otherwise a partial body could masquerade as our completed download
+    # when the rename finally happens.
+    _remove_if_exists(tmp_path)
+
+    bytes_seen = 0
+    try:
+        http = await get_session()
+        headers = _auth_headers_for(session.url)
+        logging.info(
+            "[model_downloader] starting GET %s (auth=%s)",
+            session.url, "yes" if "Authorization" in headers else "no",
+        )
+        async with http.get(
+            session.url,
+            allow_redirects=True,
+            timeout=REQUEST_TIMEOUT,
+            headers=headers,
+        ) as resp:
+            if resp.status != 200:
+                # Capture a snippet of the response body so 4xx/5xx aren't
+                # opaque in the logs — HF returns JSON or HTML with a
+                # human-readable reason on failures.
+                body_snippet = await _read_short(resp)
+                logging.warning(
+                    "[model_downloader] GET %s failed: status=%d final_url=%s body=%s",
+                    session.url, resp.status, str(resp.url), body_snippet,
+                )
+                raise DownloadError(
+                    f"unexpected HTTP {resp.status} fetching {session.url}: {body_snippet}",
+                    status=resp.status,
+                )
+
+            total = parse_content_length(resp.headers.get("Content-Length"))
+            DOWNLOAD_SERVER.update_progress(session, 0, total)
+
+            with open(tmp_path, "wb") as f:
+                async for chunk in resp.content.iter_chunked(CHUNK_SIZE):
+                    # Cancellation check between chunks. Cheap and means
+                    # cancellation latency is bounded by one chunk plus
+                    # one ``write()`` — typically well under a second
+                    # even on slow disks.
+                    if not DOWNLOAD_SERVER.is_active(session):
+                        raise DownloadCancelled()
+                    f.write(chunk)
+                    bytes_seen += len(chunk)
+                    DOWNLOAD_SERVER.update_progress(session, bytes_seen, total)
+
+        # Final cancellation check before we promote the .tmp to the real
+        # filename — avoids the awkward case where cancel arrives during
+        # the very last chunk and we'd otherwise commit anyway.
+        if not DOWNLOAD_SERVER.is_active(session):
+            raise DownloadCancelled()
+
+        # Size verification before commit. aiohttp already raises
+        # ClientPayloadError on a truncated Content-Length/chunked body,
+        # but this also catches the HTTP/1.0-style case (no Content-Length
+        # + Connection: close) where a short read can masquerade as a
+        # complete download.
+        if total is not None and bytes_seen != total:
+            raise DownloadError(
+                f"size mismatch for {session.model_id}: "
+                f"got {bytes_seen} of {total} bytes from {session.url}"
+            )
+
+        # Atomic rename. os.replace is atomic within the same filesystem,
+        # which is guaranteed here because tmp lives alongside final_path.
+        os.replace(tmp_path, final_path)
+        logging.info(
+            "[model_downloader] downloaded %s (%d bytes) from %s",
+            session.model_id, bytes_seen, session.url,
+        )
+        return final_path
+
+    except DownloadCancelled:
+        logging.info("[model_downloader] cancelled: %s", session.model_id)
+        _remove_if_exists(tmp_path)
+        raise
+    except Exception as e:
+        logging.warning(
+            "[model_downloader] failed: %s from %s: %s: %s",
+            session.model_id, session.url, type(e).__name__, e,
+            exc_info=True,
+        )
+        _remove_if_exists(tmp_path)
+        raise
+    finally:
+        # In all terminal states (success / cancel / error) drop the
+        # session from the registry. Idempotent — only removes if we're
+        # still the live epoch for this model_id.
+        DOWNLOAD_SERVER.finish(session)
+
+
+class DownloadError(Exception):
+    """Network / protocol error during a download."""
+
+    def __init__(self, message: str, status: Optional[int] = None) -> None:
+        super().__init__(message)
+        self.status = status
+
+
+async def _read_short(resp: aiohttp.ClientResponse, limit: int = 512) -> str:
+    """Read up to ``limit`` bytes of a response body for logging.
+
+    Used to surface the JSON/HTML reason from an HF non-2xx response in
+    server logs instead of just the status code. Best-effort: any
+    error here is swallowed.
+    """
+    try:
+        raw = await resp.content.read(limit)
+        return raw.decode("utf-8", errors="replace").strip()
+    except Exception:
+        return "<unreadable>"
+
+
+def _auth_headers_for(url: str) -> dict[str, str]:
+    """Return any auth headers we should add to the GET for ``url``.
+
+    For HuggingFace URLs we inject the user's OAuth access token as a
+    Bearer header — this is HF's documented way to access gated repos
+    (see ``huggingface_hub.hf_hub_download``'s wire format). For every
+    other host we send no extra headers; allowlisted public files
+    don't need them and we don't want to leak tokens to other hosts.
+    """
+    if not is_hf_url(url):
+        return {}
+    tok = HF_AUTH_STORE.get_token_sync()
+    if tok is None or not tok.access_token:
+        return {}
+    return {"Authorization": f"Bearer {tok.access_token}"}
+
+
+def _remove_if_exists(path: str) -> None:
+    try:
+        os.remove(path)
+    except FileNotFoundError:
+        pass
+    except OSError as e:
+        logging.warning("[model_downloader] could not remove %s: %s", path, e)
+
+
+async def run_batch_sequential(sessions: list[DownloadSession]) -> None:
+    """Run a list of sessions one after the other.
+
+    Each session is independent: a failure or cancellation of one does
+    not abort the rest. Cancellations are observable via the registry
+    *before* a given download starts, so a session that's been
+    pre-cancelled (cancel before the worker reached it) just gets skipped.
+    """
+    for session in sessions:
+        # If the session got cancelled before its turn, skip without
+        # touching disk. This is what makes the per-request "sequential
+        # but cancellable" semantic work.
+        if not DOWNLOAD_SERVER.is_active(session):
+            DOWNLOAD_SERVER.finish(session)
+            continue
+        try:
+            await stream_to_disk(session)
+        except DownloadCancelled:
+            # Already logged + tmp removed inside stream_to_disk.
+            continue
+        except Exception:
+            # stream_to_disk already logged. Continue with the rest of the batch.
+            continue
+
+
+def schedule_batch(sessions: list[DownloadSession]) -> asyncio.Task:
+    """Kick off ``run_batch_sequential`` on the running event loop.
+
+    Returned task is fire-and-forget; the API handler returns immediately
+    after scheduling and clients observe progress via the polling endpoints.
+    """
+    return asyncio.create_task(run_batch_sequential(sessions))

app/model_downloader/gated_detection.py

@@ -0,0 +1,245 @@
+"""Per-URL probes for the unified availability endpoint.
+
+Three cached/derived facts per URL:
+
+  - ``is_gated``         intrinsic to the model; cached forever once known.
+                         Determined by ``auth_check(repo_id, token=None)``:
+                         ``GatedRepoError`` → True, success → False.
+
+  - ``is_hf_downloadable`` depends on the *current* token; recomputed every
+                         call. For non-gated URLs this is trivially True
+                         (no HF call needed). For gated URLs we run
+                         ``auth_check`` with the stored token each call.
+
+  - ``file_size``        intrinsic to the file. Cached forever once
+                         determined (including ``None`` on transient
+                         failure — we don't retry). We only attempt the
+                         HEAD when we already know the URL is downloadable
+                         to us; that way a failed-because-gated probe
+                         never lands as a cached ``None``.
+
+Caches are per-process, in-memory; small, no eviction needed for the
+workflow-scale (~tens of URLs). Concurrent calls for the same URL
+deduplicate via per-URL ``asyncio.Lock``.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+from dataclasses import dataclass
+from typing import Optional
+
+import aiohttp
+from huggingface_hub import HfApi
+from huggingface_hub.errors import (
+    GatedRepoError,
+    HfHubHTTPError,
+    RepositoryNotFoundError,
+)
+
+from app.model_downloader.allowlist import is_url_allowed
+from app.model_downloader.hf_auth.auth_store import HF_AUTH_STORE
+from app.model_downloader.hf_url import is_hf_url, repo_id_from_url
+from app.model_downloader.http_client import get_session, parse_content_length
+
+
+_HEAD_TIMEOUT = aiohttp.ClientTimeout(total=15)
+
+
+@dataclass
+class ProbeResult:
+    file_size: Optional[int]
+    is_hf_downloadable: Optional[bool]
+
+
+# --- caches -------------------------------------------------------------- #
+
+
+# url → bool. Whether this URL's HF repo gates access. Intrinsic to the
+# model — never changes for a given URL.
+_is_gated_cache: dict[str, bool] = {}
+
+# url → Optional[int]. The file's size in bytes, ``None`` if a probe
+# was attempted and produced no answer. **Only populated when we knew
+# the URL was downloadable to us at probe time** — so gated-without-
+# access never lands a ``None`` here that we'd be stuck with after login.
+_file_size_cache: dict[str, Optional[int]] = {}
+
+# Per-URL locks for single-flight probes — when multiple polls arrive
+# in the same tick for the same URL, exactly one of them runs the HF
+# call and the others wait on the result.
+_locks: dict[str, asyncio.Lock] = {}
+
+
+def _lock_for(url: str) -> asyncio.Lock:
+    lock = _locks.get(url)
+    if lock is None:
+        lock = asyncio.Lock()
+        _locks[url] = lock
+    return lock
+
+
+def clear_caches_for_tests() -> None:
+    """Test-only: drop everything."""
+    _is_gated_cache.clear()
+    _file_size_cache.clear()
+    _locks.clear()
+
+
+# --- public entrypoint --------------------------------------------------- #
+
+
+async def probe_url(url: str) -> ProbeResult:
+    """Return downloadability + size for one URL, using caches where safe."""
+    if not is_url_allowed(url):
+        return ProbeResult(file_size=None, is_hf_downloadable=None)
+    if not is_hf_url(url):
+        # Non-HF: ``is_hf_downloadable`` is "not applicable" (None).
+        # Size we still cache so we don't HEAD on every poll.
+        size = await _get_or_probe_size(url, token=None)
+        return ProbeResult(file_size=size, is_hf_downloadable=None)
+
+    repo_id = repo_id_from_url(url)
+    if repo_id is None:
+        return ProbeResult(file_size=None, is_hf_downloadable=None)
+
+    # Determine intrinsic gating once.
+    gated = await _resolve_is_gated(url, repo_id)
+    if gated is None:
+        return ProbeResult(file_size=None, is_hf_downloadable=None)
+
+    # Compute current-token downloadability per call.
+    tok = await HF_AUTH_STORE.get_valid_token()
+    token_str: Optional[str] = tok.access_token if tok else None
+    if not gated:
+        is_hf_downloadable: Optional[bool] = True
+    else:
+        is_hf_downloadable = await _auth_check_with_token(repo_id, token_str)
+
+    if is_hf_downloadable is True:
+        size = await _get_or_probe_size(url, token=token_str)
+    else:
+        # Skip the HEAD entirely — would 401 and we'd be stuck with
+        # cached None that survives a later login.
+        size = None
+
+    return ProbeResult(file_size=size, is_hf_downloadable=is_hf_downloadable)
+
+
+# --- gated/auth probes --------------------------------------------------- #
+
+
+async def _resolve_is_gated(url: str, repo_id: str) -> Optional[bool]:
+    """Decide once whether ``repo_id`` is a gated repo."""
+    cached = _is_gated_cache.get(url)
+    if cached is not None:
+        return cached
+
+    async with _lock_for(url):
+        cached = _is_gated_cache.get(url)
+        if cached is not None:
+            return cached
+        # Probe anonymously (token=None) on purpose: an unauthenticated
+        # auth_check is what makes HF raise GatedRepoError for gated repos.
+        # With a token, a gated-but-accepted repo would succeed and look
+        # ungated.
+        try:
+            await asyncio.to_thread(_auth_check_sync, repo_id, None)
+            _is_gated_cache[url] = False
+            return False
+        except GatedRepoError:
+            _is_gated_cache[url] = True
+            return True
+        except RepositoryNotFoundError:
+            # Repo doesn't exist publicly. Treat as gated — we can't
+            # serve it without auth, and an authenticated check might
+            # still succeed if it's a private repo the user can see.
+            _is_gated_cache[url] = True
+            return True
+        except (HfHubHTTPError, Exception) as e:
+            logging.debug(
+                "[hf_auth] is_gated probe failed for %s (will retry): %s",
+                repo_id, e,
+            )
+            return None  # don't cache; retry next call
+
+
+async def _auth_check_with_token(
+    repo_id: str, token: Optional[str]
+) -> Optional[bool]:
+    """Auth-check with the supplied token. True/False/None per outcome."""
+    try:
+        await asyncio.to_thread(_auth_check_sync, repo_id, token)
+        return True
+    except GatedRepoError:
+        return False
+    except RepositoryNotFoundError:
+        return False
+    except HfHubHTTPError as e:
+        # 401/403 covers org-SSO-required, revoked tokens, and similar —
+        # all of which mean "can't fetch right now" from the user's POV.
+        status = getattr(getattr(e, "response", None), "status_code", None)
+        if status in (401, 403):
+            return False
+        logging.debug(
+            "[hf_auth] auth_check transient failure for %s: %s", repo_id, e,
+        )
+        return None
+    except Exception as e:
+        logging.warning("[hf_auth] unexpected auth_check error for %s: %s", repo_id, e)
+        return None
+
+
+def _auth_check_sync(repo_id: str, token: Optional[str]) -> None:
+    """Thin sync wrapper around ``HfApi.auth_check`` for ``asyncio.to_thread``."""
+    HfApi().auth_check(repo_id, token=token)
+
+
+# --- size probe ---------------------------------------------------------- #
+
+
+async def _get_or_probe_size(url: str, token: Optional[str]) -> Optional[int]:
+    """Return the cached size or HEAD the URL once and cache the result."""
+    if url in _file_size_cache:
+        return _file_size_cache[url]
+
+    async with _lock_for(url):
+        if url in _file_size_cache:
+            return _file_size_cache[url]
+        size = await _probe_size_once(url, token=token)
+        _file_size_cache[url] = size
+        return size
+
+
+async def _probe_size_once(url: str, token: Optional[str]) -> Optional[int]:
+    """HEAD the URL and return the file size in bytes, or None on failure.
+
+    HuggingFace serves LFS-tracked files via a 302 to a signed CDN URL.
+    The real file size lives in the non-standard ``X-Linked-Size`` header
+    on that 302 response (``Content-Length`` is the redirect-body length).
+    Disabling redirect-follow lets us read either header on the same
+    response:
+
+      - LFS files: 302 + ``X-Linked-Size``
+      - Small/non-LFS files: 200 + ``Content-Length``
+    """
+    headers = {"Authorization": f"Bearer {token}"} if token else {}
+    try:
+        session = await get_session()
+        async with session.head(
+            url, allow_redirects=False, timeout=_HEAD_TIMEOUT, headers=headers,
+        ) as resp:
+            linked = parse_content_length(resp.headers.get("X-Linked-Size"))
+            if linked is not None:
+                return linked
+            if resp.status == 200:
+                return parse_content_length(resp.headers.get("Content-Length"))
+            return None
+    except (aiohttp.ClientError, TimeoutError, OSError):
+        return None
+
+
+# Backward-compat shim so consumers that still import the old name keep
+# building during the refactor; can be removed once routes are updated.
+MetadataProbeResult = ProbeResult

app/model_downloader/hf_auth/auth_store.py

@@ -0,0 +1,121 @@
+"""In-memory token cache with lazy disk persistence + refresh.
+
+Public surface is the ``HF_AUTH_STORE`` singleton. Callers ask
+``get_valid_token()``; the store transparently refreshes from disk
+on first use, refreshes via the OAuth refresh_token if the cached
+access_token is expired, and returns ``None`` if neither path works.
+
+The refresh path imports ``oauth.refresh_access_token`` lazily to
+avoid an import cycle (oauth needs the store to save tokens it
+acquires).
+"""
+
+from __future__ import annotations
+
+import logging
+import threading
+from typing import Optional
+
+from app.model_downloader.hf_auth.token_store import (
+    Token,
+    delete_token,
+    load_token,
+    save_token,
+)
+
+
+class HfAuthStore:
+    def __init__(self) -> None:
+        self._lock = threading.Lock()
+        self._token: Optional[Token] = None
+        self._loaded_from_disk = False
+
+    def _ensure_loaded(self) -> None:
+        """Read the disk token into memory on first access."""
+        if self._loaded_from_disk:
+            return
+        with self._lock:
+            if self._loaded_from_disk:
+                return
+            self._token = load_token()
+            self._loaded_from_disk = True
+
+    def has_token(self) -> bool:
+        """Cheap check: is there any token in memory?
+
+        Does not attempt refresh; an expired-but-refreshable token still
+        counts as "logged in" from the user's perspective.
+        """
+        self._ensure_loaded()
+        return self._token is not None
+
+    def _store_token_locked(self, token: Token) -> None:
+        """Set the in-memory token and persist it to disk.
+
+        Caller must already hold ``self._lock``. Keeping the disk write inside
+        the lock means memory and disk flip together — a concurrent ``clear()``
+        or refresh can't interleave between them.
+        """
+        self._token = token
+        self._loaded_from_disk = True
+        save_token(token)
+
+    def set_token(self, token: Token) -> None:
+        """Replace the in-memory token and persist to disk (atomically)."""
+        with self._lock:
+            self._store_token_locked(token)
+
+    def clear(self) -> None:
+        """Forget the token in memory and on disk (logout)."""
+        with self._lock:
+            self._token = None
+            self._loaded_from_disk = True
+            delete_token()
+
+    def get_token_sync(self) -> Optional[Token]:
+        """Return the cached token without refreshing.
+
+        Sync callers (e.g. constructing an Authorization header in a
+        non-async path) use this. They accept an expired token over
+        ``None``; HF will simply return 401 and the caller can decide
+        what to do.
+        """
+        self._ensure_loaded()
+        return self._token
+
+    async def get_valid_token(self) -> Optional[Token]:
+        """Return a fresh token, refreshing via OAuth if necessary.
+
+        Returns ``None`` if there's no cached token at all, or if the
+        cached token is expired and refresh failed. Callers should
+        treat that as "user is not logged in".
+        """
+        self._ensure_loaded()
+        with self._lock:
+            tok = self._token
+        if tok is None:
+            return None
+        if tok.is_valid():
+            return tok
+        if not tok.refresh_token:
+            return None
+
+        # Lazy import to avoid the oauth ↔ store import cycle.
+        from app.model_downloader.hf_auth.oauth import refresh_access_token
+
+        try:
+            refreshed = await refresh_access_token(tok.refresh_token)
+        except Exception as e:
+            logging.warning("[hf_auth] token refresh failed: %s", e)
+            return None
+
+        with self._lock:
+            # If a logout (clear) or another update replaced the token while we
+            # were awaiting the refresh, don't resurrect the old session.
+            if self._token is not tok:
+                return None
+            self._store_token_locked(refreshed)
+        return refreshed
+
+
+HF_AUTH_STORE = HfAuthStore()

app/model_downloader/hf_auth/eligibility.py

@@ -0,0 +1,55 @@
+"""Whether this deployment is allowed to do interactive HF OAuth.
+
+We only let the server hold a HuggingFace token under a strict trust
+assumption: this is a *single tenant local* install. Concretely:
+
+  - The server is bound to a loopback address. SSH tunneling /
+    reverse-proxies can defeat this, but it's the strongest signal
+    we have without an authentication system.
+  - ``--multi-user`` is off. A shared token used implicitly by multiple
+    declared users would be a footgun — one user's gated downloads
+    would silently authenticate as another.
+
+Anything else and the frontend hides the HF login UI entirely; gated
+models continue to show the "acquire it manually" message.
+"""
+
+from __future__ import annotations
+
+import ipaddress
+import socket
+
+from comfy.cli_args import args
+
+
+def _is_loopback(host: str | None) -> bool:
+    """Duplicates ``server.is_loopback`` (small, no shared module yet).
+
+    Resolves a host or IP literal to whether it lives on the loopback
+    interface (127.0.0.0/8 for IPv4, ::1 for IPv6). Returns False for
+    ``0.0.0.0`` / ``::`` because those are bind-all wildcards, not
+    loopback.
+    """
+    if host is None:
+        return False
+    try:
+        return ipaddress.ip_address(host).is_loopback
+    except ValueError:
+        pass
+
+    loopback = False
+    for family in (socket.AF_INET, socket.AF_INET6):
+        try:
+            r = socket.getaddrinfo(host, None, family, socket.SOCK_STREAM)
+            for _family, _, _, _, sockaddr in r:
+                if not ipaddress.ip_address(sockaddr[0]).is_loopback:
+                    return loopback
+                loopback = True
+        except socket.gaierror:
+            pass
+    return loopback
+
+
+def is_hf_auth_eligible() -> bool:
+    """True iff this deployment may surface the HF OAuth flow."""
+    return _is_loopback(args.listen) and not args.multi_user

app/model_downloader/hf_auth/oauth.py

@@ -0,0 +1,301 @@
+"""OAuth 2.0 PKCE flow against HuggingFace's authorization server.
+
+Wired so that ``POST /api/hf-auth-login-start`` can:
+  1. Generate state + PKCE verifier/challenge in this process.
+  2. Spin up a short-lived loopback HTTP server at port 41954 to
+     receive the redirect callback from HF.
+  3. Return the ``authorize_url`` for the frontend to open in a new tab.
+
+After the user grants consent on huggingface.co, HF redirects to the
+local callback URL with ``code`` and ``state``. The callback server
+validates ``state`` (CSRF), exchanges the code for tokens via PKCE,
+hands the resulting Token to ``HF_AUTH_STORE.set_token``, and shuts
+itself down.
+
+Before this can be exercised end-to-end a maintainer must register a
+HuggingFace OAuth app and substitute the ``HF_CLIENT_ID`` placeholder
+below. See the comment above the constant for the exact steps.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import base64
+import hashlib
+import logging
+import secrets
+import threading
+import time
+
+import aiohttp
+from aiohttp import web
+
+from app.model_downloader.hf_auth.auth_store import HF_AUTH_STORE
+from app.model_downloader.hf_auth.token_store import Token
+from app.model_downloader.http_client import get_session
+
+
+# --- HF OAuth app registration -------------------------------------------- #
+# NOTE: The OAuth client_id below is a placeholder. Before this feature can be
+# exercised end-to-end, a maintainer must register a HuggingFace OAuth app
+# under a Comfy-Org-controlled HF account and substitute its client_id here.
+# Detailed walkthrough is in docs/server-side-model-downloads-handover.html
+# ("HuggingFace OAuth app setup" section). Short version:
+#   1. huggingface.co → Settings → Connected Apps → "Create app"
+#   2. Default Scopes: check ``openid`` + ``profile`` (User Info) and
+#      ``gated-repos`` (Repository Access). Leave everything else off.
+#   3. Redirect URLs: exactly ``http://127.0.0.1:41954/api/auth/huggingface/callback``
+#      — must match ``REDIRECT_URI`` below; change both in lockstep if you
+#      change ``CALLBACK_PORT``.
+#   4. Save → copy the resulting Client ID into ``HF_CLIENT_ID`` below.
+# The client_id is not a secret (it travels through the user's browser in
+# plaintext); HF's "Public app" type means there's no client secret to
+# manage — PKCE replaces it.
+HF_CLIENT_ID = "REPLACE_ME_WITH_COMFY_ORG_HF_OAUTH_CLIENT_ID"
+
+CALLBACK_HOST = "127.0.0.1"
+CALLBACK_PORT = 41954
+CALLBACK_PATH = "/api/auth/huggingface/callback"
+REDIRECT_URI = f"http://{CALLBACK_HOST}:{CALLBACK_PORT}{CALLBACK_PATH}"
+
+AUTHORIZE_URL = "https://huggingface.co/oauth/authorize"
+TOKEN_URL = "https://huggingface.co/oauth/token"
+_TOKEN_REQUEST_TIMEOUT = aiohttp.ClientTimeout(total=30)
+# Minimal scope set for the feature:
+#   - openid       : required by HF when the app uses OIDC at all
+#   - profile      : lets ``HfApi.whoami(token=...)`` return a username for the
+#                    settings UI; cosmetic but expected
+#   - gated-repos  : grants the token enough to call ``auth_check`` and
+#                    download files from public gated repos the user has
+#                    accepted the license for. The wider ``read-repos`` scope
+#                    would also work (it includes ``gated-repos``) but it
+#                    additionally grants private-repo read access, which we
+#                    don't need and which makes the consent screen scarier
+#                    for the user.
+SCOPE = "openid profile gated-repos"
+
+# Maximum time the callback server stays up waiting for the user to
+# complete consent on huggingface.co. Past this, the port closes and
+# the user has to click "Log in" again.
+CALLBACK_TIMEOUT_SECS = 300
+
+
+# Process-wide lock so two simultaneous /api/hf-auth-login-start
+# requests don't fight over port CALLBACK_PORT.
+_OAUTH_LOCK = threading.Lock()
+
+
+class OAuthInProgressError(Exception):
+    """Another OAuth attempt is already running."""
+
+
+class OAuthCallbackError(Exception):
+    """The OAuth callback returned an error (HF denied, port stolen, etc.)."""
+
+
+# --- PKCE primitives ------------------------------------------------------ #
+
+
+def _make_pkce() -> tuple[str, str, str]:
+    """Return ``(verifier, challenge, state)``.
+
+    Verifier never leaves this process. Challenge and state travel
+    through the user's browser. State is checked on the callback to
+    prevent a malicious cross-origin redirect from injecting a token.
+    """
+    verifier = secrets.token_urlsafe(64)
+    challenge = (
+        base64.urlsafe_b64encode(hashlib.sha256(verifier.encode("ascii")).digest())
+        .rstrip(b"=")
+        .decode("ascii")
+    )
+    state = secrets.token_urlsafe(32)
+    return verifier, challenge, state
+
+
+def _build_authorize_url(challenge: str, state: str) -> str:
+    from urllib.parse import urlencode
+
+    params = {
+        "client_id": HF_CLIENT_ID,
+        "redirect_uri": REDIRECT_URI,
+        "response_type": "code",
+        "scope": SCOPE,
+        "state": state,
+        "code_challenge": challenge,
+        "code_challenge_method": "S256",
+    }
+    return f"{AUTHORIZE_URL}?{urlencode(params)}"
+
+
+# --- Token exchange ------------------------------------------------------- #
+
+
+async def _exchange_code(code: str, verifier: str) -> Token:
+    """Trade the authorization code for an access+refresh token pair."""
+    data = {
+        "grant_type": "authorization_code",
+        "code": code,
+        "redirect_uri": REDIRECT_URI,
+        "client_id": HF_CLIENT_ID,
+        "code_verifier": verifier,
+    }
+    session = await get_session()
+    async with session.post(TOKEN_URL, data=data, timeout=_TOKEN_REQUEST_TIMEOUT) as resp:
+        resp.raise_for_status()
+        body = await resp.json()
+    return Token(
+        access_token=body["access_token"],
+        refresh_token=body.get("refresh_token"),
+        expires_at=time.time() + float(body.get("expires_in", 3600)),
+        scope=body.get("scope", SCOPE),
+    )
+
+
+async def refresh_access_token(refresh_token: str) -> Token:
+    """Trade a refresh_token for a new access (+ possibly refresh) token."""
+    data = {
+        "grant_type": "refresh_token",
+        "refresh_token": refresh_token,
+        "client_id": HF_CLIENT_ID,
+    }
+    session = await get_session()
+    async with session.post(TOKEN_URL, data=data, timeout=_TOKEN_REQUEST_TIMEOUT) as resp:
+        resp.raise_for_status()
+        body = await resp.json()
+    return Token(
+        access_token=body["access_token"],
+        # If HF doesn't rotate refresh tokens, keep using the existing one.
+        refresh_token=body.get("refresh_token", refresh_token),
+        expires_at=time.time() + float(body.get("expires_in", 3600)),
+        scope=body.get("scope", SCOPE),
+    )
+
+
+# --- Callback server ------------------------------------------------------ #
+
+
+async def start_login_flow() -> str:
+    """Begin one OAuth attempt: spawn the callback server, return the URL.
+
+    Returns the URL the frontend should open in a new tab. Raises
+    ``OAuthInProgressError`` if another attempt is already running.
+    The callback server runs in the background until the user
+    completes consent or until ``CALLBACK_TIMEOUT_SECS`` elapses;
+    either way the lock + port are released afterward.
+    """
+    if not _OAUTH_LOCK.acquire(blocking=False):
+        raise OAuthInProgressError()
+
+    try:
+        verifier, challenge, state = _make_pkce()
+        authorize_url = _build_authorize_url(challenge, state)
+        ready: asyncio.Future[None] = asyncio.get_event_loop().create_future()
+    except BaseException:
+        # Failed before handing the lock to the callback-server task: release it
+        # here. (Once the task is spawned, it owns releasing the lock.)
+        _OAUTH_LOCK.release()
+        raise
+
+    asyncio.create_task(_run_callback_server(verifier, state, ready))
+    # Don't return the URL until the callback server is actually bound and
+    # listening — otherwise HF could redirect to a port nothing is serving and
+    # the login would silently dead-end. ``ready`` raises if the bind failed.
+    await ready
+    return authorize_url
+
+
+async def _run_callback_server(
+    verifier: str, expected_state: str, ready: "asyncio.Future[None]"
+) -> None:
+    """Listen for HF's redirect once, capture the token, then shut down.
+
+    Signals ``ready`` once the port is bound (or with an exception if the bind
+    fails), so ``start_login_flow`` only hands back a URL on a live server.
+    """
+    received: asyncio.Future[Token] = asyncio.get_event_loop().create_future()
+
+    async def handler(request: web.Request) -> web.Response:
+        try:
+            if request.query.get("state") != expected_state:
+                return web.Response(status=400, text="state mismatch")
+            err = request.query.get("error")
+            if err:
+                received.set_exception(OAuthCallbackError(f"HF returned: {err}"))
+                return web.Response(status=400, text=f"OAuth error: {err}")
+            code = request.query.get("code")
+            if not code:
+                return web.Response(status=400, text="missing code")
+            tok = await _exchange_code(code, verifier)
+            if not received.done():
+                received.set_result(tok)
+            return web.Response(
+                content_type="text/html",
+                text=(
+                    "<html><body style='font-family:sans-serif;padding:40px'>"
+                    "<h2>HuggingFace login successful</h2>"
+                    "<p>You can close this tab and return to ComfyUI.</p>"
+                    "</body></html>"
+                ),
+            )
+        except Exception as exc:
+            if not received.done():
+                received.set_exception(exc)
+            return web.Response(status=500, text=str(exc))
+
+    app = web.Application()
+    app.router.add_get(CALLBACK_PATH, handler)
+    runner = web.AppRunner(app)
+    try:
+        await runner.setup()
+        site = web.TCPSite(runner, CALLBACK_HOST, CALLBACK_PORT, reuse_address=True)
+        await site.start()
+    except Exception as e:
+        # Couldn't bind the callback port (commonly already in use). Tell the
+        # waiting start_login_flow via ``ready`` so it surfaces a clear error
+        # instead of returning a dead URL, and release the lock for next time.
+        logging.warning("[hf_auth] could not start callback server: %s", e)
+        if not ready.done():
+            ready.set_exception(
+                OAuthCallbackError(f"could not bind callback port {CALLBACK_PORT}: {e}")
+            )
+        _OAUTH_LOCK.release()
+        return
+
+    # Bound and listening — now it's safe for start_login_flow to return the URL.
+    if not ready.done():
+        ready.set_result(None)
+
+    try:
+        token = await asyncio.wait_for(received, timeout=CALLBACK_TIMEOUT_SECS)
+    except asyncio.TimeoutError:
+        logging.info("[hf_auth] OAuth login timed out after %ds", CALLBACK_TIMEOUT_SECS)
+        return
+    except OAuthCallbackError as e:
+        logging.warning("[hf_auth] OAuth callback error: %s", e)
+        return
+    except Exception as e:
+        logging.warning("[hf_auth] unexpected OAuth failure: %s", e)
+        return
+    else:
+        HF_AUTH_STORE.set_token(token)
+        logging.info("[hf_auth] OAuth login complete")
+    finally:
+        await runner.cleanup()
+        if _OAUTH_LOCK.locked():
+            _OAUTH_LOCK.release()
+
+
+def is_login_in_progress() -> bool:
+    """True iff a callback server is currently bound + waiting."""
+    return _OAUTH_LOCK.locked()
+
+
+# Re-export for callers that only want the URL builder (e.g. tests).
+__all__ = [
+    "start_login_flow",
+    "refresh_access_token",
+    "is_login_in_progress",
+    "OAuthInProgressError",
+    "CALLBACK_TIMEOUT_SECS",
+]

app/model_downloader/hf_auth/token_store.py

@@ -0,0 +1,94 @@
+"""On-disk persistence for the HuggingFace OAuth token.
+
+The token shape mirrors what HF returns on the token exchange: an
+``access_token``, an optional ``refresh_token``, the absolute epoch at
+which the access token expires, and the granted scope. We persist
+this so logging in once survives ComfyUI restarts under the internal
+``__hf_auth`` system-user directory; the file is mode ``0600`` so only
+the OS user can read it.
+"""
+
+from __future__ import annotations
+
+import json
+import logging
+import os
+import stat
+import time
+from dataclasses import asdict, dataclass
+from typing import Optional
+
+import folder_paths
+
+
+# Treat a token as expired this many seconds before its server-reported
+# ``expires_at`` so we don't try to use a token mid-request only for it
+# to flip stale between auth_check and the actual GET.
+EXPIRY_BUFFER_SECS = 60
+
+TOKEN_FILENAME = "hf_auth_token.json"
+
+
+@dataclass
+class Token:
+    """One OAuth token + the metadata we need to use it."""
+    access_token: str
+    refresh_token: Optional[str]
+    expires_at: float  # absolute epoch seconds
+    scope: str = ""
+
+    def is_valid(self) -> bool:
+        """True iff we can use this token right now."""
+        return (
+            bool(self.access_token)
+            and (self.expires_at - time.time() > EXPIRY_BUFFER_SECS)
+        )
+
+
+def _token_dir() -> str:
+    return folder_paths.get_system_user_directory("hf_auth")
+
+
+def _token_path() -> str:
+    return os.path.join(_token_dir(), TOKEN_FILENAME)
+
+
+def load_token() -> Optional[Token]:
+    """Read the persisted token, returning ``None`` if absent or corrupt."""
+    path = _token_path()
+    if not os.path.exists(path):
+        return None
+    try:
+        with open(path, "r", encoding="utf-8") as f:
+            data = json.load(f)
+        return Token(**data)
+    except (OSError, json.JSONDecodeError, TypeError) as e:
+        logging.warning("[hf_auth] could not load token at %s: %s", path, e)
+        return None
+
+
+def save_token(token: Token) -> None:
+    """Atomically write the token with 0600 permissions."""
+    path = _token_path()
+    os.makedirs(os.path.dirname(path), exist_ok=True)
+    tmp = path + ".tmp"
+    fd = os.open(tmp, os.O_WRONLY | os.O_CREAT | os.O_TRUNC, 0o600)
+    with os.fdopen(fd, "w", encoding="utf-8") as f:
+        json.dump(asdict(token), f)
+    os.replace(tmp, path)
+    try:
+        os.chmod(path, stat.S_IRUSR | stat.S_IWUSR)
+    except OSError as e:
+        # On Windows / weird filesystems chmod may be a no-op; not fatal.
+        logging.debug("[hf_auth] chmod 0600 on %s failed: %s", path, e)
+
+
+def delete_token() -> None:
+    """Remove the persisted token; no-op if it doesn't exist."""
+    path = _token_path()
+    try:
+        os.remove(path)
+    except FileNotFoundError:
+        pass
+    except OSError as e:
+        logging.warning("[hf_auth] could not remove token at %s: %s", path, e)

app/model_downloader/hf_url.py

@@ -0,0 +1,41 @@
+"""Parsers for the ``huggingface.co`` URL shape we accept in workflows.
+
+The download API accepts URLs of the form
+``https://huggingface.co/<org>/<repo>/resolve/<rev>/<path/to/file>``.
+We need to recover ``<org>/<repo>`` (the *repo_id*) from such URLs for
+``huggingface_hub`` API calls (notably ``HfApi.auth_check``).
+"""
+
+from __future__ import annotations
+
+from typing import Optional
+from urllib.parse import urlparse
+
+_HF_HOST = "huggingface.co"
+
+
+def is_hf_url(url: str) -> bool:
+    """Cheap host check — does this URL point at huggingface.co?"""
+    try:
+        return urlparse(url).hostname == _HF_HOST
+    except ValueError:
+        return False
+
+
+def repo_id_from_url(url: str) -> Optional[str]:
+    """Extract ``<org>/<repo>`` from an HF model file URL.
+
+    Returns ``None`` if the URL isn't on huggingface.co or doesn't look
+    like a model-file URL. The expected shape is
+    ``/<org>/<repo>/resolve/<rev>/<path>`` — anything else
+    (datasets, spaces, /tree/, /blob/, …) we treat as out of scope here.
+    """
+    if not is_hf_url(url):
+        return None
+    parts = urlparse(url).path.lstrip("/").split("/")
+    if len(parts) < 4 or parts[2] != "resolve":
+        return None
+    org, repo = parts[0], parts[1]
+    if not org or not repo:
+        return None
+    return f"{org}/{repo}"

app/model_downloader/http_client.py

@@ -0,0 +1,63 @@
+"""Lazy module-level aiohttp ClientSession.
+
+A single shared session means TLS handshakes are reused across HEAD probes
+and the subsequent GETs to the same host (HuggingFace is the dominant
+case), which is a noticeable speedup on cold connections.
+
+We deliberately don't close the session at process exit — aiohttp's
+warning about unclosed sessions is benign at shutdown, and adding atexit
+plumbing buys nothing because the OS reclaims the sockets anyway. The
+session lifetime is the lifetime of the Python process.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import ssl
+from typing import Optional
+
+import aiohttp
+import certifi
+
+
+# Larger per-host pool than aiohttp's default (=100 total / =0 per host)
+# so concurrent gated probes + a download to the same host don't queue.
+_CONNECTOR_LIMIT_PER_HOST = 8
+
+_session: Optional[aiohttp.ClientSession] = None
+_lock = asyncio.Lock()
+
+
+def ssl_context() -> ssl.SSLContext:
+    """TLS context pinned to certifi's CA bundle.
+    aiohttp's default context uses the OS trust store, which isn't wired up
+    on some Python installs (python.org macOS, slim containers) — there TLS
+    to huggingface.co fails with CERTIFICATE_VERIFY_FAILED.
+    """
+    return ssl.create_default_context(cafile=certifi.where())
+
+
+async def get_session() -> aiohttp.ClientSession:
+    """Return the shared session, creating it on first call."""
+    global _session
+    if _session is not None and not _session.closed:
+        return _session
+    async with _lock:
+        if _session is None or _session.closed:
+            connector = aiohttp.TCPConnector(
+                limit_per_host=_CONNECTOR_LIMIT_PER_HOST,
+                ssl=ssl_context(),
+            )
+            _session = aiohttp.ClientSession(connector=connector)
+    return _session
+
+
+def parse_content_length(value: Optional[str]) -> Optional[int]:
+    """Parse a byte-count header value, or None if absent/malformed/negative."""
+    if not value:
+        return None
+    try:
+        n = int(value)
+    except ValueError:
+        return None
+    return n if n >= 0 else None

app/model_downloader/paths.py

@@ -0,0 +1,111 @@
+"""Path resolution for model downloads.
+
+Model identifiers used across the download API are *relative destination
+paths* of the form ``<directory>/<filename>`` (e.g. ``loras/my_lora.safetensors``).
+This module turns one of those identifiers into an absolute on-disk path
+under one of ComfyUI's registered model folders, while rejecting unknown
+folders, path traversal, and other ill-formed inputs.
+"""
+
+import os
+import re
+from typing import Optional, Tuple
+
+import folder_paths
+
+
+# Constrain components so a model_id can never escape its target directory.
+# - directory: a single path segment of safe chars
+# - filename: a single path segment of safe chars, must end with a model extension
+_SEGMENT_RE = re.compile(r"^[A-Za-z0-9._-]+$")
+
+# Destination filename must name a model file (same set as the URL allowlist),
+# so a download can't land as e.g. ``foo.txt`` that ComfyUI won't recognise.
+_MODEL_EXTENSIONS = (".safetensors", ".sft", ".ckpt", ".pth", ".pt")
+
+# Distinctive temp suffix so the startup orphan-sweep only removes files THIS
+# subsystem created — never unrelated ``*.tmp`` files in the model dirs.
+_TMP_SUFFIX = ".comfy-download.tmp"
+
+
+class InvalidModelId(ValueError):
+    """Raised when a model_id is syntactically invalid or refers to an
+    unknown model folder."""
+
+
+def parse_model_id(model_id: str) -> Tuple[str, str]:
+    """Split ``<directory>/<filename>`` and validate both components.
+
+    Returns ``(directory, filename)``. Raises ``InvalidModelId`` on
+    malformed input. Does NOT touch the filesystem.
+    """
+    if not isinstance(model_id, str) or "/" not in model_id:
+        raise InvalidModelId(f"model_id must be '<directory>/<filename>', got {model_id!r}")
+    directory, _, filename = model_id.partition("/")
+    if "/" in filename or not directory or not filename:
+        raise InvalidModelId(f"model_id must be exactly one '/' separator, got {model_id!r}")
+    if not _SEGMENT_RE.match(directory):
+        raise InvalidModelId(f"invalid directory segment {directory!r}")
+    if not _SEGMENT_RE.match(filename):
+        raise InvalidModelId(f"invalid filename segment {filename!r}")
+    if not filename.endswith(_MODEL_EXTENSIONS):
+        raise InvalidModelId(
+            f"filename must end with a model extension {_MODEL_EXTENSIONS}, got {filename!r}"
+        )
+    if directory not in folder_paths.folder_names_and_paths:
+        raise InvalidModelId(f"unknown model folder {directory!r}")
+    return directory, filename
+
+
+def resolve_existing(model_id: str) -> Optional[str]:
+    """Return the absolute path of an installed model, or None if missing.
+
+    Honours ``extra_model_paths.yaml`` transparently via
+    ``folder_paths.get_full_path``.
+    """
+    directory, filename = parse_model_id(model_id)
+    return folder_paths.get_full_path(directory, filename)
+
+
+def resolve_destination(model_id: str, epoch: int = 0) -> Tuple[str, str]:
+    """Return ``(final_path, tmp_path)`` for a download.
+
+    Downloads land at the first registered path for the model's directory
+    (the "primary" location). The temp sibling is the write target, atomically
+    renamed onto ``final_path`` on success.
+
+    ``tmp_path`` embeds the session ``epoch`` so a cancel+retry of the same
+    model never shares a temp path between the old (cancelling) worker and the
+    new attempt — otherwise the old worker's rollback could delete the new
+    worker's in-progress file. The distinctive suffix scopes the orphan sweep.
+    """
+    directory, filename = parse_model_id(model_id)
+    roots = folder_paths.get_folder_paths(directory)
+    if not roots:
+        raise InvalidModelId(f"no on-disk path registered for folder {directory!r}")
+    root = roots[0]
+    final_path = os.path.join(root, filename)
+    tmp_path = f"{final_path}.{epoch}{_TMP_SUFFIX}"
+    return final_path, tmp_path
+
+
+def iter_all_tmp_paths():
+    """Yield this subsystem's temp files under every registered model folder.
+
+    Matches only our distinctive ``_TMP_SUFFIX`` (not every ``*.tmp``) so the
+    startup orphan-sweep can't delete temp files created by other tools.
+    """
+    seen_roots: set[str] = set()
+    for directory in folder_paths.folder_names_and_paths.keys():
+        for root in folder_paths.get_folder_paths(directory):
+            if root in seen_roots or not os.path.isdir(root):
+                continue
+            seen_roots.add(root)
+            try:
+                for entry in os.scandir(root):
+                    if entry.is_file() and entry.name.endswith(_TMP_SUFFIX):
+                        yield entry.path
+            except OSError:
+                # Folder might be unreadable / missing on certain mounts —
+                # not fatal, just skip it.
+                continue

app/model_manager.py

@@ -1,5 +1,3 @@
-from __future__ import annotations
-
 import os
 import base64
 import json
@@ -52,21 +50,45 @@ class ModelFileManager:
         @routes.get("/experiment/models/preview/{folder}/{path_index}/{filename:.*}")
         async def get_model_preview(request):
             folder_name = request.match_info.get("folder", None)
-            path_index = int(request.match_info.get("path_index", None))
             filename = request.match_info.get("filename", None)
 
             if folder_name not in folder_paths.folder_names_and_paths:
                 return web.Response(status=404)
 
+            # The "{filename:.*}" capture also matches the empty string, which
+            # would resolve to the folder itself; reject it explicitly.
+            if not filename:
+                return web.Response(status=400)
+
+            try:
+                path_index = int(request.match_info.get("path_index", None))
+            except (TypeError, ValueError):
+                return web.Response(status=400)
+
             folders = folder_paths.folder_names_and_paths[folder_name]
+            if path_index < 0 or path_index >= len(folders[0]):
+                return web.Response(status=404)
             folder = folders[0][path_index]
-            full_filename = os.path.join(folder, filename)
+            full_filename = os.path.normpath(os.path.join(folder, filename))
+
+            # Prevent path traversal: the requested file must stay within the
+            # configured model folder. `filename` is an unrestricted ".*" capture,
+            # so values like "../../../../etc/passwd" would otherwise escape it.
+            if not folder_paths.is_within_directory(folder, full_filename):
+                return web.Response(status=403)
 
             previews = self.get_model_previews(full_filename)
             default_preview = previews[0] if len(previews) > 0 else None
             if default_preview is None or (isinstance(default_preview, str) and not os.path.isfile(default_preview)):
                 return web.Response(status=404)
 
+            # The preview is selected by a glob inside get_model_previews, so a
+            # companion file (e.g. "model.preview.png") could itself be a symlink
+            # resolving outside the model folder. Re-validate the file actually
+            # opened: is_within_directory realpaths it, catching symlink escape.
+            if isinstance(default_preview, str) and not folder_paths.is_within_directory(folder, default_preview):
+                return web.Response(status=403)
+
             try:
                 with Image.open(default_preview) as img:
                     img_bytes = BytesIO()

app/user_manager.py

@@ -1,4 +1,3 @@
-from __future__ import annotations
 import json
 import os
 import re
@@ -7,6 +6,7 @@ import glob
 import shutil
 import logging
 import tempfile
+import mimetypes
 from aiohttp import web
 from urllib import parse
 from comfy.cli_args import args
@@ -337,7 +337,20 @@ class UserManager():
             if not isinstance(path, str):
                 return path
 
-            return web.FileResponse(path)
+            # User data files are arbitrary user-supplied content and are never
+            # meant to render inline. Disable MIME sniffing and force a download
+            # so uploaded markup/scripts can't execute in the app origin (stored
+            # XSS). Content-Disposition: attachment is the load-bearing guard;
+            # the content-type override and nosniff are defence in depth.
+            content_type = mimetypes.guess_type(path)[0] or 'application/octet-stream'
+            if folder_paths.is_dangerous_content_type(content_type):
+                content_type = 'application/octet-stream'
+
+            return web.FileResponse(path, headers={
+                "Content-Type": content_type,
+                "X-Content-Type-Options": "nosniff",
+                "Content-Disposition": "attachment",
+            })
 
         @routes.post("/userdata/{file}")
         async def post_userdata(request):

blueprints/Canny to Image (Z-Image-Turbo).json

@@ -1553,7 +1553,7 @@
           "VHS_MetadataImage": true,
           "VHS_KeepIntermediate": true
         },
-        "category": "Image generation and editing/Canny to image",
+        "category": "Image generation and editing/Conditioned",
         "description": "Generates an image from a Canny edge map using Z-Image-Turbo, with text conditioning."
       }
     ]

blueprints/Canny to Video (LTX 2.0).json

@@ -3600,7 +3600,7 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Video generation and editing/Canny to video",
+        "category": "Video generation and editing/Conditioned",
         "description": "Generates video from Canny edge maps using LTX-2, with optional synchronized audio."
       }
     ]

blueprints/ControlNet (Z-Image-Turbo).json

@@ -1401,7 +1401,7 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Image generation and editing/ControlNet",
+        "category": "Image generation and editing/Conditioned",
         "description": "Generates images from a text prompt and ControlNet conditioning (e.g. depth, canny) using Z-Image-Turbo."
       }
     ]

blueprints/Depth to Image (Z-Image-Turbo).json

@@ -1579,7 +1579,7 @@
           "VHS_MetadataImage": true,
           "VHS_KeepIntermediate": true
         },
-        "category": "Image generation and editing/Depth to image",
+        "category": "Image generation and editing/Conditioned",
         "description": "Generates an image from a depth map using Z-Image-Turbo with text conditioning."
       },
       {

blueprints/Depth to Video (ltx 2.0).json

@@ -4233,7 +4233,7 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Video generation and editing/Depth to video",
+        "category": "Video generation and editing/Conditioned",
         "description": "Generates depth-controlled video with LTX-2: motion and structure follow a depth-reference video alongside text prompting, optional first-frame image conditioning, with optional synchronized audio."
       },
       {

blueprints/First-Last-Frame to Video (LTX-2.3).json

@@ -3350,7 +3350,7 @@
           }
         ],
         "extra": {},
-        "category": "Video generation and editing/First-Last-Frame to Video",
+        "category": "Video generation and editing/Conditioned",
         "description": "Generates a video interpolating between first and last keyframes using LTX-2.3."
       }
     ]

blueprints/First-Last-Frame to Video.json

@@ -3350,7 +3350,7 @@
           }
         ],
         "extra": {},
-        "category": "Video generation and editing/First-Last-Frame to Video",
+        "category": "Video generation and editing/FLF2V",
         "description": "Generates a video that interpolates between the first and last keyframes using LTX-2.3, including optional audio."
       }
     ]

blueprints/Image Captioning (gemini).json

@@ -310,9 +310,9 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Text generation/Image Captioning",
+        "category": "Image Tools",
         "description": "Generates descriptive captions for images using Google's Gemini multimodal LLM."
       }
     ]
   }
-}
+}

blueprints/Image Depth Estimation (Depth Anything 3).json

@@ -0,0 +1,569 @@
+{
+  "revision": 0,
+  "last_node_id": 89,
+  "last_link_id": 0,
+  "nodes": [
+    {
+      "id": 89,
+      "type": "85e595bd-af9e-40ee-85c5-b98bb15da47a",
+      "pos": [
+        320,
+        520
+      ],
+      "size": [
+        400,
+        360
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "image",
+          "name": "image",
+          "type": "IMAGE",
+          "link": null
+        },
+        {
+          "name": "resolution",
+          "type": "INT",
+          "widget": {
+            "name": "resolution"
+          },
+          "link": null
+        },
+        {
+          "name": "resize_method",
+          "type": "COMBO",
+          "widget": {
+            "name": "resize_method"
+          },
+          "link": null
+        },
+        {
+          "label": "output_type",
+          "name": "output",
+          "type": "COMFY_DYNAMICCOMBO_V3",
+          "widget": {
+            "name": "output"
+          },
+          "link": null
+        },
+        {
+          "label": "output_normalization",
+          "name": "output.normalization",
+          "type": "COMBO",
+          "widget": {
+            "name": "output.normalization"
+          },
+          "link": null
+        },
+        {
+          "label": "apply_sky_clip",
+          "name": "output.apply_sky_clip",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "output.apply_sky_clip"
+          },
+          "link": null
+        },
+        {
+          "name": "model_name",
+          "type": "COMBO",
+          "widget": {
+            "name": "model_name"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "IMAGE",
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": []
+        }
+      ],
+      "properties": {
+        "proxyWidgets": [
+          [
+            "87",
+            "resolution"
+          ],
+          [
+            "87",
+            "resize_method"
+          ],
+          [
+            "86",
+            "output"
+          ],
+          [
+            "86",
+            "output.normalization"
+          ],
+          [
+            "86",
+            "output.apply_sky_clip"
+          ],
+          [
+            "88",
+            "model_name"
+          ]
+        ],
+        "cnr_id": "comfy-core",
+        "ver": "0.24.0"
+      },
+      "widgets_values": [],
+      "title": "Image Depth Estimation (Depth Anything 3)"
+    }
+  ],
+  "links": [],
+  "version": 0.4,
+  "definitions": {
+    "subgraphs": [
+      {
+        "id": "85e595bd-af9e-40ee-85c5-b98bb15da47a",
+        "version": 1,
+        "state": {
+          "lastGroupId": 4,
+          "lastNodeId": 89,
+          "lastLinkId": 109,
+          "lastRerouteId": 0
+        },
+        "revision": 2,
+        "config": {},
+        "name": "Image Depth Estimation (Depth Anything 3)",
+        "inputNode": {
+          "id": -10,
+          "bounding": [
+            400,
+            90,
+            166.998046875,
+            188
+          ]
+        },
+        "outputNode": {
+          "id": -20,
+          "bounding": [
+            1250,
+            146,
+            128,
+            68
+          ]
+        },
+        "inputs": [
+          {
+            "id": "43cf3118-495a-487d-8eb3-a17c7e92f64f",
+            "name": "image",
+            "type": "IMAGE",
+            "linkIds": [
+              19
+            ],
+            "localized_name": "image",
+            "pos": [
+              542.998046875,
+              114
+            ]
+          },
+          {
+            "id": "1089a0a1-6db1-45a8-84b0-0bfdc2ed920a",
+            "name": "resolution",
+            "type": "INT",
+            "linkIds": [
+              22
+            ],
+            "pos": [
+              542.998046875,
+              134
+            ]
+          },
+          {
+            "id": "25fb64ac-26d5-466d-995b-6d51b9afa2c4",
+            "name": "resize_method",
+            "type": "COMBO",
+            "linkIds": [
+              23
+            ],
+            "pos": [
+              542.998046875,
+              154
+            ]
+          },
+          {
+            "id": "8acafb7c-6c8b-46b3-9d74-c563498a3af1",
+            "name": "output",
+            "type": "COMFY_DYNAMICCOMBO_V3",
+            "linkIds": [
+              24
+            ],
+            "label": "output_type",
+            "pos": [
+              542.998046875,
+              174
+            ]
+          },
+          {
+            "id": "1da5009b-4648-43e8-a257-16426630cf22",
+            "name": "output.normalization",
+            "type": "COMBO",
+            "linkIds": [
+              25
+            ],
+            "label": "output_normalization",
+            "pos": [
+              542.998046875,
+              194
+            ]
+          },
+          {
+            "id": "fd7edb33-5fb1-4538-a411-26e5039a9321",
+            "name": "output.apply_sky_clip",
+            "type": "BOOLEAN",
+            "linkIds": [
+              26
+            ],
+            "label": "apply_sky_clip",
+            "pos": [
+              542.998046875,
+              214
+            ]
+          },
+          {
+            "id": "b5be4c8a-b833-4f1e-8c94-3ed1dd722190",
+            "name": "model_name",
+            "type": "COMBO",
+            "linkIds": [
+              106
+            ],
+            "pos": [
+              542.998046875,
+              234
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "id": "478ab537-63bc-4d74-a9f0-c975f550880f",
+            "name": "IMAGE",
+            "type": "IMAGE",
+            "linkIds": [
+              7
+            ],
+            "localized_name": "IMAGE",
+            "pos": [
+              1274,
+              170
+            ]
+          }
+        ],
+        "widgets": [],
+        "nodes": [
+          {
+            "id": 86,
+            "type": "DA3Render",
+            "pos": [
+              800,
+              310
+            ],
+            "size": [
+              380,
+              130
+            ],
+            "flags": {},
+            "order": 0,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "da3_geometry",
+                "name": "da3_geometry",
+                "type": "DA3_GEOMETRY",
+                "link": 12
+              },
+              {
+                "localized_name": "output",
+                "name": "output",
+                "type": "COMFY_DYNAMICCOMBO_V3",
+                "widget": {
+                  "name": "output"
+                },
+                "link": 24
+              },
+              {
+                "localized_name": "output.normalization",
+                "name": "output.normalization",
+                "type": "COMBO",
+                "widget": {
+                  "name": "output.normalization"
+                },
+                "link": 25
+              },
+              {
+                "localized_name": "output.apply_sky_clip",
+                "name": "output.apply_sky_clip",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "output.apply_sky_clip"
+                },
+                "link": 26
+              },
+              {
+                "name": "geometry",
+                "type": "DA3_GEOMETRY",
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "IMAGE",
+                "name": "IMAGE",
+                "type": "IMAGE",
+                "slot_index": 0,
+                "links": [
+                  7
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "DA3Render",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0"
+            },
+            "widgets_values": [
+              "depth",
+              "v2_style",
+              false
+            ]
+          },
+          {
+            "id": 87,
+            "type": "DA3Inference",
+            "pos": [
+              800,
+              50
+            ],
+            "size": [
+              390,
+              130
+            ],
+            "flags": {},
+            "order": 1,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "da3_model",
+                "name": "da3_model",
+                "type": "DA3_MODEL",
+                "link": 107
+              },
+              {
+                "localized_name": "image",
+                "name": "image",
+                "type": "IMAGE",
+                "link": 19
+              },
+              {
+                "localized_name": "resolution",
+                "name": "resolution",
+                "type": "INT",
+                "widget": {
+                  "name": "resolution"
+                },
+                "link": 22
+              },
+              {
+                "localized_name": "resize_method",
+                "name": "resize_method",
+                "type": "COMBO",
+                "widget": {
+                  "name": "resize_method"
+                },
+                "link": 23
+              },
+              {
+                "localized_name": "mode",
+                "name": "mode",
+                "type": "COMFY_DYNAMICCOMBO_V3",
+                "widget": {
+                  "name": "mode"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "da3_geometry",
+                "name": "da3_geometry",
+                "type": "DA3_GEOMETRY",
+                "slot_index": 0,
+                "links": [
+                  12
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "DA3Inference",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0"
+            },
+            "widgets_values": [
+              504,
+              "upper_bound_resize",
+              "mono"
+            ]
+          },
+          {
+            "id": 88,
+            "type": "LoadDA3Model",
+            "pos": [
+              810,
+              -160
+            ],
+            "size": [
+              400,
+              140
+            ],
+            "flags": {},
+            "order": 2,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "model_name",
+                "name": "model_name",
+                "type": "COMBO",
+                "widget": {
+                  "name": "model_name"
+                },
+                "link": 106
+              },
+              {
+                "localized_name": "weight_dtype",
+                "name": "weight_dtype",
+                "type": "COMBO",
+                "widget": {
+                  "name": "weight_dtype"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "DA3_MODEL",
+                "name": "DA3_MODEL",
+                "type": "DA3_MODEL",
+                "links": [
+                  107
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "LoadDA3Model",
+              "cnr_id": "comfy-core",
+              "ver": "0.24.0",
+              "models": [
+                {
+                  "name": "depth_anything_3_mono_large.safetensors",
+                  "url": "https://huggingface.co/Comfy-Org/Depth-Anything-3/resolve/main/geometry_estimation/depth_anything_3_mono_large.safetensors",
+                  "directory": "geometry_estimation"
+                }
+              ]
+            },
+            "widgets_values": [
+              "depth_anything_3_mono_large.safetensors",
+              "default"
+            ]
+          }
+        ],
+        "groups": [],
+        "links": [
+          {
+            "id": 12,
+            "origin_id": 87,
+            "origin_slot": 0,
+            "target_id": 86,
+            "target_slot": 0,
+            "type": "DA3_GEOMETRY"
+          },
+          {
+            "id": 19,
+            "origin_id": -10,
+            "origin_slot": 0,
+            "target_id": 87,
+            "target_slot": 1,
+            "type": "IMAGE"
+          },
+          {
+            "id": 7,
+            "origin_id": 86,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 0,
+            "type": "IMAGE"
+          },
+          {
+            "id": 22,
+            "origin_id": -10,
+            "origin_slot": 1,
+            "target_id": 87,
+            "target_slot": 2,
+            "type": "INT"
+          },
+          {
+            "id": 23,
+            "origin_id": -10,
+            "origin_slot": 2,
+            "target_id": 87,
+            "target_slot": 3,
+            "type": "COMBO"
+          },
+          {
+            "id": 24,
+            "origin_id": -10,
+            "origin_slot": 3,
+            "target_id": 86,
+            "target_slot": 1,
+            "type": "COMFY_DYNAMICCOMBO_V3"
+          },
+          {
+            "id": 25,
+            "origin_id": -10,
+            "origin_slot": 4,
+            "target_id": 86,
+            "target_slot": 2,
+            "type": "COMBO"
+          },
+          {
+            "id": 26,
+            "origin_id": -10,
+            "origin_slot": 5,
+            "target_id": 86,
+            "target_slot": 3,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 106,
+            "origin_id": -10,
+            "origin_slot": 6,
+            "target_id": 88,
+            "target_slot": 0,
+            "type": "COMBO"
+          },
+          {
+            "id": 107,
+            "origin_id": 88,
+            "origin_slot": 0,
+            "target_id": 87,
+            "target_slot": 0,
+            "type": "DA3_MODEL"
+          }
+        ],
+        "extra": {},
+        "category": "Conditioning & Preprocessors/Depth",
+        "description": "This subgraph takes an input image and produces a depth map using the Depth Anything 3 model, which recovers spatially consistent geometry from any number of views. It is ideal for single or multi-view images, videos, and 3D scenes where accurate depth estimation is needed for tasks like SLAM, novel view synthesis, or spatial perception. The model uses a plain transformer backbone and supports both monocular and multi-view inputs without."
+      }
+    ]
+  },
+  "extra": {
+    "BlueprintDescription": "This subgraph takes an input image and produces a depth map using the Depth Anything 3 model, which recovers spatially consistent geometry from any number of views. It is ideal for single or multi-view images, videos, and 3D scenes where accurate depth estimation is needed for tasks like SLAM, novel view synthesis, or spatial perception. The model uses a plain transformer backbone and supports both monocular and multi-view inputs without."
+  }
+}

blueprints/Image Depth Estimation (Lotus Depth).json

@@ -1,19 +1,18 @@
 {
-  "id": "6af0a6c1-0161-4528-8685-65776e838d44",
   "revision": 0,
-  "last_node_id": 75,
-  "last_link_id": 245,
+  "last_node_id": 76,
+  "last_link_id": 0,
   "nodes": [
     {
-      "id": 75,
-      "type": "488652fd-6edf-4d06-8f9f-4d84d3a34eaf",
+      "id": 76,
+      "type": "96338968-1242-4f02-b6a1-d496af4bcffe",
       "pos": [
-        600,
-        830
+        670,
+        1280
       ],
       "size": [
         400,
-        110
+        201.3125
       ],
       "flags": {},
       "order": 0,
@@ -59,47 +58,44 @@
           "links": []
         }
       ],
+      "title": "Image Depth Estimation (Lotus Depth)",
       "properties": {
         "proxyWidgets": [
           [
-            "-1",
+            "28",
             "sigma"
           ],
           [
-            "-1",
+            "10",
             "unet_name"
           ],
           [
-            "-1",
+            "14",
             "vae_name"
           ]
         ],
         "cnr_id": "comfy-core",
         "ver": "0.14.1"
       },
-      "widgets_values": [
-        999.0000000000002,
-        "lotus-depth-d-v1-1.safetensors",
-        "vae-ft-mse-840000-ema-pruned.safetensors"
-      ]
+      "widgets_values": []
     }
   ],
   "links": [],
-  "groups": [],
+  "version": 0.4,
   "definitions": {
     "subgraphs": [
       {
-        "id": "488652fd-6edf-4d06-8f9f-4d84d3a34eaf",
+        "id": "96338968-1242-4f02-b6a1-d496af4bcffe",
         "version": 1,
         "state": {
           "lastGroupId": 1,
-          "lastNodeId": 75,
+          "lastNodeId": 76,
           "lastLinkId": 245,
           "lastRerouteId": 0
         },
         "revision": 0,
         "config": {},
-        "name": "Image to Depth Map (Lotus)",
+        "name": "Image Depth Estimation (Lotus Depth)",
         "inputNode": {
           "id": -10,
           "bounding": [
@@ -191,12 +187,12 @@
             "id": 10,
             "type": "UNETLoader",
             "pos": [
-              108.05555555555557,
-              -253.05555555555557
+              110,
+              -250
             ],
             "size": [
-              254.93706597222226,
-              82
+              260,
+              90
             ],
             "flags": {},
             "order": 4,
@@ -234,9 +230,9 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "UNETLoader",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "UNETLoader",
               "models": [
                 {
                   "name": "lotus-depth-d-v1-1.safetensors",
@@ -255,12 +251,12 @@
             "id": 18,
             "type": "DisableNoise",
             "pos": [
-              607.0641494069639,
-              -268.33337840371513
+              610,
+              -270
             ],
             "size": [
-              175,
-              33.333333333333336
+              180,
+              40
             ],
             "flags": {},
             "order": 0,
@@ -278,26 +274,25 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "DisableNoise",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "DisableNoise",
               "widget_ue_connectable": {}
-            },
-            "widgets_values": []
+            }
           },
           {
-            "id": 23,
+            "id": 74,
             "type": "VAEEncode",
             "pos": [
               620,
               160
             ],
             "size": [
-              175,
+              180,
               50
             ],
             "flags": {},
-            "order": 10,
+            "order": 11,
             "mode": 0,
             "inputs": [
               {
@@ -325,12 +320,11 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "VAEEncode",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "VAEEncode",
               "widget_ue_connectable": {}
-            },
-            "widgets_values": []
+            }
           },
           {
             "id": 21,
@@ -341,7 +335,7 @@
             ],
             "size": [
               210,
-              58
+              60
             ],
             "flags": {},
             "order": 1,
@@ -369,9 +363,9 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "KSamplerSelect",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "KSamplerSelect",
               "widget_ue_connectable": {}
             },
             "widgets_values": [
@@ -386,7 +380,7 @@
               -170
             ],
             "size": [
-              175,
+              180,
               50
             ],
             "flags": {},
@@ -418,12 +412,11 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "BasicGuider",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "BasicGuider",
               "widget_ue_connectable": {}
-            },
-            "widgets_values": []
+            }
           },
           {
             "id": 16,
@@ -433,8 +426,8 @@
               -130
             ],
             "size": [
-              295.99609375,
-              271.65798611111114
+              300,
+              280
             ],
             "flags": {},
             "order": 6,
@@ -490,12 +483,11 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "SamplerCustomAdvanced",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "SamplerCustomAdvanced",
               "widget_ue_connectable": {}
-            },
-            "widgets_values": []
+            }
           },
           {
             "id": 28,
@@ -506,10 +498,10 @@
             ],
             "size": [
               210,
-              58
+              60
             ],
             "flags": {},
-            "order": 11,
+            "order": 10,
             "mode": 0,
             "inputs": [
               {
@@ -540,9 +532,9 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "SetFirstSigma",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "SetFirstSigma",
               "widget_ue_connectable": {}
             },
             "widgets_values": [
@@ -557,7 +549,7 @@
               -120
             ],
             "size": [
-              175,
+              180,
               50
             ],
             "flags": {},
@@ -589,12 +581,11 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "VAEDecode",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "VAEDecode",
               "widget_ue_connectable": {}
-            },
-            "widgets_values": []
+            }
           },
           {
             "id": 22,
@@ -604,8 +595,8 @@
               -220
             ],
             "size": [
-              175,
-              33.333333333333336
+              180,
+              40
             ],
             "flags": {},
             "order": 9,
@@ -630,12 +621,11 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "ImageInvert",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "ImageInvert",
               "widget_ue_connectable": {}
-            },
-            "widgets_values": []
+            }
           },
           {
             "id": 14,
@@ -645,8 +635,8 @@
               -90
             ],
             "size": [
-              254.93706597222226,
-              58
+              260,
+              60
             ],
             "flags": {},
             "order": 5,
@@ -675,9 +665,9 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "VAELoader",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "VAELoader",
               "models": [
                 {
                   "name": "vae-ft-mse-840000-ema-pruned.safetensors",
@@ -692,15 +682,15 @@
             ]
           },
           {
-            "id": 68,
+            "id": 75,
             "type": "LotusConditioning",
             "pos": [
               400,
               -150
             ],
             "size": [
-              175,
-              33.333333333333336
+              180,
+              40
             ],
             "flags": {},
             "order": 2,
@@ -718,12 +708,11 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "LotusConditioning",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "LotusConditioning",
               "widget_ue_connectable": {}
-            },
-            "widgets_values": []
+            }
           },
           {
             "id": 20,
@@ -734,7 +723,7 @@
             ],
             "size": [
               210,
-              106
+              110
             ],
             "flags": {},
             "order": 8,
@@ -786,9 +775,9 @@
               }
             ],
             "properties": {
+              "Node name for S&R": "BasicScheduler",
               "cnr_id": "comfy-core",
               "ver": "0.3.34",
-              "Node name for S&R": "BasicScheduler",
               "widget_ue_connectable": {}
             },
             "widgets_values": [
@@ -850,7 +839,7 @@
           },
           {
             "id": 201,
-            "origin_id": 23,
+            "origin_id": 74,
             "origin_slot": 0,
             "target_id": 16,
             "target_slot": 4,
@@ -866,7 +855,7 @@
           },
           {
             "id": 238,
-            "origin_id": 68,
+            "origin_id": 75,
             "origin_slot": 0,
             "target_id": 19,
             "target_slot": 1,
@@ -892,7 +881,7 @@
             "id": 38,
             "origin_id": 14,
             "origin_slot": 0,
-            "target_id": 23,
+            "target_id": 74,
             "target_slot": 1,
             "type": "VAE"
           },
@@ -908,7 +897,7 @@
             "id": 37,
             "origin_id": -10,
             "origin_slot": 0,
-            "target_id": 23,
+            "target_id": 74,
             "target_slot": 0,
             "type": "IMAGE"
           },
@@ -948,12 +937,11 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Image generation and editing/Depth to image",
+        "category": "Conditioning & Preprocessors/Depth",
         "description": "Estimates a monocular depth map from an input image using the Lotus depth estimation model."
       }
     ]
   },
-  "config": {},
   "extra": {
     "ds": {
       "scale": 1.3589709866044692,
@@ -961,8 +949,6 @@
         -138.53613935617864,
         -786.0629126022195
       ]
-    },
-    "workflowRendererVersion": "LG"
-  },
-  "version": 0.4
+    }
+  }
 }

blueprints/Image Face Detection (Mediapipe).json

@@ -0,0 +1,779 @@
+{
+  "revision": 0,
+  "last_node_id": 33,
+  "last_link_id": 0,
+  "nodes": [
+    {
+      "id": 33,
+      "type": "6062babb-b649-4a71-be9e-20ebce567744",
+      "pos": [
+        -450,
+        4240
+      ],
+      "size": [
+        420,
+        400
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "image",
+          "name": "image",
+          "type": "IMAGE",
+          "link": null
+        },
+        {
+          "name": "face_landmarker",
+          "type": "FACE_LANDMARKER",
+          "link": null
+        },
+        {
+          "name": "detector_variant",
+          "type": "COMBO",
+          "widget": {
+            "name": "detector_variant"
+          },
+          "link": null
+        },
+        {
+          "name": "num_faces",
+          "type": "INT",
+          "widget": {
+            "name": "num_faces"
+          },
+          "link": null
+        },
+        {
+          "label": "custom_face_oval",
+          "name": "regions.face_oval",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "regions.face_oval"
+          },
+          "link": null
+        },
+        {
+          "label": "custom_lips",
+          "name": "regions.lips",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "regions.lips"
+          },
+          "link": null
+        },
+        {
+          "label": "custom_left_eye",
+          "name": "regions.left_eye",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "regions.left_eye"
+          },
+          "link": null
+        },
+        {
+          "label": "custom_right_eye",
+          "name": "regions.right_eye",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "regions.right_eye"
+          },
+          "link": null
+        },
+        {
+          "label": "custom_irises",
+          "name": "regions.irises",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "regions.irises"
+          },
+          "link": null
+        },
+        {
+          "name": "model_name",
+          "type": "COMBO",
+          "widget": {
+            "name": "model_name"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "face_landmarks",
+          "name": "face_landmarks",
+          "type": "FACE_LANDMARKS",
+          "links": []
+        },
+        {
+          "localized_name": "bboxes",
+          "name": "bboxes",
+          "type": "BOUNDING_BOX",
+          "links": []
+        },
+        {
+          "label": "mask",
+          "name": "MASK_1",
+          "type": "MASK",
+          "links": []
+        }
+      ],
+      "title": "Image Face Detection (Mediapipe)",
+      "properties": {
+        "proxyWidgets": [
+          [
+            "11",
+            "detector_variant"
+          ],
+          [
+            "11",
+            "num_faces"
+          ],
+          [
+            "20",
+            "regions.face_oval"
+          ],
+          [
+            "20",
+            "regions.lips"
+          ],
+          [
+            "20",
+            "regions.left_eye"
+          ],
+          [
+            "20",
+            "regions.right_eye"
+          ],
+          [
+            "20",
+            "regions.irises"
+          ],
+          [
+            "2",
+            "model_name"
+          ]
+        ],
+        "cnr_id": "comfy-core",
+        "ver": "0.22.0",
+        "enableTabs": false,
+        "tabWidth": 65,
+        "tabXOffset": 10,
+        "hasSecondTab": false,
+        "secondTabText": "Send Back",
+        "secondTabOffset": 80,
+        "secondTabWidth": 65
+      },
+      "widgets_values": []
+    }
+  ],
+  "links": [],
+  "version": 0.4,
+  "definitions": {
+    "subgraphs": [
+      {
+        "id": "6062babb-b649-4a71-be9e-20ebce567744",
+        "version": 1,
+        "state": {
+          "lastGroupId": 2,
+          "lastNodeId": 158,
+          "lastLinkId": 140,
+          "lastRerouteId": 0
+        },
+        "revision": 0,
+        "config": {},
+        "name": "Image Face Detection (Mediapipe)",
+        "description": "Detects facial landmarks from an image using MediaPipe, outputting landmark data, face bounding boxes, and an optional face-region mask.",
+        "inputNode": {
+          "id": -10,
+          "bounding": [
+            -710,
+            4300,
+            148.880859375,
+            248
+          ]
+        },
+        "outputNode": {
+          "id": -20,
+          "bounding": [
+            140,
+            4480,
+            137.677734375,
+            108
+          ]
+        },
+        "inputs": [
+          {
+            "id": "705dc1ae-6dc9-4155-92df-52f816ad451e",
+            "name": "image",
+            "type": "IMAGE",
+            "linkIds": [
+              60
+            ],
+            "localized_name": "image",
+            "pos": [
+              -585.119140625,
+              4324
+            ]
+          },
+          {
+            "id": "d6277190-732c-4604-b7cd-d3a9588bf761",
+            "name": "face_landmarker",
+            "type": "FACE_LANDMARKER",
+            "linkIds": [
+              74
+            ],
+            "pos": [
+              -585.119140625,
+              4344
+            ]
+          },
+          {
+            "id": "ac473a08-6a86-42a7-b460-e70c6c5e1e2b",
+            "name": "detector_variant",
+            "type": "COMBO",
+            "linkIds": [
+              75
+            ],
+            "pos": [
+              -585.119140625,
+              4364
+            ]
+          },
+          {
+            "id": "1bec2252-ca2d-496e-8a33-33a61d21f897",
+            "name": "num_faces",
+            "type": "INT",
+            "linkIds": [
+              76
+            ],
+            "pos": [
+              -585.119140625,
+              4384
+            ]
+          },
+          {
+            "id": "17994fa2-0ea0-4c9b-a70a-19789c459c80",
+            "name": "regions.face_oval",
+            "type": "BOOLEAN",
+            "linkIds": [
+              77
+            ],
+            "label": "custom_face_oval",
+            "pos": [
+              -585.119140625,
+              4404
+            ]
+          },
+          {
+            "id": "1c6c5893-2aee-4c37-b702-15ef2e20d863",
+            "name": "regions.lips",
+            "type": "BOOLEAN",
+            "linkIds": [
+              78
+            ],
+            "label": "custom_lips",
+            "pos": [
+              -585.119140625,
+              4424
+            ]
+          },
+          {
+            "id": "f353fcea-4b6f-42a1-8fdd-32b3aa1e1f09",
+            "name": "regions.left_eye",
+            "type": "BOOLEAN",
+            "linkIds": [
+              79
+            ],
+            "label": "custom_left_eye",
+            "pos": [
+              -585.119140625,
+              4444
+            ]
+          },
+          {
+            "id": "1387e121-c1fb-4522-8f0d-43459e11dd86",
+            "name": "regions.right_eye",
+            "type": "BOOLEAN",
+            "linkIds": [
+              80
+            ],
+            "label": "custom_right_eye",
+            "pos": [
+              -585.119140625,
+              4464
+            ]
+          },
+          {
+            "id": "14acb0a0-d1f4-48f3-ba31-811b26236ef9",
+            "name": "regions.irises",
+            "type": "BOOLEAN",
+            "linkIds": [
+              81
+            ],
+            "label": "custom_irises",
+            "pos": [
+              -585.119140625,
+              4484
+            ]
+          },
+          {
+            "id": "25a82859-87de-42c8-8431-09948665546e",
+            "name": "model_name",
+            "type": "COMBO",
+            "linkIds": [
+              86
+            ],
+            "pos": [
+              -585.119140625,
+              4504
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "id": "d2ba3f92-e8b1-49c3-9590-cfad56c54cf4",
+            "name": "face_landmarks",
+            "type": "FACE_LANDMARKS",
+            "linkIds": [
+              44
+            ],
+            "localized_name": "face_landmarks",
+            "pos": [
+              164,
+              4504
+            ]
+          },
+          {
+            "id": "4f356bb0-d4c4-4f93-b4cf-0845a65c4e6d",
+            "name": "bboxes",
+            "type": "BOUNDING_BOX",
+            "linkIds": [
+              25
+            ],
+            "localized_name": "bboxes",
+            "pos": [
+              164,
+              4524
+            ]
+          },
+          {
+            "id": "f6309e1d-6397-4363-b38f-778a122abc51",
+            "name": "MASK_1",
+            "type": "MASK",
+            "linkIds": [
+              83
+            ],
+            "label": "mask",
+            "pos": [
+              164,
+              4544
+            ]
+          }
+        ],
+        "widgets": [],
+        "nodes": [
+          {
+            "id": 11,
+            "type": "MediaPipeFaceLandmarker",
+            "pos": [
+              -280,
+              4280
+            ],
+            "size": [
+              350,
+              220
+            ],
+            "flags": {},
+            "order": 1,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "face_detection_model",
+                "name": "face_detection_model",
+                "type": "FACE_DETECTION_MODEL",
+                "link": 66
+              },
+              {
+                "localized_name": "image",
+                "name": "image",
+                "type": "IMAGE",
+                "link": 60
+              },
+              {
+                "localized_name": "detector_variant",
+                "name": "detector_variant",
+                "type": "COMBO",
+                "widget": {
+                  "name": "detector_variant"
+                },
+                "link": 75
+              },
+              {
+                "localized_name": "num_faces",
+                "name": "num_faces",
+                "type": "INT",
+                "widget": {
+                  "name": "num_faces"
+                },
+                "link": 76
+              },
+              {
+                "localized_name": "min_confidence",
+                "name": "min_confidence",
+                "type": "FLOAT",
+                "widget": {
+                  "name": "min_confidence"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "missing_frame_fallback",
+                "name": "missing_frame_fallback",
+                "type": "COMBO",
+                "widget": {
+                  "name": "missing_frame_fallback"
+                },
+                "link": null
+              },
+              {
+                "name": "face_landmarker",
+                "type": "FACE_LANDMARKER",
+                "link": 74
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "face_landmarks",
+                "name": "face_landmarks",
+                "type": "FACE_LANDMARKS",
+                "links": [
+                  44,
+                  46
+                ]
+              },
+              {
+                "localized_name": "bboxes",
+                "name": "bboxes",
+                "type": "BOUNDING_BOX",
+                "links": [
+                  25
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "MediaPipeFaceLandmarker",
+              "cnr_id": "comfy-core",
+              "ver": "0.22.0",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65
+            },
+            "widgets_values": [
+              "full",
+              0,
+              0.5,
+              "empty"
+            ]
+          },
+          {
+            "id": 2,
+            "type": "LoadMediaPipeFaceLandmarker",
+            "pos": [
+              -290,
+              4060
+            ],
+            "size": [
+              350,
+              140
+            ],
+            "flags": {},
+            "order": 0,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "model_name",
+                "name": "model_name",
+                "type": "COMBO",
+                "widget": {
+                  "name": "model_name"
+                },
+                "link": 86
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "FACE_DETECTION_MODEL",
+                "name": "FACE_DETECTION_MODEL",
+                "type": "FACE_DETECTION_MODEL",
+                "links": [
+                  66
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "LoadMediaPipeFaceLandmarker",
+              "cnr_id": "comfy-core",
+              "ver": "0.22.0",
+              "models": [
+                {
+                  "name": "mediapipe_face_fp32.safetensors",
+                  "url": "https://huggingface.co/Comfy-Org/mediapipe/resolve/main/detection/mediapipe_face_fp32.safetensors",
+                  "directory": "detection"
+                }
+              ],
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65
+            },
+            "widgets_values": [
+              "mediapipe_face_fp32.safetensors"
+            ]
+          },
+          {
+            "id": 20,
+            "type": "MediaPipeFaceMask",
+            "pos": [
+              -290,
+              4560
+            ],
+            "size": [
+              360,
+              180
+            ],
+            "flags": {},
+            "order": 2,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "face_landmarks",
+                "name": "face_landmarks",
+                "type": "FACE_LANDMARKS",
+                "link": 46
+              },
+              {
+                "localized_name": "regions",
+                "name": "regions",
+                "type": "COMFY_DYNAMICCOMBO_V3",
+                "widget": {
+                  "name": "regions"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "regions.face_oval",
+                "name": "regions.face_oval",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "regions.face_oval"
+                },
+                "link": 77
+              },
+              {
+                "localized_name": "regions.lips",
+                "name": "regions.lips",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "regions.lips"
+                },
+                "link": 78
+              },
+              {
+                "localized_name": "regions.left_eye",
+                "name": "regions.left_eye",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "regions.left_eye"
+                },
+                "link": 79
+              },
+              {
+                "localized_name": "regions.right_eye",
+                "name": "regions.right_eye",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "regions.right_eye"
+                },
+                "link": 80
+              },
+              {
+                "localized_name": "regions.irises",
+                "name": "regions.irises",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "regions.irises"
+                },
+                "link": 81
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "MASK",
+                "name": "MASK",
+                "type": "MASK",
+                "links": [
+                  83
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "MediaPipeFaceMask",
+              "cnr_id": "comfy-core",
+              "ver": "0.22.0",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65
+            },
+            "widgets_values": [
+              "custom",
+              true,
+              false,
+              false,
+              false,
+              false
+            ]
+          }
+        ],
+        "groups": [],
+        "links": [
+          {
+            "id": 66,
+            "origin_id": 2,
+            "origin_slot": 0,
+            "target_id": 11,
+            "target_slot": 0,
+            "type": "FACE_DETECTION_MODEL"
+          },
+          {
+            "id": 46,
+            "origin_id": 11,
+            "origin_slot": 0,
+            "target_id": 20,
+            "target_slot": 0,
+            "type": "FACE_LANDMARKS"
+          },
+          {
+            "id": 60,
+            "origin_id": -10,
+            "origin_slot": 0,
+            "target_id": 11,
+            "target_slot": 1,
+            "type": "IMAGE"
+          },
+          {
+            "id": 44,
+            "origin_id": 11,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 0,
+            "type": "FACE_LANDMARKS"
+          },
+          {
+            "id": 25,
+            "origin_id": 11,
+            "origin_slot": 1,
+            "target_id": -20,
+            "target_slot": 1,
+            "type": "BOUNDING_BOX"
+          },
+          {
+            "id": 74,
+            "origin_id": -10,
+            "origin_slot": 1,
+            "target_id": 11,
+            "target_slot": 6,
+            "type": "FACE_LANDMARKER"
+          },
+          {
+            "id": 75,
+            "origin_id": -10,
+            "origin_slot": 2,
+            "target_id": 11,
+            "target_slot": 2,
+            "type": "COMBO"
+          },
+          {
+            "id": 76,
+            "origin_id": -10,
+            "origin_slot": 3,
+            "target_id": 11,
+            "target_slot": 3,
+            "type": "INT"
+          },
+          {
+            "id": 77,
+            "origin_id": -10,
+            "origin_slot": 4,
+            "target_id": 20,
+            "target_slot": 2,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 78,
+            "origin_id": -10,
+            "origin_slot": 5,
+            "target_id": 20,
+            "target_slot": 3,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 79,
+            "origin_id": -10,
+            "origin_slot": 6,
+            "target_id": 20,
+            "target_slot": 4,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 80,
+            "origin_id": -10,
+            "origin_slot": 7,
+            "target_id": 20,
+            "target_slot": 5,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 81,
+            "origin_id": -10,
+            "origin_slot": 8,
+            "target_id": 20,
+            "target_slot": 6,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 83,
+            "origin_id": 20,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 2,
+            "type": "MASK"
+          },
+          {
+            "id": 86,
+            "origin_id": -10,
+            "origin_slot": 9,
+            "target_id": 2,
+            "target_slot": 0,
+            "type": "COMBO"
+          }
+        ],
+        "extra": {},
+        "category": "Conditioning & Preprocessors/Face Detection"
+      }
+    ]
+  },
+  "extra": {}
+}

blueprints/Image Segmentation (SAM3).json

@@ -703,7 +703,7 @@
           }
         ],
         "extra": {},
-        "category": "Image Tools/Image Segmentation",
+        "category": "Conditioning & Preprocessors/Segmentation & Mask",
         "description": "Segments images into masks using Meta SAM3 from text prompts, points, or boxes."
       }
     ]

blueprints/Image Upscale(Z-image-Turbo).json

@@ -1302,7 +1302,7 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Image generation and editing/Enhance",
+        "category": "Image generation and editing/Upscale",
         "description": "Upscales images to higher resolution using Z-Image-Turbo."
       }
     ]
@@ -1312,4 +1312,4 @@
     "workflowRendererVersion": "LG"
   },
   "version": 0.4
-}
+}

blueprints/Image to Pose Map (SDPose-OOD).json

@@ -0,0 +1,888 @@
+{
+  "revision": 0,
+  "last_node_id": 675,
+  "last_link_id": 0,
+  "nodes": [
+    {
+      "id": 675,
+      "type": "01b6a731-fb78-4070-9a38-c87146da9604",
+      "pos": [
+        -2480,
+        3400
+      ],
+      "size": [
+        360,
+        433.3125
+      ],
+      "flags": {},
+      "order": 2,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "input",
+          "name": "input",
+          "type": "IMAGE,MASK",
+          "link": null
+        },
+        {
+          "label": "resize_target_longer_size",
+          "name": "resize_type.longer_size",
+          "type": "INT",
+          "widget": {
+            "name": "resize_type.longer_size"
+          },
+          "link": null
+        },
+        {
+          "name": "scale_method",
+          "type": "COMBO",
+          "widget": {
+            "name": "scale_method"
+          },
+          "link": null
+        },
+        {
+          "name": "draw_body",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "draw_body"
+          },
+          "link": null
+        },
+        {
+          "name": "draw_hands",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "draw_hands"
+          },
+          "link": null
+        },
+        {
+          "name": "draw_face",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "draw_face"
+          },
+          "link": null
+        },
+        {
+          "name": "draw_feet",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "draw_feet"
+          },
+          "link": null
+        },
+        {
+          "name": "stick_width",
+          "type": "INT",
+          "widget": {
+            "name": "stick_width"
+          },
+          "link": null
+        },
+        {
+          "name": "face_point_size",
+          "type": "INT",
+          "widget": {
+            "name": "face_point_size"
+          },
+          "link": null
+        },
+        {
+          "name": "score_threshold",
+          "type": "FLOAT",
+          "widget": {
+            "name": "score_threshold"
+          },
+          "link": null
+        },
+        {
+          "name": "ckpt_name",
+          "type": "COMBO",
+          "widget": {
+            "name": "ckpt_name"
+          },
+          "link": null
+        },
+        {
+          "name": "bboxes",
+          "shape": 7,
+          "type": "BOUNDING_BOX",
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "IMAGE",
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": []
+        },
+        {
+          "name": "keypoints",
+          "type": "POSE_KEYPOINT",
+          "links": null
+        }
+      ],
+      "properties": {
+        "proxyWidgets": [
+          [
+            "674",
+            "resize_type.longer_size"
+          ],
+          [
+            "674",
+            "scale_method"
+          ],
+          [
+            "672",
+            "draw_body"
+          ],
+          [
+            "672",
+            "draw_hands"
+          ],
+          [
+            "672",
+            "draw_face"
+          ],
+          [
+            "672",
+            "draw_feet"
+          ],
+          [
+            "672",
+            "stick_width"
+          ],
+          [
+            "672",
+            "face_point_size"
+          ],
+          [
+            "672",
+            "score_threshold"
+          ],
+          [
+            "673",
+            "ckpt_name"
+          ]
+        ],
+        "cnr_id": "comfy-core",
+        "ver": "0.15.1",
+        "ue_properties": {
+          "widget_ue_connectable": {},
+          "version": "7.7",
+          "input_ue_unconnectable": {}
+        }
+      },
+      "widgets_values": [],
+      "title": "Image to Pose Map (SDPose-OOD)"
+    }
+  ],
+  "links": [],
+  "version": 0.4,
+  "definitions": {
+    "subgraphs": [
+      {
+        "id": "01b6a731-fb78-4070-9a38-c87146da9604",
+        "version": 1,
+        "state": {
+          "lastGroupId": 0,
+          "lastNodeId": 676,
+          "lastLinkId": 1715,
+          "lastRerouteId": 0
+        },
+        "revision": 0,
+        "config": {},
+        "name": "Image to Pose Map (SDPose-OOD)",
+        "inputNode": {
+          "id": -10,
+          "bounding": [
+            -3290,
+            3590,
+            190.8984375,
+            288
+          ]
+        },
+        "outputNode": {
+          "id": -20,
+          "bounding": [
+            -1756.2451602089645,
+            3366,
+            128,
+            88
+          ]
+        },
+        "inputs": [
+          {
+            "id": "e24699c3-1356-4634-9eb4-19bb58e5c0b0",
+            "name": "input",
+            "type": "IMAGE,MASK",
+            "linkIds": [
+              1700
+            ],
+            "localized_name": "input",
+            "pos": [
+              -3123.1015625,
+              3614
+            ]
+          },
+          {
+            "id": "088eefc1-cd8a-4573-993f-9e4da008a12d",
+            "name": "resize_type.longer_size",
+            "type": "INT",
+            "linkIds": [
+              1704
+            ],
+            "label": "resize_target_longer_size",
+            "pos": [
+              -3123.1015625,
+              3634
+            ]
+          },
+          {
+            "id": "b6449bd3-73d4-41c8-b81f-cf8d33f76a2e",
+            "name": "scale_method",
+            "type": "COMBO",
+            "linkIds": [
+              1705
+            ],
+            "pos": [
+              -3123.1015625,
+              3654
+            ]
+          },
+          {
+            "id": "4cff52ad-ed07-4c97-8803-fcbd89554fd0",
+            "name": "draw_body",
+            "type": "BOOLEAN",
+            "linkIds": [
+              1706
+            ],
+            "pos": [
+              -3123.1015625,
+              3674
+            ]
+          },
+          {
+            "id": "7af63dce-f7df-4d7e-8215-d7c7f60bf81c",
+            "name": "draw_hands",
+            "type": "BOOLEAN",
+            "linkIds": [
+              1707
+            ],
+            "pos": [
+              -3123.1015625,
+              3694
+            ]
+          },
+          {
+            "id": "af3a9bce-61f9-4aca-b530-9f65e028b35e",
+            "name": "draw_face",
+            "type": "BOOLEAN",
+            "linkIds": [
+              1708
+            ],
+            "pos": [
+              -3123.1015625,
+              3714
+            ]
+          },
+          {
+            "id": "4620f6a3-2c85-4b79-ad8f-35d0326b568f",
+            "name": "draw_feet",
+            "type": "BOOLEAN",
+            "linkIds": [
+              1709
+            ],
+            "pos": [
+              -3123.1015625,
+              3734
+            ]
+          },
+          {
+            "id": "fee5d0c9-8d4b-4934-81d8-ba2206dc56cb",
+            "name": "stick_width",
+            "type": "INT",
+            "linkIds": [
+              1710
+            ],
+            "pos": [
+              -3123.1015625,
+              3754
+            ]
+          },
+          {
+            "id": "aafdd060-ba81-4324-a9cc-b656e1ebc133",
+            "name": "face_point_size",
+            "type": "INT",
+            "linkIds": [
+              1711
+            ],
+            "pos": [
+              -3123.1015625,
+              3774
+            ]
+          },
+          {
+            "id": "514c5503-f9e6-4d23-b1ae-1d3291acb2a3",
+            "name": "score_threshold",
+            "type": "FLOAT",
+            "linkIds": [
+              1712
+            ],
+            "pos": [
+              -3123.1015625,
+              3794
+            ]
+          },
+          {
+            "id": "ae46de61-2cc6-483e-8ee9-87e4144a2ffa",
+            "name": "ckpt_name",
+            "type": "COMBO",
+            "linkIds": [
+              1713
+            ],
+            "pos": [
+              -3123.1015625,
+              3814
+            ]
+          },
+          {
+            "id": "41bec0c6-dffa-4c78-9289-ee678715ae54",
+            "name": "bboxes",
+            "type": "BOUNDING_BOX",
+            "linkIds": [
+              1714
+            ],
+            "pos": [
+              -3123.1015625,
+              3834
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "id": "f05ed8cc-9403-4f14-8085-4364b06f8a48",
+            "name": "IMAGE",
+            "type": "IMAGE",
+            "linkIds": [
+              1701
+            ],
+            "localized_name": "IMAGE",
+            "pos": [
+              -1732.2451602089645,
+              3390
+            ]
+          },
+          {
+            "id": "29a6584e-4685-4986-8ffd-e6d8539953fd",
+            "name": "keypoints",
+            "type": "POSE_KEYPOINT",
+            "linkIds": [
+              1715
+            ],
+            "pos": [
+              -1732.2451602089645,
+              3410
+            ]
+          }
+        ],
+        "widgets": [],
+        "nodes": [
+          {
+            "id": 671,
+            "type": "SDPoseKeypointExtractor",
+            "pos": [
+              -2470,
+              3250
+            ],
+            "size": [
+              270,
+              180
+            ],
+            "flags": {},
+            "order": 0,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "model",
+                "name": "model",
+                "type": "MODEL",
+                "link": 1696
+              },
+              {
+                "localized_name": "vae",
+                "name": "vae",
+                "type": "VAE",
+                "link": 1697
+              },
+              {
+                "localized_name": "image",
+                "name": "image",
+                "type": "IMAGE",
+                "link": 1698
+              },
+              {
+                "localized_name": "bboxes",
+                "name": "bboxes",
+                "shape": 7,
+                "type": "BOUNDING_BOX",
+                "link": 1714
+              },
+              {
+                "localized_name": "batch_size",
+                "name": "batch_size",
+                "type": "INT",
+                "widget": {
+                  "name": "batch_size"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "keypoints",
+                "name": "keypoints",
+                "type": "POSE_KEYPOINT",
+                "links": [
+                  1699,
+                  1715
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "SDPoseKeypointExtractor",
+              "cnr_id": "comfy-core",
+              "ver": "0.15.0",
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              16
+            ]
+          },
+          {
+            "id": 674,
+            "type": "ResizeImageMaskNode",
+            "pos": [
+              -2960,
+              3490
+            ],
+            "size": [
+              270,
+              110
+            ],
+            "flags": {},
+            "order": 3,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "input",
+                "name": "input",
+                "type": "IMAGE,MASK",
+                "link": 1700
+              },
+              {
+                "localized_name": "resize_type",
+                "name": "resize_type",
+                "type": "COMFY_DYNAMICCOMBO_V3",
+                "widget": {
+                  "name": "resize_type"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "resize_type.longer_size",
+                "name": "resize_type.longer_size",
+                "type": "INT",
+                "widget": {
+                  "name": "resize_type.longer_size"
+                },
+                "link": 1704
+              },
+              {
+                "localized_name": "scale_method",
+                "name": "scale_method",
+                "type": "COMBO",
+                "widget": {
+                  "name": "scale_method"
+                },
+                "link": 1705
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "resized",
+                "name": "resized",
+                "type": "*",
+                "links": [
+                  1698
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "ResizeImageMaskNode",
+              "cnr_id": "comfy-core",
+              "ver": "0.15.0",
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              "scale longer dimension",
+              1024,
+              "area"
+            ]
+          },
+          {
+            "id": 672,
+            "type": "SDPoseDrawKeypoints",
+            "pos": [
+              -2120,
+              3260
+            ],
+            "size": [
+              270,
+              280
+            ],
+            "flags": {},
+            "order": 1,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "keypoints",
+                "name": "keypoints",
+                "type": "POSE_KEYPOINT",
+                "link": 1699
+              },
+              {
+                "localized_name": "draw_body",
+                "name": "draw_body",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "draw_body"
+                },
+                "link": 1706
+              },
+              {
+                "localized_name": "draw_hands",
+                "name": "draw_hands",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "draw_hands"
+                },
+                "link": 1707
+              },
+              {
+                "localized_name": "draw_face",
+                "name": "draw_face",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "draw_face"
+                },
+                "link": 1708
+              },
+              {
+                "localized_name": "draw_feet",
+                "name": "draw_feet",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "draw_feet"
+                },
+                "link": 1709
+              },
+              {
+                "localized_name": "stick_width",
+                "name": "stick_width",
+                "type": "INT",
+                "widget": {
+                  "name": "stick_width"
+                },
+                "link": 1710
+              },
+              {
+                "localized_name": "face_point_size",
+                "name": "face_point_size",
+                "type": "INT",
+                "widget": {
+                  "name": "face_point_size"
+                },
+                "link": 1711
+              },
+              {
+                "localized_name": "score_threshold",
+                "name": "score_threshold",
+                "type": "FLOAT",
+                "widget": {
+                  "name": "score_threshold"
+                },
+                "link": 1712
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "IMAGE",
+                "name": "IMAGE",
+                "type": "IMAGE",
+                "links": [
+                  1701
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "SDPoseDrawKeypoints",
+              "cnr_id": "comfy-core",
+              "ver": "0.15.0",
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              true,
+              true,
+              true,
+              true,
+              4,
+              2,
+              0.5
+            ]
+          },
+          {
+            "id": 673,
+            "type": "CheckpointLoaderSimple",
+            "pos": [
+              -2960,
+              3250
+            ],
+            "size": [
+              390,
+              190
+            ],
+            "flags": {},
+            "order": 2,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "ckpt_name",
+                "name": "ckpt_name",
+                "type": "COMBO",
+                "widget": {
+                  "name": "ckpt_name"
+                },
+                "link": 1713
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "MODEL",
+                "name": "MODEL",
+                "type": "MODEL",
+                "links": [
+                  1696
+                ]
+              },
+              {
+                "localized_name": "CLIP",
+                "name": "CLIP",
+                "type": "CLIP",
+                "links": []
+              },
+              {
+                "localized_name": "VAE",
+                "name": "VAE",
+                "type": "VAE",
+                "links": [
+                  1697
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "CheckpointLoaderSimple",
+              "cnr_id": "comfy-core",
+              "ver": "0.15.0",
+              "models": [
+                {
+                  "name": "sdpose_wholebody_fp16.safetensors",
+                  "url": "https://huggingface.co/Comfy-Org/SDPose/resolve/main/checkpoints/sdpose_wholebody_fp16.safetensors",
+                  "directory": "checkpoints"
+                }
+              ],
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              "sdpose_wholebody_fp16.safetensors"
+            ]
+          }
+        ],
+        "groups": [],
+        "links": [
+          {
+            "id": 1696,
+            "origin_id": 673,
+            "origin_slot": 0,
+            "target_id": 671,
+            "target_slot": 0,
+            "type": "MODEL"
+          },
+          {
+            "id": 1697,
+            "origin_id": 673,
+            "origin_slot": 2,
+            "target_id": 671,
+            "target_slot": 1,
+            "type": "VAE"
+          },
+          {
+            "id": 1698,
+            "origin_id": 674,
+            "origin_slot": 0,
+            "target_id": 671,
+            "target_slot": 2,
+            "type": "IMAGE"
+          },
+          {
+            "id": 1699,
+            "origin_id": 671,
+            "origin_slot": 0,
+            "target_id": 672,
+            "target_slot": 0,
+            "type": "POSE_KEYPOINT"
+          },
+          {
+            "id": 1700,
+            "origin_id": -10,
+            "origin_slot": 0,
+            "target_id": 674,
+            "target_slot": 0,
+            "type": "IMAGE,MASK"
+          },
+          {
+            "id": 1701,
+            "origin_id": 672,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 0,
+            "type": "IMAGE"
+          },
+          {
+            "id": 1704,
+            "origin_id": -10,
+            "origin_slot": 1,
+            "target_id": 674,
+            "target_slot": 2,
+            "type": "INT"
+          },
+          {
+            "id": 1705,
+            "origin_id": -10,
+            "origin_slot": 2,
+            "target_id": 674,
+            "target_slot": 3,
+            "type": "COMBO"
+          },
+          {
+            "id": 1706,
+            "origin_id": -10,
+            "origin_slot": 3,
+            "target_id": 672,
+            "target_slot": 1,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 1707,
+            "origin_id": -10,
+            "origin_slot": 4,
+            "target_id": 672,
+            "target_slot": 2,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 1708,
+            "origin_id": -10,
+            "origin_slot": 5,
+            "target_id": 672,
+            "target_slot": 3,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 1709,
+            "origin_id": -10,
+            "origin_slot": 6,
+            "target_id": 672,
+            "target_slot": 4,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 1710,
+            "origin_id": -10,
+            "origin_slot": 7,
+            "target_id": 672,
+            "target_slot": 5,
+            "type": "INT"
+          },
+          {
+            "id": 1711,
+            "origin_id": -10,
+            "origin_slot": 8,
+            "target_id": 672,
+            "target_slot": 6,
+            "type": "INT"
+          },
+          {
+            "id": 1712,
+            "origin_id": -10,
+            "origin_slot": 9,
+            "target_id": 672,
+            "target_slot": 7,
+            "type": "FLOAT"
+          },
+          {
+            "id": 1713,
+            "origin_id": -10,
+            "origin_slot": 10,
+            "target_id": 673,
+            "target_slot": 0,
+            "type": "COMBO"
+          },
+          {
+            "id": 1714,
+            "origin_id": -10,
+            "origin_slot": 11,
+            "target_id": 671,
+            "target_slot": 3,
+            "type": "BOUNDING_BOX"
+          },
+          {
+            "id": 1715,
+            "origin_id": 671,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 1,
+            "type": "POSE_KEYPOINT"
+          }
+        ],
+        "extra": {
+          "workflowRendererVersion": "LG"
+        },
+        "category": "Conditioning & Preprocessors/Pose",
+        "description": "Extracts human pose keypoints and stick-figure visuals from an image using SDPose-OOD, with optional bounding-box input per subject."
+      }
+    ]
+  },
+  "extra": {
+    "ue_links": []
+  }
+}

blueprints/Pose to Image (Z-Image-Turbo).json

@@ -1298,7 +1298,7 @@
           "VHS_MetadataImage": true,
           "VHS_KeepIntermediate": true
         },
-        "category": "Image generation and editing/Pose to image",
+        "category": "Image generation and editing/Conditioned",
         "description": "Generates an image from pose keypoints using Z-Image-Turbo with text conditioning."
       }
     ]

blueprints/Pose to Video (LTX 2.0).json

@@ -3870,7 +3870,7 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Video generation and editing/Pose to video",
+        "category": "Video generation and editing/Conditioned",
         "description": "Generates video from pose reference frames using LTX-2, with optional synchronized audio."
       }
     ]

blueprints/Prompt Enhance.json

@@ -270,7 +270,7 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Text generation/Prompt enhance",
+        "category": "Text Tools",
         "description": "Expands short text prompts into detailed descriptions using a text generation model for better generation quality."
       }
     ]

blueprints/Remove Background (BiRefNet).json

@@ -389,7 +389,7 @@
           }
         ],
         "extra": {},
-        "category": "Image generation and editing/Background Removal"
+        "category": "Image Tools/Background Removal"
       }
     ]
   },

blueprints/Select Per-Line Text by Index.json

@@ -0,0 +1,485 @@
+{
+  "revision": 0,
+  "last_node_id": 10,
+  "last_link_id": 0,
+  "nodes": [
+    {
+      "id": 10,
+      "type": "3fb7557a-470d-4983-9d8c-6d5caa9788f0",
+      "pos": [
+        -250,
+        8590
+      ],
+      "size": [
+        280,
+        360
+      ],
+      "flags": {},
+      "order": 0,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "text_per_line",
+          "name": "text_per_line",
+          "type": "STRING",
+          "widget": {
+            "name": "text_per_line"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "index",
+          "name": "index",
+          "type": "INT",
+          "widget": {
+            "name": "index"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "selected_line",
+          "name": "selected_line",
+          "type": "STRING",
+          "links": []
+        }
+      ],
+      "properties": {
+        "proxyWidgets": [
+          [
+            "2",
+            "string"
+          ],
+          [
+            "3",
+            "value"
+          ]
+        ],
+        "cnr_id": "comfy-core",
+        "ver": "0.19.0",
+        "ue_properties": {
+          "widget_ue_connectable": {},
+          "input_ue_unconnectable": {}
+        }
+      },
+      "widgets_values": [],
+      "title": "Select Per-Line Text by Index"
+    }
+  ],
+  "links": [],
+  "version": 0.4,
+  "definitions": {
+    "subgraphs": [
+      {
+        "id": "3fb7557a-470d-4983-9d8c-6d5caa9788f0",
+        "version": 1,
+        "state": {
+          "lastGroupId": 0,
+          "lastNodeId": 10,
+          "lastLinkId": 14,
+          "lastRerouteId": 0
+        },
+        "revision": 0,
+        "config": {},
+        "name": "Select Per-Line Text by Index",
+        "inputNode": {
+          "id": -10,
+          "bounding": [
+            -990,
+            8595,
+            128,
+            88
+          ]
+        },
+        "outputNode": {
+          "id": -20,
+          "bounding": [
+            710,
+            8585,
+            128,
+            68
+          ]
+        },
+        "inputs": [
+          {
+            "id": "75417d82-a934-4ac9-b667-d8dcd5a3bfb3",
+            "name": "text_per_line",
+            "type": "STRING",
+            "linkIds": [
+              13
+            ],
+            "localized_name": "text_per_line",
+            "pos": [
+              -886,
+              8619
+            ]
+          },
+          {
+            "id": "46e69a73-1804-4ca6-9175-31445bf0be96",
+            "name": "index",
+            "type": "INT",
+            "linkIds": [
+              14
+            ],
+            "localized_name": "index",
+            "pos": [
+              -886,
+              8639
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "id": "e34e8ad1-84d2-4bd2-a460-eb7de6067c10",
+            "name": "selected_line",
+            "type": "STRING",
+            "linkIds": [
+              10
+            ],
+            "localized_name": "selected_line",
+            "pos": [
+              734,
+              8609
+            ]
+          }
+        ],
+        "widgets": [],
+        "nodes": [
+          {
+            "id": 1,
+            "type": "PreviewAny",
+            "pos": [
+              -500,
+              8400
+            ],
+            "size": [
+              230,
+              180
+            ],
+            "flags": {},
+            "order": 0,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "source",
+                "name": "source",
+                "type": "*",
+                "link": 1
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "STRING",
+                "name": "STRING",
+                "type": "STRING",
+                "links": [
+                  6
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "PreviewAny",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0",
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              null,
+              null,
+              null
+            ]
+          },
+          {
+            "id": 2,
+            "type": "RegexExtract",
+            "pos": [
+              -240,
+              8740
+            ],
+            "size": [
+              470,
+              460
+            ],
+            "flags": {},
+            "order": 1,
+            "mode": 0,
+            "showAdvanced": false,
+            "inputs": [
+              {
+                "localized_name": "string",
+                "name": "string",
+                "type": "STRING",
+                "widget": {
+                  "name": "string"
+                },
+                "link": 13
+              },
+              {
+                "localized_name": "regex_pattern",
+                "name": "regex_pattern",
+                "type": "STRING",
+                "widget": {
+                  "name": "regex_pattern"
+                },
+                "link": 9
+              },
+              {
+                "localized_name": "mode",
+                "name": "mode",
+                "type": "COMBO",
+                "widget": {
+                  "name": "mode"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "case_insensitive",
+                "name": "case_insensitive",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "case_insensitive"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "multiline",
+                "name": "multiline",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "multiline"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "dotall",
+                "name": "dotall",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "dotall"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "group_index",
+                "name": "group_index",
+                "type": "INT",
+                "widget": {
+                  "name": "group_index"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "STRING",
+                "name": "STRING",
+                "type": "STRING",
+                "links": [
+                  10
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "RegexExtract",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0",
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              "",
+              "",
+              "First Group",
+              false,
+              false,
+              false,
+              1
+            ]
+          },
+          {
+            "id": 3,
+            "type": "PrimitiveInt",
+            "pos": [
+              -810,
+              8400
+            ],
+            "size": [
+              270,
+              110
+            ],
+            "flags": {},
+            "order": 2,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "value",
+                "name": "value",
+                "type": "INT",
+                "widget": {
+                  "name": "value"
+                },
+                "link": 14
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "INT",
+                "name": "INT",
+                "type": "INT",
+                "links": [
+                  1
+                ]
+              }
+            ],
+            "title": "Int (line index)",
+            "properties": {
+              "Node name for S&R": "Int (line index)",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0",
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              0,
+              "fixed"
+            ]
+          },
+          {
+            "id": 8,
+            "type": "StringReplace",
+            "pos": [
+              -240,
+              8400
+            ],
+            "size": [
+              400,
+              280
+            ],
+            "flags": {},
+            "order": 3,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "string",
+                "name": "string",
+                "type": "STRING",
+                "widget": {
+                  "name": "string"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "find",
+                "name": "find",
+                "type": "STRING",
+                "widget": {
+                  "name": "find"
+                },
+                "link": null
+              },
+              {
+                "localized_name": "replace",
+                "name": "replace",
+                "type": "STRING",
+                "widget": {
+                  "name": "replace"
+                },
+                "link": 6
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "STRING",
+                "name": "STRING",
+                "type": "STRING",
+                "links": [
+                  9
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "StringReplace",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0",
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              "^(?:[^\\n]*\\n){index}([^\\n]*)(?:\\n|$)",
+              "index",
+              ""
+            ]
+          }
+        ],
+        "groups": [],
+        "links": [
+          {
+            "id": 1,
+            "origin_id": 3,
+            "origin_slot": 0,
+            "target_id": 1,
+            "target_slot": 0,
+            "type": "INT"
+          },
+          {
+            "id": 9,
+            "origin_id": 8,
+            "origin_slot": 0,
+            "target_id": 2,
+            "target_slot": 1,
+            "type": "STRING"
+          },
+          {
+            "id": 6,
+            "origin_id": 1,
+            "origin_slot": 0,
+            "target_id": 8,
+            "target_slot": 2,
+            "type": "STRING"
+          },
+          {
+            "id": 10,
+            "origin_id": 2,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 0,
+            "type": "STRING"
+          },
+          {
+            "id": 13,
+            "origin_id": -10,
+            "origin_slot": 0,
+            "target_id": 2,
+            "target_slot": 0,
+            "type": "STRING"
+          },
+          {
+            "id": 14,
+            "origin_id": -10,
+            "origin_slot": 1,
+            "target_id": 3,
+            "target_slot": 0,
+            "type": "INT"
+          }
+        ],
+        "extra": {},
+        "category": "Text Tools",
+        "description": "Selects one line from multiline text by zero-based index for batch or list-driven prompt workflows."
+      }
+    ]
+  },
+  "extra": {
+    "ue_links": [],
+    "links_added_by_ue": []
+  }
+}

blueprints/Split Image Grid to Tiles.json

@@ -0,0 +1,714 @@
+{
+  "revision": 0,
+  "last_node_id": 251,
+  "last_link_id": 0,
+  "nodes": [
+    {
+      "id": 251,
+      "type": "609e1fd1-b731-4b78-89ac-d19b1156b025",
+      "pos": [
+        -1490,
+        130
+      ],
+      "size": [
+        230,
+        164
+      ],
+      "flags": {},
+      "order": 1,
+      "mode": 0,
+      "inputs": [
+        {
+          "localized_name": "source_image",
+          "name": "source_image",
+          "type": "IMAGE",
+          "link": null
+        },
+        {
+          "localized_name": "columns",
+          "name": "columns",
+          "type": "INT",
+          "widget": {
+            "name": "columns"
+          },
+          "link": null
+        },
+        {
+          "localized_name": "rows",
+          "name": "rows",
+          "type": "INT",
+          "widget": {
+            "name": "rows"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "tiles",
+          "name": "tiles",
+          "type": "IMAGE",
+          "links": []
+        }
+      ],
+      "properties": {
+        "proxyWidgets": [
+          [
+            "228",
+            "value"
+          ],
+          [
+            "252",
+            "value"
+          ]
+        ],
+        "cnr_id": "comfy-core",
+        "ver": "0.20.1",
+        "enableTabs": false,
+        "tabWidth": 65,
+        "tabXOffset": 10,
+        "hasSecondTab": false,
+        "secondTabText": "Send Back",
+        "secondTabOffset": 80,
+        "secondTabWidth": 65
+      },
+      "widgets_values": [],
+      "title": "Split Image Grid to Tiles"
+    }
+  ],
+  "links": [],
+  "version": 0.4,
+  "definitions": {
+    "subgraphs": [
+      {
+        "id": "609e1fd1-b731-4b78-89ac-d19b1156b025",
+        "version": 1,
+        "state": {
+          "lastGroupId": 9,
+          "lastNodeId": 252,
+          "lastLinkId": 429,
+          "lastRerouteId": 0
+        },
+        "revision": 0,
+        "config": {},
+        "name": "Split Image Grid to Tiles",
+        "inputNode": {
+          "id": -10,
+          "bounding": [
+            -1690,
+            260,
+            128,
+            108
+          ]
+        },
+        "outputNode": {
+          "id": -20,
+          "bounding": [
+            -510,
+            590,
+            128,
+            68
+          ]
+        },
+        "inputs": [
+          {
+            "id": "866ac798-cfbc-450a-b755-e704f86404d9",
+            "name": "source_image",
+            "type": "IMAGE",
+            "linkIds": [
+              386,
+              389
+            ],
+            "localized_name": "source_image",
+            "pos": [
+              -1586,
+              284
+            ]
+          },
+          {
+            "id": "bc37b1f8-8ab2-4f19-bd00-75d4fbc4feb3",
+            "name": "columns",
+            "type": "INT",
+            "linkIds": [
+              427
+            ],
+            "localized_name": "columns",
+            "pos": [
+              -1586,
+              304
+            ]
+          },
+          {
+            "id": "d45915da-e848-43dd-9ccc-e3161e9c99d9",
+            "name": "rows",
+            "type": "INT",
+            "linkIds": [
+              428
+            ],
+            "localized_name": "rows",
+            "pos": [
+              -1586,
+              324
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "id": "18bc780f-064b-4038-87c6-67dba71deb08",
+            "name": "tiles",
+            "type": "IMAGE",
+            "linkIds": [
+              394
+            ],
+            "localized_name": "tiles",
+            "shape": 6,
+            "pos": [
+              -486,
+              614
+            ]
+          }
+        ],
+        "widgets": [],
+        "nodes": [
+          {
+            "id": 225,
+            "type": "SplitImageToTileList",
+            "pos": [
+              -1010,
+              620
+            ],
+            "size": [
+              290,
+              170
+            ],
+            "flags": {},
+            "order": 0,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "image",
+                "name": "image",
+                "type": "IMAGE",
+                "link": 386
+              },
+              {
+                "localized_name": "tile_width",
+                "name": "tile_width",
+                "type": "INT",
+                "widget": {
+                  "name": "tile_width"
+                },
+                "link": 403
+              },
+              {
+                "localized_name": "tile_height",
+                "name": "tile_height",
+                "type": "INT",
+                "widget": {
+                  "name": "tile_height"
+                },
+                "link": 404
+              },
+              {
+                "localized_name": "overlap",
+                "name": "overlap",
+                "type": "INT",
+                "widget": {
+                  "name": "overlap"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "IMAGE",
+                "name": "IMAGE",
+                "shape": 6,
+                "type": "IMAGE",
+                "links": [
+                  394
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "SplitImageToTileList",
+              "cnr_id": "comfy-core",
+              "ver": "0.20.1",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65
+            },
+            "widgets_values": [
+              1024,
+              1024,
+              0
+            ]
+          },
+          {
+            "id": 231,
+            "type": "ComfyMathExpression",
+            "pos": [
+              -1080,
+              330
+            ],
+            "size": [
+              370,
+              190
+            ],
+            "flags": {},
+            "order": 4,
+            "mode": 0,
+            "inputs": [
+              {
+                "label": "a",
+                "localized_name": "values.a",
+                "name": "values.a",
+                "type": "FLOAT,INT,BOOLEAN",
+                "link": 390
+              },
+              {
+                "label": "b",
+                "localized_name": "values.b",
+                "name": "values.b",
+                "shape": 7,
+                "type": "FLOAT,INT,BOOLEAN",
+                "link": 429
+              },
+              {
+                "label": "c",
+                "localized_name": "values.c",
+                "name": "values.c",
+                "shape": 7,
+                "type": "FLOAT,INT,BOOLEAN",
+                "link": null
+              },
+              {
+                "localized_name": "expression",
+                "name": "expression",
+                "type": "STRING",
+                "widget": {
+                  "name": "expression"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "FLOAT",
+                "name": "FLOAT",
+                "type": "FLOAT",
+                "links": null
+              },
+              {
+                "localized_name": "INT",
+                "name": "INT",
+                "type": "INT",
+                "links": [
+                  404
+                ]
+              },
+              {
+                "localized_name": "BOOL",
+                "name": "BOOL",
+                "type": "BOOLEAN",
+                "links": null
+              }
+            ],
+            "title": "Math Expression （Height）",
+            "properties": {
+              "Node name for S&R": "ComfyMathExpression",
+              "cnr_id": "comfy-core",
+              "ver": "0.18.1",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65,
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              "max(1, (int(a) + int(b) - 1) // int(b))"
+            ]
+          },
+          {
+            "id": 229,
+            "type": "ComfyMathExpression",
+            "pos": [
+              -1090,
+              -30
+            ],
+            "size": [
+              370,
+              190
+            ],
+            "flags": {},
+            "order": 2,
+            "mode": 0,
+            "inputs": [
+              {
+                "label": "a",
+                "localized_name": "values.a",
+                "name": "values.a",
+                "type": "FLOAT,INT,BOOLEAN",
+                "link": 387
+              },
+              {
+                "label": "b",
+                "localized_name": "values.b",
+                "name": "values.b",
+                "shape": 7,
+                "type": "FLOAT,INT,BOOLEAN",
+                "link": 388
+              },
+              {
+                "label": "c",
+                "localized_name": "values.c",
+                "name": "values.c",
+                "shape": 7,
+                "type": "FLOAT,INT,BOOLEAN",
+                "link": null
+              },
+              {
+                "localized_name": "expression",
+                "name": "expression",
+                "type": "STRING",
+                "widget": {
+                  "name": "expression"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "FLOAT",
+                "name": "FLOAT",
+                "type": "FLOAT",
+                "links": null
+              },
+              {
+                "localized_name": "INT",
+                "name": "INT",
+                "type": "INT",
+                "links": [
+                  403
+                ]
+              },
+              {
+                "localized_name": "BOOL",
+                "name": "BOOL",
+                "type": "BOOLEAN",
+                "links": null
+              }
+            ],
+            "title": "Math Expression （Width）",
+            "properties": {
+              "Node name for S&R": "ComfyMathExpression",
+              "cnr_id": "comfy-core",
+              "ver": "0.18.1",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65,
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              "max(1, (int(a) + int(b) - 1) // int(b))"
+            ]
+          },
+          {
+            "id": 228,
+            "type": "PrimitiveInt",
+            "pos": [
+              -1380,
+              90
+            ],
+            "size": [
+              230,
+              110
+            ],
+            "flags": {},
+            "order": 1,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "value",
+                "name": "value",
+                "type": "INT",
+                "widget": {
+                  "name": "value"
+                },
+                "link": 427
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "INT",
+                "name": "INT",
+                "type": "INT",
+                "links": [
+                  388
+                ]
+              }
+            ],
+            "title": "Int (grid columns)",
+            "properties": {
+              "Node name for S&R": "Int (grid columns)",
+              "cnr_id": "comfy-core",
+              "ver": "0.18.1",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65,
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              2,
+              "fixed"
+            ]
+          },
+          {
+            "id": 230,
+            "type": "GetImageSize",
+            "pos": [
+              -1380,
+              290
+            ],
+            "size": [
+              230,
+              100
+            ],
+            "flags": {},
+            "order": 3,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "image",
+                "name": "image",
+                "type": "IMAGE",
+                "link": 389
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "width",
+                "name": "width",
+                "type": "INT",
+                "links": [
+                  387
+                ]
+              },
+              {
+                "localized_name": "height",
+                "name": "height",
+                "type": "INT",
+                "links": [
+                  390
+                ]
+              },
+              {
+                "localized_name": "batch_size",
+                "name": "batch_size",
+                "type": "INT",
+                "links": null
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "GetImageSize",
+              "cnr_id": "comfy-core",
+              "ver": "0.18.1",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65,
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            }
+          },
+          {
+            "id": 252,
+            "type": "PrimitiveInt",
+            "pos": [
+              -1380,
+              470
+            ],
+            "size": [
+              230,
+              110
+            ],
+            "flags": {},
+            "order": 5,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "value",
+                "name": "value",
+                "type": "INT",
+                "widget": {
+                  "name": "value"
+                },
+                "link": 428
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "INT",
+                "name": "INT",
+                "type": "INT",
+                "links": [
+                  429
+                ]
+              }
+            ],
+            "title": "Int (grid rows)",
+            "properties": {
+              "Node name for S&R": "Int (grid rows)",
+              "cnr_id": "comfy-core",
+              "ver": "0.18.1",
+              "enableTabs": false,
+              "tabWidth": 65,
+              "tabXOffset": 10,
+              "hasSecondTab": false,
+              "secondTabText": "Send Back",
+              "secondTabOffset": 80,
+              "secondTabWidth": 65,
+              "ue_properties": {
+                "widget_ue_connectable": {},
+                "version": "7.7",
+                "input_ue_unconnectable": {}
+              }
+            },
+            "widgets_values": [
+              3,
+              "fixed"
+            ]
+          }
+        ],
+        "groups": [],
+        "links": [
+          {
+            "id": 403,
+            "origin_id": 229,
+            "origin_slot": 1,
+            "target_id": 225,
+            "target_slot": 1,
+            "type": "INT"
+          },
+          {
+            "id": 404,
+            "origin_id": 231,
+            "origin_slot": 1,
+            "target_id": 225,
+            "target_slot": 2,
+            "type": "INT"
+          },
+          {
+            "id": 390,
+            "origin_id": 230,
+            "origin_slot": 1,
+            "target_id": 231,
+            "target_slot": 0,
+            "type": "INT"
+          },
+          {
+            "id": 387,
+            "origin_id": 230,
+            "origin_slot": 0,
+            "target_id": 229,
+            "target_slot": 0,
+            "type": "INT"
+          },
+          {
+            "id": 388,
+            "origin_id": 228,
+            "origin_slot": 0,
+            "target_id": 229,
+            "target_slot": 1,
+            "type": "INT"
+          },
+          {
+            "id": 386,
+            "origin_id": -10,
+            "origin_slot": 0,
+            "target_id": 225,
+            "target_slot": 0,
+            "type": "IMAGE"
+          },
+          {
+            "id": 389,
+            "origin_id": -10,
+            "origin_slot": 0,
+            "target_id": 230,
+            "target_slot": 0,
+            "type": "IMAGE"
+          },
+          {
+            "id": 394,
+            "origin_id": 225,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 0,
+            "type": "IMAGE"
+          },
+          {
+            "id": 427,
+            "origin_id": -10,
+            "origin_slot": 1,
+            "target_id": 228,
+            "target_slot": 0,
+            "type": "INT"
+          },
+          {
+            "id": 428,
+            "origin_id": -10,
+            "origin_slot": 2,
+            "target_id": 252,
+            "target_slot": 0,
+            "type": "INT"
+          },
+          {
+            "id": 429,
+            "origin_id": 252,
+            "origin_slot": 0,
+            "target_id": 231,
+            "target_slot": 1,
+            "type": "INT"
+          }
+        ],
+        "extra": {},
+        "category": "Image Tools/Crop",
+        "description": "Splits an image into a configurable columns×rows grid of equal tiles for tiled generation or processing."
+      }
+    ]
+  },
+  "extra": {}
+}

blueprints/Video Captioning (Gemini).json

@@ -307,9 +307,9 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Text generation/Video Captioning",
+        "category": "Video Tools",
         "description": "Generates descriptive captions for video input using Google's Gemini multimodal LLM."
       }
     ]
   }
-}
+}

blueprints/Video Depth Estimation (Depth Anything 3).json

@@ -0,0 +1,825 @@
+{
+  "revision": 0,
+  "last_node_id": 97,
+  "last_link_id": 0,
+  "nodes": [
+    {
+      "id": 97,
+      "type": "253ec5ca-8333-4ddf-a036-9fc0923651b9",
+      "pos": [
+        410,
+        500
+      ],
+      "size": [
+        400,
+        400
+      ],
+      "flags": {},
+      "order": 3,
+      "mode": 0,
+      "inputs": [
+        {
+          "name": "video",
+          "type": "VIDEO",
+          "link": null
+        },
+        {
+          "name": "start_time",
+          "type": "FLOAT",
+          "widget": {
+            "name": "start_time"
+          },
+          "link": null
+        },
+        {
+          "name": "duration",
+          "type": "FLOAT",
+          "widget": {
+            "name": "duration"
+          },
+          "link": null
+        },
+        {
+          "name": "resolution",
+          "type": "INT",
+          "widget": {
+            "name": "resolution"
+          },
+          "link": null
+        },
+        {
+          "name": "resize_method",
+          "type": "COMBO",
+          "widget": {
+            "name": "resize_method"
+          },
+          "link": null
+        },
+        {
+          "label": "output_type",
+          "name": "output",
+          "type": "COMFY_DYNAMICCOMBO_V3",
+          "widget": {
+            "name": "output"
+          },
+          "link": null
+        },
+        {
+          "label": "normalization",
+          "name": "output.normalization",
+          "type": "COMBO",
+          "widget": {
+            "name": "output.normalization"
+          },
+          "link": null
+        },
+        {
+          "name": "output.apply_sky_clip",
+          "type": "BOOLEAN",
+          "widget": {
+            "name": "output.apply_sky_clip"
+          },
+          "link": null
+        },
+        {
+          "name": "model_name",
+          "type": "COMBO",
+          "widget": {
+            "name": "model_name"
+          },
+          "link": null
+        }
+      ],
+      "outputs": [
+        {
+          "localized_name": "IMAGE",
+          "name": "IMAGE",
+          "type": "IMAGE",
+          "links": []
+        },
+        {
+          "name": "audio",
+          "type": "AUDIO",
+          "links": []
+        },
+        {
+          "name": "fps",
+          "type": "FLOAT",
+          "links": []
+        }
+      ],
+      "properties": {
+        "proxyWidgets": [
+          [
+            "96",
+            "start_time"
+          ],
+          [
+            "96",
+            "duration"
+          ],
+          [
+            "93",
+            "resolution"
+          ],
+          [
+            "93",
+            "resize_method"
+          ],
+          [
+            "92",
+            "output"
+          ],
+          [
+            "92",
+            "output.normalization"
+          ],
+          [
+            "92",
+            "output.apply_sky_clip"
+          ],
+          [
+            "94",
+            "model_name"
+          ]
+        ],
+        "cnr_id": "comfy-core",
+        "ver": "0.24.0"
+      },
+      "widgets_values": [],
+      "title": "Video Depth Estimation (Depth Anything 3)"
+    }
+  ],
+  "links": [],
+  "version": 0.4,
+  "definitions": {
+    "subgraphs": [
+      {
+        "id": "253ec5ca-8333-4ddf-a036-9fc0923651b9",
+        "version": 1,
+        "state": {
+          "lastGroupId": 4,
+          "lastNodeId": 97,
+          "lastLinkId": 129,
+          "lastRerouteId": 0
+        },
+        "revision": 2,
+        "config": {},
+        "name": "Video Depth Estimation (Depth Anything 3)",
+        "inputNode": {
+          "id": -10,
+          "bounding": [
+            -230,
+            130,
+            167.912109375,
+            228
+          ]
+        },
+        "outputNode": {
+          "id": -20,
+          "bounding": [
+            1520,
+            140,
+            128,
+            108
+          ]
+        },
+        "inputs": [
+          {
+            "id": "698c28c6-cf92-4039-8b39-f3062868ea7c",
+            "name": "video",
+            "type": "VIDEO",
+            "linkIds": [
+              119
+            ],
+            "pos": [
+              -86.087890625,
+              154
+            ]
+          },
+          {
+            "id": "97a1f63e-1585-4a40-9dec-e2700120d84a",
+            "name": "start_time",
+            "type": "FLOAT",
+            "linkIds": [
+              121
+            ],
+            "pos": [
+              -86.087890625,
+              174
+            ]
+          },
+          {
+            "id": "4dbbd3b3-c5ee-4a56-a0d3-3268d3b2fd64",
+            "name": "duration",
+            "type": "FLOAT",
+            "linkIds": [
+              122
+            ],
+            "pos": [
+              -86.087890625,
+              194
+            ]
+          },
+          {
+            "id": "16f55101-f99d-4c0c-bebf-c3b31c54f13e",
+            "name": "resolution",
+            "type": "INT",
+            "linkIds": [
+              124
+            ],
+            "pos": [
+              -86.087890625,
+              214
+            ]
+          },
+          {
+            "id": "d9cd7693-4bb3-4ed7-9a75-276b997abcd9",
+            "name": "resize_method",
+            "type": "COMBO",
+            "linkIds": [
+              125
+            ],
+            "pos": [
+              -86.087890625,
+              234
+            ]
+          },
+          {
+            "id": "a6e90532-323b-462e-ba9c-1672384d5b31",
+            "name": "output",
+            "type": "COMFY_DYNAMICCOMBO_V3",
+            "linkIds": [
+              126
+            ],
+            "label": "output_type",
+            "pos": [
+              -86.087890625,
+              254
+            ]
+          },
+          {
+            "id": "69e6aeef-437d-4fde-b2fc-d5ab9369238d",
+            "name": "output.normalization",
+            "type": "COMBO",
+            "linkIds": [
+              127
+            ],
+            "label": "normalization",
+            "pos": [
+              -86.087890625,
+              274
+            ]
+          },
+          {
+            "id": "73206f72-f89a-4698-885e-5d9277df2998",
+            "name": "output.apply_sky_clip",
+            "type": "BOOLEAN",
+            "linkIds": [
+              128
+            ],
+            "pos": [
+              -86.087890625,
+              294
+            ]
+          },
+          {
+            "id": "dddbc7fc-9431-448a-9ed3-9aa62404288b",
+            "name": "model_name",
+            "type": "COMBO",
+            "linkIds": [
+              129
+            ],
+            "pos": [
+              -86.087890625,
+              314
+            ]
+          }
+        ],
+        "outputs": [
+          {
+            "id": "478ab537-63bc-4d74-a9f0-c975f550880f",
+            "name": "IMAGE",
+            "type": "IMAGE",
+            "linkIds": [
+              7
+            ],
+            "localized_name": "IMAGE",
+            "pos": [
+              1544,
+              164
+            ]
+          },
+          {
+            "id": "cdaf037e-79bc-4a94-b06c-0fd32e76f615",
+            "name": "audio",
+            "type": "AUDIO",
+            "linkIds": [
+              112
+            ],
+            "pos": [
+              1544,
+              184
+            ]
+          },
+          {
+            "id": "4c0e5484-d193-49c7-b107-92619628880a",
+            "name": "fps",
+            "type": "FLOAT",
+            "linkIds": [
+              113
+            ],
+            "pos": [
+              1544,
+              204
+            ]
+          }
+        ],
+        "widgets": [],
+        "nodes": [
+          {
+            "id": 92,
+            "type": "DA3Render",
+            "pos": [
+              740,
+              230
+            ],
+            "size": [
+              380,
+              130
+            ],
+            "flags": {},
+            "order": 0,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "da3_geometry",
+                "name": "da3_geometry",
+                "type": "DA3_GEOMETRY",
+                "link": 12
+              },
+              {
+                "localized_name": "output",
+                "name": "output",
+                "type": "COMFY_DYNAMICCOMBO_V3",
+                "widget": {
+                  "name": "output"
+                },
+                "link": 126
+              },
+              {
+                "localized_name": "output.normalization",
+                "name": "output.normalization",
+                "type": "COMBO",
+                "widget": {
+                  "name": "output.normalization"
+                },
+                "link": 127
+              },
+              {
+                "localized_name": "output.apply_sky_clip",
+                "name": "output.apply_sky_clip",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "output.apply_sky_clip"
+                },
+                "link": 128
+              },
+              {
+                "name": "geometry",
+                "type": "DA3_GEOMETRY",
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "IMAGE",
+                "name": "IMAGE",
+                "type": "IMAGE",
+                "slot_index": 0,
+                "links": [
+                  7
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "DA3Render",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0"
+            },
+            "widgets_values": [
+              "depth",
+              "v2_style",
+              false
+            ]
+          },
+          {
+            "id": 93,
+            "type": "DA3Inference",
+            "pos": [
+              740,
+              -30
+            ],
+            "size": [
+              390,
+              130
+            ],
+            "flags": {},
+            "order": 1,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "da3_model",
+                "name": "da3_model",
+                "type": "DA3_MODEL",
+                "link": 107
+              },
+              {
+                "localized_name": "image",
+                "name": "image",
+                "type": "IMAGE",
+                "link": 111
+              },
+              {
+                "localized_name": "resolution",
+                "name": "resolution",
+                "type": "INT",
+                "widget": {
+                  "name": "resolution"
+                },
+                "link": 124
+              },
+              {
+                "localized_name": "resize_method",
+                "name": "resize_method",
+                "type": "COMBO",
+                "widget": {
+                  "name": "resize_method"
+                },
+                "link": 125
+              },
+              {
+                "localized_name": "mode",
+                "name": "mode",
+                "type": "COMFY_DYNAMICCOMBO_V3",
+                "widget": {
+                  "name": "mode"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "da3_geometry",
+                "name": "da3_geometry",
+                "type": "DA3_GEOMETRY",
+                "slot_index": 0,
+                "links": [
+                  12
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "DA3Inference",
+              "cnr_id": "comfy-core",
+              "ver": "0.19.0"
+            },
+            "widgets_values": [
+              504,
+              "lower_bound_resize",
+              "mono"
+            ]
+          },
+          {
+            "id": 94,
+            "type": "LoadDA3Model",
+            "pos": [
+              50,
+              410
+            ],
+            "size": [
+              400,
+              140
+            ],
+            "flags": {},
+            "order": 2,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "model_name",
+                "name": "model_name",
+                "type": "COMBO",
+                "widget": {
+                  "name": "model_name"
+                },
+                "link": 129
+              },
+              {
+                "localized_name": "weight_dtype",
+                "name": "weight_dtype",
+                "type": "COMBO",
+                "widget": {
+                  "name": "weight_dtype"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "DA3_MODEL",
+                "name": "DA3_MODEL",
+                "type": "DA3_MODEL",
+                "links": [
+                  107
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "LoadDA3Model",
+              "cnr_id": "comfy-core",
+              "ver": "0.24.0",
+              "models": [
+                {
+                  "name": "depth_anything_3_mono_large.safetensors",
+                  "url": "https://huggingface.co/Comfy-Org/Depth-Anything-3/resolve/main/geometry_estimation/depth_anything_3_mono_large.safetensors",
+                  "directory": "geometry_estimation"
+                }
+              ]
+            },
+            "widgets_values": [
+              "depth_anything_3_mono_large.safetensors",
+              "default"
+            ]
+          },
+          {
+            "id": 95,
+            "type": "GetVideoComponents",
+            "pos": [
+              70,
+              -140
+            ],
+            "size": [
+              260,
+              120
+            ],
+            "flags": {},
+            "order": 3,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "video",
+                "name": "video",
+                "type": "VIDEO",
+                "link": 120
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "images",
+                "name": "images",
+                "type": "IMAGE",
+                "links": [
+                  111
+                ]
+              },
+              {
+                "localized_name": "audio",
+                "name": "audio",
+                "type": "AUDIO",
+                "links": [
+                  112
+                ]
+              },
+              {
+                "localized_name": "fps",
+                "name": "fps",
+                "type": "FLOAT",
+                "links": [
+                  113
+                ]
+              },
+              {
+                "localized_name": "bit_depth",
+                "name": "bit_depth",
+                "type": "INT",
+                "links": null
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "GetVideoComponents",
+              "cnr_id": "comfy-core",
+              "ver": "0.24.0"
+            }
+          },
+          {
+            "id": 96,
+            "type": "Video Slice",
+            "pos": [
+              70,
+              -360
+            ],
+            "size": [
+              270,
+              170
+            ],
+            "flags": {},
+            "order": 4,
+            "mode": 0,
+            "inputs": [
+              {
+                "localized_name": "video",
+                "name": "video",
+                "type": "VIDEO",
+                "link": 119
+              },
+              {
+                "localized_name": "start_time",
+                "name": "start_time",
+                "type": "FLOAT",
+                "widget": {
+                  "name": "start_time"
+                },
+                "link": 121
+              },
+              {
+                "localized_name": "duration",
+                "name": "duration",
+                "type": "FLOAT",
+                "widget": {
+                  "name": "duration"
+                },
+                "link": 122
+              },
+              {
+                "localized_name": "strict_duration",
+                "name": "strict_duration",
+                "type": "BOOLEAN",
+                "widget": {
+                  "name": "strict_duration"
+                },
+                "link": null
+              }
+            ],
+            "outputs": [
+              {
+                "localized_name": "VIDEO",
+                "name": "VIDEO",
+                "type": "VIDEO",
+                "links": [
+                  120
+                ]
+              }
+            ],
+            "properties": {
+              "Node name for S&R": "Video Slice",
+              "cnr_id": "comfy-core",
+              "ver": "0.24.0"
+            },
+            "widgets_values": [
+              0,
+              5,
+              false
+            ]
+          }
+        ],
+        "groups": [],
+        "links": [
+          {
+            "id": 12,
+            "origin_id": 93,
+            "origin_slot": 0,
+            "target_id": 92,
+            "target_slot": 0,
+            "type": "DA3_GEOMETRY"
+          },
+          {
+            "id": 7,
+            "origin_id": 92,
+            "origin_slot": 0,
+            "target_id": -20,
+            "target_slot": 0,
+            "type": "IMAGE"
+          },
+          {
+            "id": 107,
+            "origin_id": 94,
+            "origin_slot": 0,
+            "target_id": 93,
+            "target_slot": 0,
+            "type": "DA3_MODEL"
+          },
+          {
+            "id": 111,
+            "origin_id": 95,
+            "origin_slot": 0,
+            "target_id": 93,
+            "target_slot": 1,
+            "type": "IMAGE"
+          },
+          {
+            "id": 112,
+            "origin_id": 95,
+            "origin_slot": 1,
+            "target_id": -20,
+            "target_slot": 1,
+            "type": "AUDIO"
+          },
+          {
+            "id": 113,
+            "origin_id": 95,
+            "origin_slot": 2,
+            "target_id": -20,
+            "target_slot": 2,
+            "type": "FLOAT"
+          },
+          {
+            "id": 119,
+            "origin_id": -10,
+            "origin_slot": 0,
+            "target_id": 96,
+            "target_slot": 0,
+            "type": "VIDEO"
+          },
+          {
+            "id": 120,
+            "origin_id": 96,
+            "origin_slot": 0,
+            "target_id": 95,
+            "target_slot": 0,
+            "type": "VIDEO"
+          },
+          {
+            "id": 121,
+            "origin_id": -10,
+            "origin_slot": 1,
+            "target_id": 96,
+            "target_slot": 1,
+            "type": "FLOAT"
+          },
+          {
+            "id": 122,
+            "origin_id": -10,
+            "origin_slot": 2,
+            "target_id": 96,
+            "target_slot": 2,
+            "type": "FLOAT"
+          },
+          {
+            "id": 124,
+            "origin_id": -10,
+            "origin_slot": 3,
+            "target_id": 93,
+            "target_slot": 2,
+            "type": "INT"
+          },
+          {
+            "id": 125,
+            "origin_id": -10,
+            "origin_slot": 4,
+            "target_id": 93,
+            "target_slot": 3,
+            "type": "COMBO"
+          },
+          {
+            "id": 126,
+            "origin_id": -10,
+            "origin_slot": 5,
+            "target_id": 92,
+            "target_slot": 1,
+            "type": "COMFY_DYNAMICCOMBO_V3"
+          },
+          {
+            "id": 127,
+            "origin_id": -10,
+            "origin_slot": 6,
+            "target_id": 92,
+            "target_slot": 2,
+            "type": "COMBO"
+          },
+          {
+            "id": 128,
+            "origin_id": -10,
+            "origin_slot": 7,
+            "target_id": 92,
+            "target_slot": 3,
+            "type": "BOOLEAN"
+          },
+          {
+            "id": 129,
+            "origin_id": -10,
+            "origin_slot": 8,
+            "target_id": 94,
+            "target_slot": 0,
+            "type": "COMBO"
+          }
+        ],
+        "extra": {},
+        "category": "Conditioning & Preprocessors/Depth",
+        "description": "This subgraph processes a video input through Depth Anything 3 to produce temporally consistent depth maps for each frame, outputting a depth video. It is ideal for video content requiring spatial geometry estimation, such as 3D reconstruction, SLAM, or novel view synthesis from moving cameras. The model uses a plain transformer backbone trained with a depth-ray representation, supporting any number of views without requiring known camera poses."
+      }
+    ]
+  },
+  "extra": {
+    "BlueprintDescription": "This subgraph processes a video input through Depth Anything 3 to produce temporally consistent depth maps for each frame, outputting a depth video. It is ideal for video content requiring spatial geometry estimation, such as 3D reconstruction, SLAM, or novel view synthesis from moving cameras. The model uses a plain transformer backbone trained with a depth-ray representation, supporting any number of views without requiring known camera poses."
+  }
+}

blueprints/Video Segmentation (SAM3).json

@@ -818,7 +818,7 @@
           }
         ],
         "extra": {},
-        "category": "Video Tools",
+        "category": "Conditioning & Preprocessors/Segmentation & Mask",
         "description": "Segments video into temporally consistent masks using Meta SAM3 from text or interactive prompts."
       }
     ]

blueprints/Video Upscale(GAN x4).json

@@ -412,7 +412,7 @@
         "extra": {
           "workflowRendererVersion": "LG"
         },
-        "category": "Video generation and editing/Enhance video",
+        "category": "Video generation and editing/Upscale",
         "description": "Upscales video to 4× resolution using a GAN-based upscaling model."
       }
     ]

comfy/background_removal/birefnet.py

@@ -105,7 +105,7 @@ class WindowAttention(nn.Module):
 
         relative_position_bias = self.relative_position_bias_table[self.relative_position_index.long().view(-1)].view(
             self.window_size[0] * self.window_size[1], self.window_size[0] * self.window_size[1], -1)  # Wh*Ww,Wh*Ww,nH
-        relative_position_bias = relative_position_bias.permute(2, 0, 1).contiguous()  # nH, Wh*Ww, Wh*Ww
+        relative_position_bias = comfy.ops.cast_to_input(relative_position_bias.permute(2, 0, 1).contiguous(), attn)  # nH, Wh*Ww, Wh*Ww
         attn = attn + relative_position_bias.unsqueeze(0)
 
         if mask is not None:

comfy/bg_removal_model.py

@@ -55,12 +55,7 @@ class BackgroundRemovalModel():
         out = torch.nn.functional.interpolate(out, size=(H, W), mode="bicubic", antialias=False)
 
         mask = out.sigmoid().to(device=comfy.model_management.intermediate_device(), dtype=comfy.model_management.intermediate_dtype())
-        if mask.ndim == 3:
-            mask = mask.unsqueeze(0)
-        if mask.shape[1] != 1:
-            mask = mask.movedim(-1, 1)
-
-        return mask
+        return mask.squeeze(1)  # (B, 1, H, W) -> (B, H, W)
 
 
 def load_background_removal_model(sd):

comfy/cli_args.py

@@ -49,7 +49,7 @@ parser.add_argument("--temp-directory", type=str, default=None, help="Set the Co
 parser.add_argument("--input-directory", type=str, default=None, help="Set the ComfyUI input directory. Overrides --base-directory.")
 parser.add_argument("--auto-launch", action="store_true", help="Automatically launch ComfyUI in the default browser.")
 parser.add_argument("--disable-auto-launch", action="store_true", help="Disable auto launching the browser.")
-parser.add_argument("--cuda-device", type=int, default=None, metavar="DEVICE_ID", help="Set the id of the cuda device this instance will use. All other devices will not be visible.")
+parser.add_argument("--cuda-device", type=str, default=None, metavar="DEVICE_ID", help="Set the ids of cuda devices this instance will use, as a comma-separated list (e.g. '0' or '0,1'). All other devices will not be visible.")
 parser.add_argument("--default-device", type=int, default=None, metavar="DEFAULT_DEVICE_ID", help="Set the id of the default device, all other devices will stay visible.")
 cm_group = parser.add_mutually_exclusive_group()
 cm_group.add_argument("--cuda-malloc", action="store_true", help="Enable cudaMallocAsync (enabled by default for torch 2.0 and up).")
@@ -110,13 +110,12 @@ parser.add_argument("--preview-method", type=LatentPreviewMethod, default=Latent
 
 parser.add_argument("--preview-size", type=int, default=512, help="Sets the maximum preview size for sampler nodes.")
 
-CACHE_RAM_AUTO_GB = -1.0
-
 cache_group = parser.add_mutually_exclusive_group()
+cache_group.add_argument("--cache-ram", nargs='*', type=float, default=[], metavar="GB", help="Use RAM pressure caching with the specified headroom thresholds. This is the default caching mode. The first value sets the active-cache threshold; the optional second value sets the inactive-cache/pin threshold. Defaults when no values are provided: active 10%% of system RAM (min 2GB, max 10GB), inactive 100%% of system RAM (max 96GB).")
 cache_group.add_argument("--cache-classic", action="store_true", help="Use the old style (aggressive) caching.")
 cache_group.add_argument("--cache-lru", type=int, default=0, help="Use LRU caching with a maximum of N node results cached. May use more RAM/VRAM.")
 cache_group.add_argument("--cache-none", action="store_true", help="Reduced RAM/VRAM usage at the expense of executing every node for each run.")
-cache_group.add_argument("--cache-ram", nargs='?', const=CACHE_RAM_AUTO_GB, type=float, default=0, help="Use RAM pressure caching with the specified headroom threshold. If available RAM drops below the threshold the cache removes large items to free RAM. Default (when no value is provided): 25%% of system RAM (min 4GB, max 32GB).")
+cache_group.add_argument("--high-ram", action="store_true", help="Can improve performance slightly on high RAM or on systems where pagefile use is preferred over model loading.")
 
 attn_group = parser.add_mutually_exclusive_group()
 attn_group.add_argument("--use-split-cross-attention", action="store_true", help="Use the split cross attention optimization. Ignored when xformers is used.")
@@ -135,7 +134,7 @@ upcast.add_argument("--dont-upcast-attention", action="store_true", help="Disabl
 parser.add_argument("--enable-manager", action="store_true", help="Enable the ComfyUI-Manager feature.")
 manager_group = parser.add_mutually_exclusive_group()
 manager_group.add_argument("--disable-manager-ui", action="store_true", help="Disables only the ComfyUI-Manager UI and endpoints. Scheduled installations and similar background tasks will still operate.")
-manager_group.add_argument("--enable-manager-legacy-ui", action="store_true", help="Enables the legacy UI of ComfyUI-Manager")
+manager_group.add_argument("--enable-manager-legacy-ui", action="store_true", help="Enables the legacy UI of ComfyUI-Manager. Implies --enable-manager.")
 
 
 vram_group = parser.add_mutually_exclusive_group()
@@ -146,11 +145,13 @@ vram_group.add_argument("--novram", action="store_true", help="When lowvram isn'
 vram_group.add_argument("--cpu", action="store_true", help="To use the CPU for everything (slow).")
 
 parser.add_argument("--reserve-vram", type=float, default=None, help="Set the amount of vram in GB you want to reserve for use by your OS/other software. By default some amount is reserved depending on your OS.")
+parser.add_argument("--vram-headroom", type=float, default=0, help="Set the amount of vram in GB for DynamicVRAM to maintain as extra headroom above default. ComfyUI will try and keep this much VRAM completely free and unused, even counting VRAM from other apps.")
 
 parser.add_argument("--async-offload", nargs='?', const=2, type=int, default=None, metavar="NUM_STREAMS", help="Use async weight offloading. An optional argument controls the amount of offload streams. Default is 2. Enabled by default on Nvidia.")
 parser.add_argument("--disable-async-offload", action="store_true", help="Disable async weight offloading.")
 parser.add_argument("--disable-dynamic-vram", action="store_true", help="Disable dynamic VRAM and use estimate based model loading.")
 parser.add_argument("--enable-dynamic-vram", action="store_true", help="Enable dynamic VRAM on systems where it's not enabled by default.")
+parser.add_argument("--fast-disk", action="store_true", help="Prefer disk-backed dynamic loading and offload over unpinned RAM. Can be faster for users with fast NVME disks.")
 
 parser.add_argument("--force-non-blocking", action="store_true", help="Force ComfyUI to use non-blocking operations for all applicable tensors. This may improve performance on some non-Nvidia systems but can cause issues with some workflows.")
 
@@ -167,6 +168,8 @@ class PerformanceFeature(enum.Enum):
 
 parser.add_argument("--fast", nargs="*", type=PerformanceFeature, help="Enable some untested and potentially quality deteriorating optimizations. This is used to test new features so using it might crash your comfyui. --fast with no arguments enables everything. You can pass a list specific optimizations if you only want to enable specific ones. Current valid optimizations: {}".format(" ".join(map(lambda c: c.value, PerformanceFeature))))
 
+parser.add_argument("--debug-hang", action="store_true", help="Enable stack trace dumps on Ctrl-C for debugging hangs.")
+
 parser.add_argument("--disable-pinned-memory", action="store_true", help="Disable pinned memory use.")
 
 parser.add_argument("--mmap-torch-files", action="store_true", help="Use mmap when loading ckpt/pt files.")
@@ -237,6 +240,7 @@ database_default_path = os.path.abspath(
 )
 parser.add_argument("--database-url", type=str, default=f"sqlite:///{database_default_path}", help="Specify the database URL, e.g. for an in-memory database you can use 'sqlite:///:memory:'.")
 parser.add_argument("--enable-assets", action="store_true", help="Enable the assets system (API routes, database synchronization, and background scanning).")
+parser.add_argument("--enable-asset-hashing", action="store_true", help="Compute blake3 content hashes when scanning assets. Hashing enables future asset-portability features (deduplication, cross-machine model resolution) but adds startup cost and per-output cost on large models directories. Off by default; enable to opt in.")
 parser.add_argument("--feature-flag", type=str, action='append', default=[], metavar="KEY[=VALUE]", help="Set a server feature flag. Use KEY=VALUE to set an explicit value, or bare KEY to set it to true. Can be specified multiple times. Boolean values (true/false) and numbers are auto-converted. Examples: --feature-flag show_signin_button=true  or  --feature-flag show_signin_button")
 parser.add_argument("--list-feature-flags", action="store_true", help="Print the registry of known CLI-settable feature flags as JSON and exit.")
 
@@ -245,6 +249,12 @@ if comfy.options.args_parsing:
 else:
     args = parser.parse_args([])
 
+if args.cache_ram is not None and len(args.cache_ram) > 2:
+    parser.error("--cache-ram accepts at most two values: active GB and inactive GB")
+
+if args.high_ram:
+    args.cache_classic = True
+
 if args.windows_standalone_build:
     args.auto_launch = True
 
@@ -254,6 +264,10 @@ if args.disable_auto_launch:
 if args.force_fp16:
     args.fp16_unet = True
 
+# '--enable-manager-legacy-ui' is meaningless unless the manager is enabled, so imply '--enable-manager'.
+if args.enable_manager_legacy_ui:
+    args.enable_manager = True
+
 
 # '--fast' is not provided, use an empty set
 if args.fast is None:

comfy/clip_vision.py

@@ -9,6 +9,7 @@ import comfy.model_management
 import comfy.utils
 import comfy.clip_model
 import comfy.image_encoders.dino2
+import comfy.image_encoders.dino3
 
 class Output:
     def __getitem__(self, key):
@@ -23,12 +24,16 @@ IMAGE_ENCODERS = {
     "siglip_vision_model": comfy.clip_model.CLIPVisionModelProjection,
     "siglip2_vision_model": comfy.clip_model.CLIPVisionModelProjection,
     "dinov2": comfy.image_encoders.dino2.Dinov2Model,
+    "dinov3": comfy.image_encoders.dino3.DINOv3ViTModel,
 }
 
 class ClipVisionModel():
     def __init__(self, json_config):
-        with open(json_config) as f:
-            config = json.load(f)
+        if isinstance(json_config, dict):
+            config = json_config
+        else:
+            with open(json_config) as f:
+                config = json.load(f)
 
         self.image_size = config.get("image_size", 224)
         self.image_mean = config.get("image_mean", [0.48145466, 0.4578275, 0.40821073])
@@ -134,6 +139,8 @@ def load_clipvision_from_sd(sd, prefix="", convert_keys=False):
         json_config = os.path.join(os.path.join(os.path.dirname(os.path.realpath(__file__)), "image_encoders"), "dino2_giant.json")
     elif 'encoder.layer.23.layer_scale2.lambda1' in sd:
         json_config = os.path.join(os.path.join(os.path.dirname(os.path.realpath(__file__)), "image_encoders"), "dino2_large.json")
+    elif 'layer.0.mlp.gate_proj.weight' in sd and 'layer.31.norm1.weight' in sd: # Dinov3 ViT-H/16+ (SwiGLU gated MLP, 32 layers)
+        json_config = comfy.image_encoders.dino3.DINOV3_VITH_CONFIG
     else:
         return None
 

comfy/comfy_types/node_typing.py

@@ -1,6 +1,5 @@
 """Comfy-specific type hinting"""
 
-from __future__ import annotations
 from typing import Literal, TypedDict, Optional
 from typing_extensions import NotRequired
 from abc import ABC, abstractmethod

comfy/context_windows.py

@@ -8,6 +8,8 @@ from abc import ABC, abstractmethod
 import logging
 import comfy.model_management
 import comfy.patcher_extension
+import comfy.utils
+import comfy.conds
 if TYPE_CHECKING:
     from comfy.model_base import BaseModel
     from comfy.model_patcher import ModelPatcher
@@ -51,12 +53,18 @@ class ContextHandlerABC(ABC):
 
 
 class IndexListContextWindow(ContextWindowABC):
-    def __init__(self, index_list: list[int], dim: int=0, total_frames: int=0):
+    def __init__(self, index_list: list[int], dim: int=0, total_frames: int=0, modality_windows: dict=None, context_overlap: int=0):
         self.index_list = index_list
         self.context_length = len(index_list)
+        self.context_overlap = context_overlap
         self.dim = dim
         self.total_frames = total_frames
         self.center_ratio = (min(index_list) + max(index_list)) / (2 * total_frames)
+        self.modality_windows = modality_windows  # dict of {mod_idx: IndexListContextWindow}
+        self.guide_frames_indices: list[int] = []
+        self.guide_overlap_info: list[tuple[int, int]] = []
+        self.guide_kf_local_positions: list[int] = []
+        self.guide_downscale_factors: list[int] = []
 
     def get_tensor(self, full: torch.Tensor, device=None, dim=None, retain_index_list=[]) -> torch.Tensor:
         if dim is None:
@@ -85,6 +93,11 @@ class IndexListContextWindow(ContextWindowABC):
         region_idx = int(self.center_ratio * num_regions)
         return min(max(region_idx, 0), num_regions - 1)
 
+    def get_window_for_modality(self, modality_idx: int) -> 'IndexListContextWindow':
+        if modality_idx == 0:
+            return self
+        return self.modality_windows[modality_idx]
+
 
 class IndexListCallbacks:
     EVALUATE_CONTEXT_WINDOWS = "evaluate_context_windows"
@@ -148,6 +161,172 @@ def slice_cond(cond_value, window: IndexListContextWindow, x_in: torch.Tensor, d
     return cond_value._copy_with(sliced)
 
 
+def compute_guide_overlap(guide_entries: list[dict], keyframe_idxs: torch.Tensor, temporal_downscale_ratio: int, window_index_list: list[int]):
+    """Compute which concatenated guide frames overlap with a context window.
+
+    Each guide's latent-space start is derived from its first token's pixel-t-start
+    in keyframe_idxs (shape (B, [t,h,w], num_tokens, [start, end])), divided by the
+    model's temporal_downscale_ratio.
+
+    Args:
+        guide_entries: list of guide_attention_entry dicts
+        keyframe_idxs: per-token pixel coords cond tensor for the modality
+        temporal_downscale_ratio: model's pixel-to-latent temporal compression ratio
+        window_index_list: the window's frame indices into the video portion
+
+    Returns:
+        suffix_indices: indices into the guide_frames tensor for frame selection
+        overlap_info: list of (entry_idx, overlap_count) for guide_attention_entries adjustment
+        kf_local_positions: window-local frame positions for keyframe_idxs regeneration
+        total_overlap: total number of overlapping guide frames
+    """
+    window_set = set(window_index_list)
+    window_list = list(window_index_list)
+    suffix_indices = []
+    overlap_info = []
+    kf_local_positions = []
+    suffix_base = 0
+    token_offset = 0
+
+    for entry_idx, entry in enumerate(guide_entries):
+        first_t_pixel = int(keyframe_idxs[0, 0, token_offset, 0].item())
+        latent_start = (first_t_pixel + temporal_downscale_ratio - 1) // temporal_downscale_ratio
+        guide_len = entry["latent_shape"][0]
+        entry_overlap = 0
+
+        for local_offset in range(guide_len):
+            video_pos = latent_start + local_offset
+            if video_pos in window_set:
+                suffix_indices.append(suffix_base + local_offset)
+                kf_local_positions.append(window_list.index(video_pos))
+                entry_overlap += 1
+
+        if entry_overlap > 0:
+            overlap_info.append((entry_idx, entry_overlap))
+        suffix_base += guide_len
+        token_offset += entry["pre_filter_count"]
+
+    return suffix_indices, overlap_info, kf_local_positions, len(suffix_indices)
+
+
+@dataclass
+class WindowingState:
+    """Per-modality context windowing state for each step,
+    built using IndexListContextHandler._build_window_state().
+    For non-multimodal models the lists are length 1
+    """
+    latents: list[torch.Tensor]                  # per-modality working latents (guide frames stripped)
+    guide_latents: list[torch.Tensor | None]     # per-modality guide frames stripped from latents
+    guide_entries: list[list[dict] | None]       # per-modality guide_attention_entry metadata
+    keyframe_idxs: list[torch.Tensor | None]     # per-modality keyframe_idxs tensor for guide latent_start derivation
+    latent_shapes: list | None                   # original packed shapes for unpack/pack (None if not multimodal)
+    dim: int = 0                                 # primary modality temporal dim for context windowing
+    is_multimodal: bool = False
+    temporal_downscale_ratio: int = 1            # model's pixel-to-latent temporal compression ratio
+
+    def prepare_window(self, window: IndexListContextWindow, model) -> IndexListContextWindow:
+        """Reformat window for multimodal contexts by deriving per-modality index lists.
+        Non-multimodal contexts return the input window unchanged.
+        """
+        if not self.is_multimodal:
+            return window
+
+        x = self.latents[0]
+        primary_total = self.latent_shapes[0][self.dim]
+        primary_overlap = window.context_overlap
+        map_shapes = self.latent_shapes
+        if x.size(self.dim) != primary_total:
+            map_shapes = list(self.latent_shapes)
+            video_shape = list(self.latent_shapes[0])
+            video_shape[self.dim] = x.size(self.dim)
+            map_shapes[0] = torch.Size(video_shape)
+        try:
+            per_modality_indices = model.map_context_window_to_modalities(
+                window.index_list, map_shapes, self.dim)
+        except AttributeError:
+            raise NotImplementedError(
+                f"{type(model).__name__} must implement map_context_window_to_modalities for multimodal context windows.")
+        modality_windows = {}
+        for mod_idx in range(1, len(self.latents)):
+            modality_total_frames = self.latents[mod_idx].shape[self.dim]
+            ratio = modality_total_frames / primary_total if primary_total > 0 else 1
+            modality_overlap = max(round(primary_overlap * ratio), 0)
+            modality_windows[mod_idx] = IndexListContextWindow(
+                per_modality_indices[mod_idx], dim=self.dim,
+                total_frames=modality_total_frames,
+                context_overlap=modality_overlap)
+        return IndexListContextWindow(
+            window.index_list, dim=self.dim, total_frames=x.shape[self.dim],
+            modality_windows=modality_windows, context_overlap=primary_overlap)
+
+    def slice_for_window(self, window: IndexListContextWindow, retain_index_list: list[int], device=None) -> tuple[list[torch.Tensor], list[int]]:
+        """Slice latents for a context window, injecting guide frames where applicable.
+        For multimodal contexts, uses the modality-specific windows derived in prepare_window().
+        """
+        sliced = []
+        guide_frame_counts = []
+        for idx in range(len(self.latents)):
+            modality_window = window.get_window_for_modality(idx)
+            retain = retain_index_list if idx == 0 else []
+            s = modality_window.get_tensor(self.latents[idx], device, retain_index_list=retain)
+            if self.guide_entries[idx] is not None:
+                s, ng = self._inject_guide_frames(s, modality_window, modality_idx=idx)
+            else:
+                ng = 0
+            sliced.append(s)
+            guide_frame_counts.append(ng)
+        return sliced, guide_frame_counts
+
+    def strip_guide_frames(self, out_per_modality: list[list[torch.Tensor]], guide_frame_counts: list[int], window: IndexListContextWindow):
+        """Strip injected guide frames from per-cond, per-modality outputs in place."""
+        for idx in range(len(self.latents)):
+            if guide_frame_counts[idx] > 0:
+                window_len = len(window.get_window_for_modality(idx).index_list)
+                for ci in range(len(out_per_modality)):
+                    out_per_modality[ci][idx] = out_per_modality[ci][idx].narrow(self.dim, 0, window_len)
+
+    def _inject_guide_frames(self, latent_slice: torch.Tensor, window: IndexListContextWindow, modality_idx: int = 0) -> tuple[torch.Tensor, int]:
+        guide_entries = self.guide_entries[modality_idx]
+        guide_frames = self.guide_latents[modality_idx]
+        keyframe_idxs = self.keyframe_idxs[modality_idx]
+        suffix_idx, overlap_info, kf_local_pos, guide_frame_count = compute_guide_overlap(
+            guide_entries, keyframe_idxs, self.temporal_downscale_ratio, window.index_list)
+        # Shift keyframe positions to account for causal_window_fix anchor occupying sub-pos 0.
+        anchor_idx = getattr(window, 'causal_anchor_index', None)
+        if anchor_idx is not None and anchor_idx >= 0:
+            kf_local_pos = [p + 1 for p in kf_local_pos]
+        window.guide_frames_indices = suffix_idx
+        window.guide_overlap_info = overlap_info
+        window.guide_kf_local_positions = kf_local_pos
+
+        # Derive per-overlap-entry latent_downscale_factor from guide entry latent_shape vs guide frame spatial dims.
+        # guide_frames has full (post-dilation) spatial dims; entry["latent_shape"] has pre-dilation dims.
+        guide_downscale_factors = []
+        if guide_frame_count > 0:
+            full_H = guide_frames.shape[3]
+            for entry_idx, _ in overlap_info:
+                entry_H = guide_entries[entry_idx]["latent_shape"][1]
+                guide_downscale_factors.append(full_H // entry_H)
+        window.guide_downscale_factors = guide_downscale_factors
+
+        if guide_frame_count > 0:
+            idx = tuple([slice(None)] * self.dim + [suffix_idx])
+            return torch.cat([latent_slice, guide_frames[idx]], dim=self.dim), guide_frame_count
+        return latent_slice, 0
+
+    def patch_latent_shapes(self, sub_conds, new_shapes):
+        if not self.is_multimodal:
+            return
+
+        for cond_list in sub_conds:
+            if cond_list is None:
+                continue
+            for cond_dict in cond_list:
+                model_conds = cond_dict.get('model_conds', {})
+                if 'latent_shapes' in model_conds:
+                    model_conds['latent_shapes'] = comfy.conds.CONDConstant(new_shapes)
+
+
 @dataclass
 class ContextSchedule:
     name: str
@@ -162,7 +341,7 @@ ContextResults = collections.namedtuple("ContextResults", ['window_idx', 'sub_co
 class IndexListContextHandler(ContextHandlerABC):
     def __init__(self, context_schedule: ContextSchedule, fuse_method: ContextFuseMethod, context_length: int=1, context_overlap: int=0, context_stride: int=1,
                  closed_loop: bool=False, dim:int=0, freenoise: bool=False, cond_retain_index_list: list[int]=[], split_conds_to_windows: bool=False,
-                 causal_window_fix: bool=True):
+                 latent_retain_index_list: list[int]=[], causal_window_fix: bool=True):
         self.context_schedule = context_schedule
         self.fuse_method = fuse_method
         self.context_length = context_length
@@ -174,17 +353,118 @@ class IndexListContextHandler(ContextHandlerABC):
         self.freenoise = freenoise
         self.cond_retain_index_list = [int(x.strip()) for x in cond_retain_index_list.split(",")] if cond_retain_index_list else []
         self.split_conds_to_windows = split_conds_to_windows
+        self.latent_retain_index_list = [int(x.strip()) for x in latent_retain_index_list.split(",")] if latent_retain_index_list else []
         self.causal_window_fix = causal_window_fix
 
         self.callbacks = {}
 
+    @staticmethod
+    def _get_latent_shapes(conds):
+        for cond_list in conds:
+            if cond_list is None:
+                continue
+            for cond_dict in cond_list:
+                model_conds = cond_dict.get('model_conds', {})
+                if 'latent_shapes' in model_conds:
+                    return model_conds['latent_shapes'].cond
+        return None
+
+    @staticmethod
+    def _get_guide_entries(conds):
+        for cond_list in conds:
+            if cond_list is None:
+                continue
+            for cond_dict in cond_list:
+                model_conds = cond_dict.get('model_conds', {})
+                entries = model_conds.get('guide_attention_entries')
+                if entries is not None and hasattr(entries, 'cond') and entries.cond:
+                    return entries.cond
+        return None
+
+    @staticmethod
+    def _get_keyframe_idxs(conds):
+        for cond_list in conds:
+            if cond_list is None:
+                continue
+            for cond_dict in cond_list:
+                model_conds = cond_dict.get('model_conds', {})
+                kf = model_conds.get('keyframe_idxs')
+                if kf is not None and hasattr(kf, 'cond') and kf.cond is not None:
+                    return kf.cond
+        return None
+
+    def _apply_freenoise(self, noise: torch.Tensor, conds: list[list[dict]], seed: int) -> torch.Tensor:
+        """Apply FreeNoise shuffling, scaling context length/overlap per-modality by frame ratio.
+        If guide frames are present on the primary modality, only the video portion is shuffled.
+        """
+        guide_entries = self._get_guide_entries(conds)
+        guide_count = sum(e["latent_shape"][0] for e in guide_entries) if guide_entries else 0
+
+        latent_shapes = self._get_latent_shapes(conds)
+        if latent_shapes is not None and len(latent_shapes) > 1:
+            modalities = comfy.utils.unpack_latents(noise, latent_shapes)
+            primary_total = latent_shapes[0][self.dim]
+            primary_video_count = modalities[0].size(self.dim) - guide_count
+            apply_freenoise(modalities[0].narrow(self.dim, 0, primary_video_count), self.dim, self.context_length, self.context_overlap, seed)
+            for i in range(1, len(modalities)):
+                mod_total = latent_shapes[i][self.dim]
+                ratio = mod_total / primary_total if primary_total > 0 else 1
+                mod_ctx_len = max(round(self.context_length * ratio), 1)
+                mod_ctx_overlap = max(round(self.context_overlap * ratio), 0)
+                modalities[i] = apply_freenoise(modalities[i], self.dim, mod_ctx_len, mod_ctx_overlap, seed)
+            noise, _ = comfy.utils.pack_latents(modalities)
+            return noise
+        video_count = noise.size(self.dim) - guide_count
+        apply_freenoise(noise.narrow(self.dim, 0, video_count), self.dim, self.context_length, self.context_overlap, seed)
+        return noise
+
+    def _build_window_state(self, x_in: torch.Tensor, conds: list[list[dict]], model: BaseModel) -> WindowingState:
+        """Build windowing state for the current step, including unpacking latents and extracting guide frame info from conds."""
+        latent_shapes = self._get_latent_shapes(conds)
+        is_multimodal = latent_shapes is not None and len(latent_shapes) > 1
+        unpacked_latents = comfy.utils.unpack_latents(x_in, latent_shapes) if is_multimodal else [x_in]
+
+        unpacked_latents_list = list(unpacked_latents)
+        guide_latents_list = [None] * len(unpacked_latents)
+        guide_entries_list = [None] * len(unpacked_latents)
+        keyframe_idxs_list = [None] * len(unpacked_latents)
+
+        extracted_guide_entries = self._get_guide_entries(conds)
+        extracted_keyframe_idxs = self._get_keyframe_idxs(conds)
+
+        # Strip guide frames (only from first modality for now)
+        if extracted_guide_entries is not None:
+            guide_count = sum(e["latent_shape"][0] for e in extracted_guide_entries)
+            if guide_count > 0:
+                x = unpacked_latents[0]
+                latent_count = x.size(self.dim) - guide_count
+                unpacked_latents_list[0] = x.narrow(self.dim, 0, latent_count)
+                guide_latents_list[0] = x.narrow(self.dim, latent_count, guide_count)
+                guide_entries_list[0] = extracted_guide_entries
+                keyframe_idxs_list[0] = extracted_keyframe_idxs
+
+
+        return WindowingState(
+            latents=unpacked_latents_list,
+            guide_latents=guide_latents_list,
+            guide_entries=guide_entries_list,
+            keyframe_idxs=keyframe_idxs_list,
+            latent_shapes=latent_shapes,
+            dim=self.dim,
+            is_multimodal=is_multimodal,
+            temporal_downscale_ratio=model.latent_format.temporal_downscale_ratio)
+
     def should_use_context(self, model: BaseModel, conds: list[list[dict]], x_in: torch.Tensor, timestep: torch.Tensor, model_options: dict[str]) -> bool:
-        # for now, assume first dim is batch - should have stored on BaseModel in actual implementation
-        if x_in.size(self.dim) > self.context_length:
-            logging.info(f"Using context windows {self.context_length} with overlap {self.context_overlap} for {x_in.size(self.dim)} frames.")
+        window_state = self._build_window_state(x_in, conds, model) # build window_state to check frame counts, will be built again in execute
+        total_frame_count = window_state.latents[0].size(self.dim)
+        if total_frame_count > self.context_length:
+            logging.info(f"\nUsing context windows: Context length {self.context_length} with overlap {self.context_overlap} for {total_frame_count} frames.")
             if self.cond_retain_index_list:
                 logging.info(f"Retaining original cond for indexes: {self.cond_retain_index_list}")
+            if self.latent_retain_index_list:
+                logging.info(f"Retaining original latent for indexes: {self.latent_retain_index_list}")
             return True
+        logging.info(f"\nNot using context windows since context length ({self.context_length}) exceeds input frames ({total_frame_count}).")
         return False
 
     def prepare_control_objects(self, control: ControlBase, device=None) -> ControlBase:
@@ -275,7 +555,9 @@ class IndexListContextHandler(ContextHandlerABC):
         return resized_cond
 
     def set_step(self, timestep: torch.Tensor, model_options: dict[str]):
-        mask = torch.isclose(model_options["transformer_options"]["sample_sigmas"], timestep[0], rtol=0.0001)
+        sample_sigmas = model_options["transformer_options"]["sample_sigmas"]
+        current_timestep = timestep[0].to(sample_sigmas.dtype)
+        mask = torch.isclose(sample_sigmas, current_timestep, rtol=0.0001)
         matches = torch.nonzero(mask)
         if torch.numel(matches) == 0:
             return  # substep from multi-step sampler: keep self._step from the last full step
@@ -284,54 +566,98 @@ class IndexListContextHandler(ContextHandlerABC):
     def get_context_windows(self, model: BaseModel, x_in: torch.Tensor, model_options: dict[str]) -> list[IndexListContextWindow]:
         full_length = x_in.size(self.dim) # TODO: choose dim based on model
         context_windows = self.context_schedule.func(full_length, self, model_options)
-        context_windows = [IndexListContextWindow(window, dim=self.dim, total_frames=full_length) for window in context_windows]
+        context_windows = [IndexListContextWindow(window, dim=self.dim, total_frames=full_length, context_overlap=self.context_overlap) for window in context_windows]
         return context_windows
 
     def execute(self, calc_cond_batch: Callable, model: BaseModel, conds: list[list[dict]], x_in: torch.Tensor, timestep: torch.Tensor, model_options: dict[str]):
         self._model = model
         self.set_step(timestep, model_options)
-        context_windows = self.get_context_windows(model, x_in, model_options)
-        enumerated_context_windows = list(enumerate(context_windows))
 
-        conds_final = [torch.zeros_like(x_in) for _ in conds]
+        window_state = self._build_window_state(x_in, conds, model)
+        num_modalities = len(window_state.latents)
+
+        context_windows = self.get_context_windows(model, window_state.latents[0], model_options)
+        enumerated_context_windows = list(enumerate(context_windows))
+        total_windows = len(enumerated_context_windows)
+
+        # Initialize per-modality accumulators (length 1 for single-modality)
+        accum = [[torch.zeros_like(m) for _ in conds] for m in window_state.latents]
         if self.fuse_method.name == ContextFuseMethods.RELATIVE:
-            counts_final = [torch.ones(get_shape_for_dim(x_in, self.dim), device=x_in.device) for _ in conds]
+            counts = [[torch.ones(get_shape_for_dim(m, self.dim), device=m.device) for _ in conds] for m in window_state.latents]
         else:
-            counts_final = [torch.zeros(get_shape_for_dim(x_in, self.dim), device=x_in.device) for _ in conds]
-        biases_final = [([0.0] * x_in.shape[self.dim]) for _ in conds]
+            counts = [[torch.zeros(get_shape_for_dim(m, self.dim), device=m.device) for _ in conds] for m in window_state.latents]
+        biases = [[([0.0] * m.shape[self.dim]) for _ in conds] for m in window_state.latents]
 
         for callback in comfy.patcher_extension.get_all_callbacks(IndexListCallbacks.EXECUTE_START, self.callbacks):
             callback(self, model, x_in, conds, timestep, model_options)
 
+        # accumulate results from each context window
         for enum_window in enumerated_context_windows:
-            results = self.evaluate_context_windows(calc_cond_batch, model, x_in, conds, timestep, [enum_window], model_options)
+            results = self.evaluate_context_windows(
+                calc_cond_batch, model, x_in, conds, timestep, [enum_window],
+                model_options, window_state=window_state, total_windows=total_windows)
             for result in results:
-                self.combine_context_window_results(x_in, result.sub_conds_out, result.sub_conds, result.window, result.window_idx, len(enumerated_context_windows), timestep,
-                                            conds_final, counts_final, biases_final)
+                # result.sub_conds_out is per-cond, per-modality: list[list[Tensor]]
+                for mod_idx in range(num_modalities):
+                    mod_out = [result.sub_conds_out[ci][mod_idx] for ci in range(len(conds))]
+                    modality_window = result.window.get_window_for_modality(mod_idx)
+                    self.combine_context_window_results(
+                        window_state.latents[mod_idx], mod_out, result.sub_conds, modality_window,
+                        result.window_idx, total_windows, timestep,
+                        accum[mod_idx], counts[mod_idx], biases[mod_idx])
+
+        # fuse accumulated results into final conds
         try:
-            # finalize conds
-            if self.fuse_method.name == ContextFuseMethods.RELATIVE:
-                # relative is already normalized, so return as is
-                del counts_final
-                return conds_final
-            else:
-                # normalize conds via division by context usage counts
-                for i in range(len(conds_final)):
-                    conds_final[i] /= counts_final[i]
-                del counts_final
-                return conds_final
+            result_out = []
+            for ci in range(len(conds)):
+                finalized = []
+                for mod_idx in range(num_modalities):
+                    if self.fuse_method.name != ContextFuseMethods.RELATIVE:
+                        accum[mod_idx][ci] /= counts[mod_idx][ci]
+                    f = accum[mod_idx][ci]
+
+                    # if guide frames were injected, append them to the end of the fused latents for the next step
+                    if window_state.guide_latents[mod_idx] is not None:
+                        f = torch.cat([f, window_state.guide_latents[mod_idx]], dim=self.dim)
+                    finalized.append(f)
+
+                # pack modalities together if needed
+                if window_state.is_multimodal and len(finalized) > 1:
+                    packed, _ = comfy.utils.pack_latents(finalized)
+                else:
+                    packed = finalized[0]
+
+                result_out.append(packed)
+            return result_out
         finally:
             for callback in comfy.patcher_extension.get_all_callbacks(IndexListCallbacks.EXECUTE_CLEANUP, self.callbacks):
                 callback(self, model, x_in, conds, timestep, model_options)
 
-    def evaluate_context_windows(self, calc_cond_batch: Callable, model: BaseModel, x_in: torch.Tensor, conds, timestep: torch.Tensor, enumerated_context_windows: list[tuple[int, IndexListContextWindow]],
-                                model_options, device=None, first_device=None):
+    def evaluate_context_windows(self, calc_cond_batch: Callable, model: BaseModel, x_in: torch.Tensor, conds,
+                                timestep: torch.Tensor, enumerated_context_windows: list[tuple[int, IndexListContextWindow]],
+                                model_options, window_state: WindowingState, total_windows: int = None,
+                                device=None, first_device=None):
+        """Evaluate context windows and return per-cond, per-modality outputs in ContextResults.sub_conds_out
+
+        For each window:
+        1. Builds windows (for each modality if multimodal)
+        2. Slices window for each modality
+        3. Injects concatenated latent guide frames where present
+        4. Packs together if needed and calls model
+        5. Unpacks and strips any guides from outputs
+        """
+        x = window_state.latents[0]
+
         results: list[ContextResults] = []
         for window_idx, window in enumerated_context_windows:
             # allow processing to end between context window executions for faster Cancel
             comfy.model_management.throw_exception_if_processing_interrupted()
 
-            # causal_window_fix: prepend a pre-window frame that will be stripped post-forward
+            # prepare the window accounting for multimodal windows
+            window = window_state.prepare_window(window, model)
+
+            # causal_window_fix: prepend a pre-window frame that will be stripped post-forward.
+            # Set anchor before slice_for_window so the latent slice and downstream cond slices both pick it up.
             anchor_applied = False
             if self.causal_window_fix:
                 anchor_idx = window.index_list[0] - 1
@@ -339,27 +665,46 @@ class IndexListContextHandler(ContextHandlerABC):
                     window.causal_anchor_index = anchor_idx
                     anchor_applied = True
 
+            # slice the window for each modality, injecting guide frames where applicable
+            sliced, guide_frame_counts_per_modality = window_state.slice_for_window(window, self.latent_retain_index_list, device)
+
             for callback in comfy.patcher_extension.get_all_callbacks(IndexListCallbacks.EVALUATE_CONTEXT_WINDOWS, self.callbacks):
                 callback(self, model, x_in, conds, timestep, model_options, window_idx, window, model_options, device, first_device)
 
-            # update exposed params
+            logging.info(f"Context window {window_idx + 1}/{total_windows or len(enumerated_context_windows)}: frames {window.index_list[0]}-{window.index_list[-1]} of {x.shape[self.dim]}"
+                         + (f" (+{guide_frame_counts_per_modality[0]} guide frames)" if guide_frame_counts_per_modality[0] > 0 else "")
+                         )
+
+            # if multimodal, pack modalities together
+            if window_state.is_multimodal and len(sliced) > 1:
+                sub_x, sub_shapes = comfy.utils.pack_latents(sliced)
+            else:
+                sub_x, sub_shapes = sliced[0], [sliced[0].shape]
+
+            # get resized conds for window
             model_options["transformer_options"]["context_window"] = window
-            # get subsections of x, timestep, conds
-            sub_x = window.get_tensor(x_in, device)
-            sub_timestep = window.get_tensor(timestep, device, dim=0)
-            sub_conds = [self.get_resized_cond(cond, x_in, window, device) for cond in conds]
+            sub_timestep = window.get_tensor(timestep, dim=0)
+            sub_conds = [self.get_resized_cond(cond, x, window) for cond in conds]
 
+            # if multimodal, patch latent_shapes in conds for correct unpacking in model
+            window_state.patch_latent_shapes(sub_conds, sub_shapes)
+
+            # call model on window
             sub_conds_out = calc_cond_batch(model, sub_conds, sub_x, sub_timestep, model_options)
-            if device is not None:
-                for i in range(len(sub_conds_out)):
-                    sub_conds_out[i] = sub_conds_out[i].to(x_in.device)
 
-            # strip causal_window_fix anchor if applied
+            # unpack outputs
+            out_per_modality = [comfy.utils.unpack_latents(sub_conds_out[i], sub_shapes) for i in range(len(sub_conds_out))]
+
+            # strip causal_window_fix anchor from primary modality before guide strip so window_len math stays correct
             if anchor_applied:
-                for i in range(len(sub_conds_out)):
-                    sub_conds_out[i] = sub_conds_out[i].narrow(self.dim, 1, sub_conds_out[i].shape[self.dim] - 1)
+                for ci in range(len(out_per_modality)):
+                    t = out_per_modality[ci][0]
+                    out_per_modality[ci][0] = t.narrow(self.dim, 1, t.shape[self.dim] - 1)
 
-            results.append(ContextResults(window_idx, sub_conds_out, sub_conds, window))
+            # strip injected guide frames
+            window_state.strip_guide_frames(out_per_modality, guide_frame_counts_per_modality, window)
+
+            results.append(ContextResults(window_idx, out_per_modality, sub_conds, window))
         return results
 
 
@@ -383,7 +728,7 @@ class IndexListContextHandler(ContextHandlerABC):
                     biases_final[i][idx] = bias_total + bias
         else:
             # add conds and counts based on weights of fuse method
-            weights = get_context_weights(window.context_length, x_in.shape[self.dim], window.index_list, self, sigma=timestep)
+            weights = get_context_weights(window.context_length, x_in.shape[self.dim], window.index_list, self, sigma=timestep, context_overlap=window.context_overlap)
             weights_tensor = match_weights_to_dim(weights, x_in, self.dim, device=x_in.device)
             for i in range(len(sub_conds_out)):
                 window.add_window(conds_final[i], sub_conds_out[i] * weights_tensor)
@@ -393,16 +738,22 @@ class IndexListContextHandler(ContextHandlerABC):
             callback(self, x_in, sub_conds_out, sub_conds, window, window_idx, total_windows, timestep, conds_final, counts_final, biases_final)
 
 
-def _prepare_sampling_wrapper(executor, model, noise_shape: torch.Tensor, *args, **kwargs):
-    # limit noise_shape length to context_length for more accurate vram use estimation
+def _prepare_sampling_wrapper(executor, model, noise_shape: torch.Tensor, conds, *args, **kwargs):
+    # Scale noise_shape to a single context window so VRAM estimation budgets per-window.
     model_options = kwargs.get("model_options", None)
     if model_options is None:
         raise Exception("model_options not found in prepare_sampling_wrapper; this should never happen, something went wrong.")
     handler: IndexListContextHandler = model_options.get("context_handler", None)
     if handler is not None:
         noise_shape = list(noise_shape)
-        noise_shape[handler.dim] = min(noise_shape[handler.dim], handler.context_length)
-    return executor(model, noise_shape, *args, **kwargs)
+        is_packed = len(noise_shape) == 3 and noise_shape[1] == 1
+        if is_packed:
+            # TODO: latent_shapes cond isn't attached yet at this point, so we can't compute a
+            # per-window flat latent here. Skipping the clamp over-estimates but prevents immediate OOM.
+            pass
+        elif handler.dim < len(noise_shape) and noise_shape[handler.dim] > handler.context_length:
+            noise_shape[handler.dim] = min(noise_shape[handler.dim], handler.context_length)
+    return executor(model, noise_shape, conds, *args, **kwargs)
 
 
 def create_prepare_sampling_wrapper(model: ModelPatcher):
@@ -422,11 +773,12 @@ def _sampler_sample_wrapper(executor, guider, sigmas, extra_args, callback, nois
         raise Exception("context_handler not found in sampler_sample_wrapper; this should never happen, something went wrong.")
     if not handler.freenoise:
         return executor(guider, sigmas, extra_args, callback, noise, *args, **kwargs)
-    noise = apply_freenoise(noise, handler.dim, handler.context_length, handler.context_overlap, extra_args["seed"])
+
+    conds = [guider.conds.get('positive', guider.conds.get('negative', []))]
+    noise = handler._apply_freenoise(noise, conds, extra_args["seed"])
 
     return executor(guider, sigmas, extra_args, callback, noise, *args, **kwargs)
 
-
 def create_sampler_sample_wrapper(model: ModelPatcher):
     model.add_wrapper_with_key(
         comfy.patcher_extension.WrappersMP.SAMPLER_SAMPLE,
@@ -434,7 +786,6 @@ def create_sampler_sample_wrapper(model: ModelPatcher):
         _sampler_sample_wrapper
     )
 
-
 def match_weights_to_dim(weights: list[float], x_in: torch.Tensor, dim: int, device=None) -> torch.Tensor:
     total_dims = len(x_in.shape)
     weights_tensor = torch.Tensor(weights).to(device=device)
@@ -580,8 +931,9 @@ def get_matching_context_schedule(context_schedule: str) -> ContextSchedule:
     return ContextSchedule(context_schedule, func)
 
 
-def get_context_weights(length: int, full_length: int, idxs: list[int], handler: IndexListContextHandler, sigma: torch.Tensor=None):
-    return handler.fuse_method.func(length, sigma=sigma, handler=handler, full_length=full_length, idxs=idxs)
+def get_context_weights(length: int, full_length: int, idxs: list[int], handler: IndexListContextHandler, sigma: torch.Tensor=None, context_overlap: int=None):
+    context_overlap = handler.context_overlap if context_overlap is None else context_overlap
+    return handler.fuse_method.func(length, sigma=sigma, handler=handler, full_length=full_length, idxs=idxs, context_overlap=context_overlap)
 
 
 def create_weights_flat(length: int, **kwargs) -> list[float]:
@@ -599,18 +951,18 @@ def create_weights_pyramid(length: int, **kwargs) -> list[float]:
         weight_sequence = list(range(1, max_weight, 1)) + [max_weight] + list(range(max_weight - 1, 0, -1))
     return weight_sequence
 
-def create_weights_overlap_linear(length: int, full_length: int, idxs: list[int], handler: IndexListContextHandler, **kwargs):
+def create_weights_overlap_linear(length: int, full_length: int, idxs: list[int], context_overlap: int, **kwargs):
     # based on code in Kijai's WanVideoWrapper: https://github.com/kijai/ComfyUI-WanVideoWrapper/blob/dbb2523b37e4ccdf45127e5ae33e31362f755c8e/nodes.py#L1302
     # only expected overlap is given different weights
     weights_torch = torch.ones((length))
     # blend left-side on all except first window
     if min(idxs) > 0:
-        ramp_up = torch.linspace(1e-37, 1, handler.context_overlap)
-        weights_torch[:handler.context_overlap] = ramp_up
+        ramp_up = torch.linspace(1e-37, 1, context_overlap)
+        weights_torch[:context_overlap] = ramp_up
     # blend right-side on all except last window
     if max(idxs) < full_length-1:
-        ramp_down = torch.linspace(1, 1e-37, handler.context_overlap)
-        weights_torch[-handler.context_overlap:] = ramp_down
+        ramp_down = torch.linspace(1, 1e-37, context_overlap)
+        weights_torch[-context_overlap:] = ramp_down
     return weights_torch
 
 class ContextFuseMethods:

comfy/controlnet.py

@@ -15,13 +15,14 @@
     You should have received a copy of the GNU General Public License
     along with this program.  If not, see <https://www.gnu.org/licenses/>.
 """
-
+from __future__ import annotations
 
 import torch
 from enum import Enum
 import math
 import os
 import logging
+import copy
 import comfy.utils
 import comfy.model_management
 import comfy.model_detection
@@ -38,7 +39,7 @@ import comfy.ldm.hydit.controlnet
 import comfy.ldm.flux.controlnet
 import comfy.ldm.qwen_image.controlnet
 import comfy.cldm.dit_embedder
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Union
 if TYPE_CHECKING:
     from comfy.hooks import HookGroup
 
@@ -64,6 +65,18 @@ class StrengthType(Enum):
     CONSTANT = 1
     LINEAR_UP = 2
 
+class ControlIsolation:
+    '''Temporarily set a ControlBase object's previous_controlnet to None to prevent cascading calls.'''
+    def __init__(self, control: ControlBase):
+        self.control = control
+        self.orig_previous_controlnet = control.previous_controlnet
+
+    def __enter__(self):
+        self.control.previous_controlnet = None
+
+    def __exit__(self, *args):
+        self.control.previous_controlnet = self.orig_previous_controlnet
+
 class ControlBase:
     def __init__(self):
         self.cond_hint_original = None
@@ -77,7 +90,7 @@ class ControlBase:
         self.compression_ratio = 8
         self.upscale_algorithm = 'nearest-exact'
         self.extra_args = {}
-        self.previous_controlnet = None
+        self.previous_controlnet: Union[ControlBase, None] = None
         self.extra_conds = []
         self.strength_type = StrengthType.CONSTANT
         self.concat_mask = False
@@ -85,6 +98,7 @@ class ControlBase:
         self.extra_concat = None
         self.extra_hooks: HookGroup = None
         self.preprocess_image = lambda a: a
+        self.multigpu_clones: dict[torch.device, ControlBase] = {}
 
     def set_cond_hint(self, cond_hint, strength=1.0, timestep_percent_range=(0.0, 1.0), vae=None, extra_concat=[]):
         self.cond_hint_original = cond_hint
@@ -111,17 +125,38 @@ class ControlBase:
     def cleanup(self):
         if self.previous_controlnet is not None:
             self.previous_controlnet.cleanup()
-
+        for device_cnet in self.multigpu_clones.values():
+            with ControlIsolation(device_cnet):
+                device_cnet.cleanup()
         self.cond_hint = None
         self.extra_concat = None
         self.timestep_range = None
 
     def get_models(self):
         out = []
+        for device_cnet in self.multigpu_clones.values():
+            out += device_cnet.get_models_only_self()
         if self.previous_controlnet is not None:
             out += self.previous_controlnet.get_models()
         return out
 
+    def get_models_only_self(self):
+        'Calls get_models, but temporarily sets previous_controlnet to None.'
+        with ControlIsolation(self):
+            return self.get_models()
+
+    def get_instance_for_device(self, device):
+        'Returns instance of this Control object intended for selected device.'
+        return self.multigpu_clones.get(device, self)
+
+    def deepclone_multigpu(self, load_device, autoregister=False):
+        '''
+        Create deep clone of Control object where model(s) is set to other devices.
+
+        When autoregister is set to True, the deep clone is also added to multigpu_clones dict.
+        '''
+        raise NotImplementedError("Classes inheriting from ControlBase should define their own deepclone_multigpu funtion.")
+
     def get_extra_hooks(self):
         out = []
         if self.extra_hooks is not None:
@@ -130,7 +165,7 @@ class ControlBase:
             out += self.previous_controlnet.get_extra_hooks()
         return out
 
-    def copy_to(self, c):
+    def copy_to(self, c: ControlBase):
         c.cond_hint_original = self.cond_hint_original
         c.strength = self.strength
         c.timestep_percent_range = self.timestep_percent_range
@@ -284,6 +319,14 @@ class ControlNet(ControlBase):
         self.copy_to(c)
         return c
 
+    def deepclone_multigpu(self, load_device, autoregister=False):
+        c = self.copy()
+        c.control_model = copy.deepcopy(c.control_model)
+        c.control_model_wrapped = comfy.model_patcher.ModelPatcher(c.control_model, load_device=load_device, offload_device=comfy.model_management.unet_offload_device())
+        if autoregister:
+            self.multigpu_clones[load_device] = c
+        return c
+
     def get_models(self):
         out = super().get_models()
         out.append(self.control_model_wrapped)
@@ -314,6 +357,10 @@ class QwenFunControlNet(ControlNet):
         super().pre_run(model, percent_to_timestep_function)
         self.set_extra_arg("base_model", model.diffusion_model)
 
+    def cleanup(self):
+        self.extra_args.pop("base_model", None)
+        super().cleanup()
+
     def copy(self):
         c = QwenFunControlNet(None, global_average_pooling=self.global_average_pooling, load_device=self.load_device, manual_cast_dtype=self.manual_cast_dtype)
         c.control_model = self.control_model
@@ -906,6 +953,14 @@ class T2IAdapter(ControlBase):
         self.copy_to(c)
         return c
 
+    def deepclone_multigpu(self, load_device, autoregister=False):
+        c = self.copy()
+        c.t2i_model = copy.deepcopy(c.t2i_model)
+        c.device = load_device
+        if autoregister:
+            self.multigpu_clones[load_device] = c
+        return c
+
 def load_t2i_adapter(t2i_data, model_options={}): #TODO: model_options
     compression_ratio = 8
     upscale_algorithm = 'nearest-exact'

comfy/float.py

@@ -1,5 +1,20 @@
+import logging
+
 import torch
 
+_CK_STOCHASTIC_ROUNDING_AVAILABLE = False
+try:
+    import comfy_kitchen as ck
+    _ck_stochastic_rounding_fp8 = ck.stochastic_rounding_fp8
+    _CK_STOCHASTIC_ROUNDING_AVAILABLE = True
+except (AttributeError, ImportError):
+    logging.warning("comfy_kitchen does not support stochastic FP8 rounding, please update comfy_kitchen.")
+
+if not _CK_STOCHASTIC_ROUNDING_AVAILABLE:
+    def _ck_stochastic_rounding_fp8(value, rng, dtype):
+        raise NotImplementedError("comfy_kitchen does not support stochastic FP8 rounding")
+
+
 def calc_mantissa(abs_x, exponent, normal_mask, MANTISSA_BITS, EXPONENT_BIAS, generator=None):
     mantissa_scaled = torch.where(
         normal_mask,
@@ -57,6 +72,10 @@ def stochastic_rounding(value, dtype, seed=0):
     if dtype == torch.float8_e4m3fn or dtype == torch.float8_e5m2:
         generator = torch.Generator(device=value.device)
         generator.manual_seed(seed)
+        if _CK_STOCHASTIC_ROUNDING_AVAILABLE:
+            rng = torch.randint(0, 256, value.size(), dtype=torch.uint8, layout=value.layout, device=value.device, generator=generator)
+            return _ck_stochastic_rounding_fp8(value, rng, dtype)
+
         output = torch.empty_like(value, dtype=dtype)
         num_slices = max(1, (value.numel() / (4096 * 4096)))
         slice_size = max(1, round(value.shape[0] / num_slices))

comfy/image_encoders/dino2.py

@@ -1,7 +1,13 @@
 import torch
+import torch.nn.functional as F
+
 from comfy.text_encoders.bert import BertAttention
 import comfy.model_management
 from comfy.ldm.modules.attention import optimized_attention_for_device
+from comfy.ldm.depth_anything_3.reference_view_selector import (
+    select_reference_view, reorder_by_reference, restore_original_order,
+    THRESH_FOR_REF_SELECTION,
+)
 
 
 class Dino2AttentionOutput(torch.nn.Module):
@@ -14,13 +20,41 @@ class Dino2AttentionOutput(torch.nn.Module):
 
 
 class Dino2AttentionBlock(torch.nn.Module):
-    def __init__(self, embed_dim, heads, layer_norm_eps, dtype, device, operations):
+    def __init__(self, embed_dim, heads, layer_norm_eps, dtype, device, operations,
+                 qk_norm=False):
         super().__init__()
+        self.heads = heads
+        self.head_dim = embed_dim // heads
         self.attention = BertAttention(embed_dim, heads, dtype, device, operations)
         self.output = Dino2AttentionOutput(embed_dim, embed_dim, layer_norm_eps, dtype, device, operations)
+        if qk_norm:
+            self.q_norm = operations.LayerNorm(self.head_dim, dtype=dtype, device=device)
+            self.k_norm = operations.LayerNorm(self.head_dim, dtype=dtype, device=device)
+        else:
+            self.q_norm = None
+            self.k_norm = None
 
-    def forward(self, x, mask, optimized_attention):
-        return self.output(self.attention(x, mask, optimized_attention))
+    def forward(self, x, mask, optimized_attention, pos=None, rope=None):
+        # Fast path used by the existing CLIP-vision DINOv2 (no DA3 extensions).
+        if self.q_norm is None and rope is None:
+            return self.output(self.attention(x, mask, optimized_attention))
+
+        # DA3 path: do QKV manually so we can apply per-head QK-norm and 2D RoPE.
+        attn = self.attention
+        B, N, C = x.shape
+        h = self.heads
+        d = self.head_dim
+        q = attn.query(x).view(B, N, h, d).transpose(1, 2)
+        k = attn.key(x).view(B, N, h, d).transpose(1, 2)
+        v = attn.value(x).view(B, N, h, d).transpose(1, 2)
+        if self.q_norm is not None:
+            q = self.q_norm(q)
+            k = self.k_norm(k)
+        if rope is not None and pos is not None:
+            q = rope(q, pos)
+            k = rope(k, pos)
+        out = optimized_attention(q, k, v, h, mask=mask, skip_reshape=True)
+        return self.output(out)
 
 
 class LayerScale(torch.nn.Module):
@@ -64,9 +98,11 @@ class SwiGLUFFN(torch.nn.Module):
 
 
 class Dino2Block(torch.nn.Module):
-    def __init__(self, dim, num_heads, layer_norm_eps, dtype, device, operations, use_swiglu_ffn):
+    def __init__(self, dim, num_heads, layer_norm_eps, dtype, device, operations, use_swiglu_ffn,
+                 qk_norm=False):
         super().__init__()
-        self.attention = Dino2AttentionBlock(dim, num_heads, layer_norm_eps, dtype, device, operations)
+        self.attention = Dino2AttentionBlock(dim, num_heads, layer_norm_eps, dtype, device, operations,
+                                             qk_norm=qk_norm)
         self.layer_scale1 = LayerScale(dim, dtype, device, operations)
         self.layer_scale2 = LayerScale(dim, dtype, device, operations)
         if use_swiglu_ffn:
@@ -76,19 +112,90 @@ class Dino2Block(torch.nn.Module):
         self.norm1 = operations.LayerNorm(dim, eps=layer_norm_eps, dtype=dtype, device=device)
         self.norm2 = operations.LayerNorm(dim, eps=layer_norm_eps, dtype=dtype, device=device)
 
-    def forward(self, x, optimized_attention):
-        x = x + self.layer_scale1(self.attention(self.norm1(x), None, optimized_attention))
+    def forward(self, x, optimized_attention, pos=None, rope=None, attn_mask=None):
+        x = x + self.layer_scale1(self.attention(self.norm1(x), attn_mask, optimized_attention,
+                                                 pos=pos, rope=rope))
         x = x + self.layer_scale2(self.mlp(self.norm2(x)))
         return x
 
 
-class Dino2Encoder(torch.nn.Module):
-    def __init__(self, dim, num_heads, layer_norm_eps, num_layers, dtype, device, operations, use_swiglu_ffn):
+# -----------------------------------------------------------------------------
+# 2D Rotary position embedding (DA3 extension)
+# -----------------------------------------------------------------------------
+
+
+class _PositionGetter:
+    """Cache (h, w) -> flat (y, x) position grid used to feed ``rope``."""
+
+    def __init__(self):
+        self._cache: dict = {}
+
+    def __call__(self, batch_size: int, height: int, width: int, device) -> torch.Tensor:
+        key = (height, width, device)
+        if key not in self._cache:
+            y = torch.arange(height, device=device)
+            x = torch.arange(width, device=device)
+            self._cache[key] = torch.cartesian_prod(y, x)
+        cached = self._cache[key]
+        return cached.view(1, height * width, 2).expand(batch_size, -1, -1).clone()
+
+
+class RotaryPositionEmbedding2D(torch.nn.Module):
+    """2D RoPE used by DA3-Small/Base. No learnable parameters."""
+
+    def __init__(self, frequency: float = 100.0):
         super().__init__()
-        self.layer = torch.nn.ModuleList([Dino2Block(dim, num_heads, layer_norm_eps, dtype, device, operations, use_swiglu_ffn = use_swiglu_ffn)
-                                          for _ in range(num_layers)])
+        self.base_frequency = frequency
+        self._freq_cache: dict = {}
+
+    def _components(self, dim: int, seq_len: int, device, dtype):
+        key = (dim, seq_len, device, dtype)
+        if key not in self._freq_cache:
+            exp = torch.arange(0, dim, 2, device=device).float() / dim
+            inv_freq = 1.0 / (self.base_frequency ** exp)
+            pos = torch.arange(seq_len, device=device, dtype=inv_freq.dtype)
+            ang = torch.einsum("i,j->ij", pos, inv_freq)
+            ang = ang.to(dtype)
+            ang = torch.cat((ang, ang), dim=-1)
+            self._freq_cache[key] = (ang.cos().to(dtype), ang.sin().to(dtype))
+        return self._freq_cache[key]
+
+    @staticmethod
+    def _rotate(x: torch.Tensor) -> torch.Tensor:
+        d = x.shape[-1]
+        x1, x2 = x[..., : d // 2], x[..., d // 2:]
+        return torch.cat((-x2, x1), dim=-1)
+
+    def _apply_1d(self, tokens, positions, cos_c, sin_c):
+        cos = F.embedding(positions, cos_c)[:, None, :, :]
+        sin = F.embedding(positions, sin_c)[:, None, :, :]
+        return (tokens * cos) + (self._rotate(tokens) * sin)
+
+    def forward(self, tokens: torch.Tensor, positions: torch.Tensor) -> torch.Tensor:
+        feature_dim = tokens.size(-1) // 2
+        max_pos = int(positions.max()) + 1
+        cos_c, sin_c = self._components(feature_dim, max_pos, tokens.device, tokens.dtype)
+        v, h = tokens.chunk(2, dim=-1)
+        v = self._apply_1d(v, positions[..., 0], cos_c, sin_c)
+        h = self._apply_1d(h, positions[..., 1], cos_c, sin_c)
+        return torch.cat((v, h), dim=-1)
+
+
+class Dino2Encoder(torch.nn.Module):
+    def __init__(self, dim, num_heads, layer_norm_eps, num_layers, dtype, device, operations, use_swiglu_ffn,
+                 qknorm_start: int = -1):
+        super().__init__()
+        self.layer = torch.nn.ModuleList([
+            Dino2Block(
+                dim, num_heads, layer_norm_eps, dtype, device, operations,
+                use_swiglu_ffn=use_swiglu_ffn,
+                qk_norm=(qknorm_start != -1 and i >= qknorm_start),
+            )
+            for i in range(num_layers)
+        ])
 
     def forward(self, x, intermediate_output=None):
+        # Backward-compat path used by ``ClipVisionModel`` (no DA3 extensions).
         optimized_attention = optimized_attention_for_device(x.device, False, small_input=True)
 
         if intermediate_output is not None:
@@ -122,16 +229,27 @@ class Dino2PatchEmbeddings(torch.nn.Module):
 
 
 class Dino2Embeddings(torch.nn.Module):
-    def __init__(self, dim, dtype, device, operations):
+    def __init__(self, dim, dtype, device, operations,
+                 patch_size: int = 14, image_size: int = 518,
+                 use_mask_token: bool = True,
+                 num_camera_tokens: int = 0):
         super().__init__()
-        patch_size = 14
-        image_size = 518
         self.patch_size = patch_size
+        self.image_size = image_size
 
         self.patch_embeddings = Dino2PatchEmbeddings(dim, patch_size=patch_size, image_size=image_size, dtype=dtype, device=device, operations=operations)
         self.position_embeddings = torch.nn.Parameter(torch.empty(1, (image_size // patch_size) ** 2 + 1, dim, dtype=dtype, device=device))
         self.cls_token = torch.nn.Parameter(torch.empty(1, 1, dim, dtype=dtype, device=device)) # mask_token is a pre-training param, kept only so strict loading accepts the key.
-        self.mask_token = torch.nn.Parameter(torch.empty(1, dim, dtype=dtype, device=device))
+        if use_mask_token:
+            self.mask_token = torch.nn.Parameter(torch.empty(1, dim, dtype=dtype, device=device))
+        else:
+            self.mask_token = None
+        if num_camera_tokens > 0:
+            # DA3 stores (ref_token, src_token) pairs that get injected at the
+            # alt-attn boundary; see ``Dinov2Model._inject_camera_token``.
+            self.camera_token = torch.nn.Parameter(torch.empty(1, num_camera_tokens, dim, dtype=dtype, device=device))
+        else:
+            self.camera_token = None
 
     def interpolate_pos_encoding(self, x, h_pixels, w_pixels):
         pos_embed = comfy.model_management.cast_to_device(self.position_embeddings, x.device, torch.float32)
@@ -140,12 +258,22 @@ class Dino2Embeddings(torch.nn.Module):
         patch_pos = pos_embed[:, 1:]
         N = patch_pos.shape[1]
         M = int(N ** 0.5)
+        assert N == M * M, f"DINOv2 position grid must be square, got N={N} patches (sqrt={M})"
         h0 = h_pixels // self.patch_size
         w0 = w_pixels // self.patch_size
-        scale_factor = ((h0 + 0.1) / M, (w0 + 0.1) / M)  # +0.1 matches upstream DINOv2's FP-rounding workaround so the interpolate output size lands on (h0, w0).
+        # +0.1 matches upstream DINOv2's FP-rounding workaround so the interpolate output size lands on (h0, w0).
+        # scale_factor is (height_scale, width_scale) -- height MUST come first;
+        # swapping these only happens to work for square inputs and breaks
+        # non-square paths like DA3-Small / DA3-Base multi-view.
+        scale_factor = ((h0 + 0.1) / M, (w0 + 0.1) / M)
 
         patch_pos = patch_pos.reshape(1, M, M, -1).permute(0, 3, 1, 2)
         patch_pos = torch.nn.functional.interpolate(patch_pos, scale_factor=scale_factor, mode="bicubic", antialias=False)
+        assert (h0, w0) == patch_pos.shape[-2:], (
+            f"Interpolated pos-embed grid {tuple(patch_pos.shape[-2:])} does not match "
+            f"target patch grid ({h0}, {w0}) for input {h_pixels}x{w_pixels} (patch_size={self.patch_size}); "
+            f"check scale_factor axis order and +0.1 rounding workaround"
+        )
         patch_pos = patch_pos.permute(0, 2, 3, 1).flatten(1, 2)
         return torch.cat((class_pos, patch_pos), dim=1).to(x.dtype)
 
@@ -168,12 +296,51 @@ class Dinov2Model(torch.nn.Module):
         heads = config_dict["num_attention_heads"]
         layer_norm_eps = config_dict["layer_norm_eps"]
         use_swiglu_ffn = config_dict["use_swiglu_ffn"]
+        patch_size = config_dict.get("patch_size", 14)
+        image_size = config_dict.get("image_size", 518)
+        use_mask_token = config_dict.get("use_mask_token", True)
 
-        self.embeddings = Dino2Embeddings(dim, dtype, device, operations)
-        self.encoder = Dino2Encoder(dim, heads, layer_norm_eps, num_layers, dtype, device, operations, use_swiglu_ffn = use_swiglu_ffn)
+        # DA3 extensions (all default to disabled).
+        self.alt_start = config_dict.get("alt_start", -1)
+        self.qknorm_start = config_dict.get("qknorm_start", -1)
+        self.rope_start = config_dict.get("rope_start", -1)
+        self.cat_token = config_dict.get("cat_token", False)
+        rope_freq = config_dict.get("rope_freq", 100.0)
+
+        self.embed_dim = dim
+        self.patch_size = patch_size
+        self.num_register_tokens = 0
+        self.patch_start_idx = 1
+
+        if self.rope_start != -1 and rope_freq > 0:
+            self.rope = RotaryPositionEmbedding2D(frequency=rope_freq)
+            self._position_getter = _PositionGetter()
+        else:
+            self.rope = None
+            self._position_getter = None
+
+        # camera_token shape: (1, 2, dim) -> (ref_token, src_token).
+        num_cam_tokens = 2 if self.alt_start != -1 else 0
+
+        self.embeddings = Dino2Embeddings(
+            dim, dtype, device, operations,
+            patch_size=patch_size, image_size=image_size,
+            use_mask_token=use_mask_token, num_camera_tokens=num_cam_tokens,
+        )
+        self.encoder = Dino2Encoder(
+            dim, heads, layer_norm_eps, num_layers, dtype, device, operations,
+            use_swiglu_ffn=use_swiglu_ffn,
+            qknorm_start=self.qknorm_start,
+        )
         self.layernorm = operations.LayerNorm(dim, eps=layer_norm_eps, dtype=dtype, device=device)
 
     def forward(self, pixel_values, attention_mask=None, intermediate_output=None):
+        if self.alt_start != -1:
+            raise RuntimeError(
+                "Dinov2Model.forward() is the backward-compatible CLIP-vision path and does not "
+                "apply DA3 extensions (RoPE, alternating attention, camera-token injection). "
+                "Use get_intermediate_layers_da3() for Depth Anything 3 models."
+            )
         x = self.embeddings(pixel_values)
         x, i = self.encoder(x, intermediate_output=intermediate_output)
         x = self.layernorm(x)
@@ -181,6 +348,7 @@ class Dinov2Model(torch.nn.Module):
         return x, i, pooled_output, None
 
     def get_intermediate_layers(self, pixel_values, indices, apply_norm=True):
+        """Single-view multi-layer feature extraction."""
         x = self.embeddings(pixel_values)
         optimized_attention = optimized_attention_for_device(x.device, False, small_input=True)
         n_layers = len(self.encoder.layer)
@@ -197,3 +365,132 @@ class Dinov2Model(torch.nn.Module):
             if i >= max_idx:
                 break
         return [cache[i] for i in resolved]
+
+    # ------------------------------------------------------------------
+    # Depth Anything 3 forward
+    # ------------------------------------------------------------------
+    def _prepare_rope_positions(self, B, S, H, W, device):
+        if self.rope is None:
+            return None, None
+        ph, pw = H // self.patch_size, W // self.patch_size
+        pos = self._position_getter(B * S, ph, pw, device=device)
+        # Shift so the cls/cam token at position 0 is reserved for "no diff".
+        pos = pos + 1
+        cls_pos = torch.zeros(B * S, self.patch_start_idx, 2, device=device, dtype=pos.dtype)
+        # Per-view local: real grid positions for patches, 0 for cls token.
+        pos_local = torch.cat([cls_pos, pos], dim=1)
+        # Global (across views): same grid positions; cls token still at 0,
+        # but patches share the same positions in every view.
+        pos_global = torch.cat([cls_pos, torch.zeros_like(pos) + 1], dim=1)
+        return pos_local, pos_global
+
+    def _inject_camera_token(self, x: torch.Tensor, B: int, S: int, cam_token: "torch.Tensor | None") -> torch.Tensor:
+        # x: (B, S, N, C). Replace token at index 0 with the camera token.
+        if cam_token is not None:
+            inj = cam_token
+        else:
+            ct = comfy.model_management.cast_to_device(self.embeddings.camera_token, x.device, x.dtype)
+            ref_token = ct[:, :1].expand(B, -1, -1)
+            src_token = ct[:, 1:].expand(B, max(S - 1, 0), -1)
+            inj = torch.cat([ref_token, src_token], dim=1)
+        x = x.clone()
+        x[:, :, 0] = inj
+        return x
+
+    def get_intermediate_layers_da3(self, pixel_values, out_layers, cam_token=None, ref_view_strategy="saddle_balanced", export_feat_layers=None):
+        """Multi-view multi-layer feature extraction used by Depth Anything 3."""
+        if pixel_values.ndim == 4:
+            pixel_values = pixel_values.unsqueeze(1)
+        assert pixel_values.ndim == 5 and pixel_values.shape[2] == 3, \
+            f"expected (B,3,H,W) or (B,S,3,H,W); got {tuple(pixel_values.shape)}"
+        B, S, _, H, W = pixel_values.shape
+
+        # Patch + cls + (interpolated) pos embed for each view.
+        x = pixel_values.reshape(B * S, 3, H, W)
+        x = self.embeddings(x)                          # (B*S, 1+N, C)
+        x = x.reshape(B, S, x.shape[-2], x.shape[-1])    # (B, S, 1+N, C)
+
+        pos_local, pos_global = self._prepare_rope_positions(B, S, H, W, x.device)
+        # optimized_attention is only used by blocks without QK-norm/RoPE
+        # (vanilla DINOv2 path); enabling-aware blocks fall through to SDPA.
+        optimized_attention = optimized_attention_for_device(x.device, False, small_input=True)
+
+        out_set = set(out_layers)
+        export_set = set(export_feat_layers) if export_feat_layers else set()
+        outputs: list[torch.Tensor] = []
+        aux_outputs: list[torch.Tensor] = []
+        local_x = x
+        b_idx = None
+
+
+        for i, blk in enumerate(self.encoder.layer):
+            apply_rope = self.rope is not None and i >= self.rope_start
+            block_rope = self.rope if apply_rope else None
+            l_pos = pos_local if apply_rope else None
+            g_pos = pos_global if apply_rope else None
+
+            # Reference-view selection threshold: matches the upstream constant
+            # THRESH_FOR_REF_SELECTION = 3. Skipped when a user-supplied
+            # cam_token is provided (camera info already pins the geometry).
+            if (self.alt_start != -1 and i == self.alt_start - 1 and S >= THRESH_FOR_REF_SELECTION and cam_token is None):
+                b_idx = select_reference_view(x, strategy=ref_view_strategy)
+                x = reorder_by_reference(x, b_idx)
+                local_x = reorder_by_reference(local_x, b_idx)
+
+            if self.alt_start != -1 and i == self.alt_start:
+                x = self._inject_camera_token(x, B, S, cam_token)
+
+            if self.alt_start != -1 and i >= self.alt_start and (i % 2 == 1):
+                # Global attention across views: flatten S into the seq dim.
+                t = x.reshape(B, S * x.shape[-2], x.shape[-1])
+                p = g_pos.reshape(B, S * g_pos.shape[-2], g_pos.shape[-1]) if g_pos is not None else None
+                t = blk(t, optimized_attention=optimized_attention, pos=p, rope=block_rope)
+                x = t.reshape(B, S, x.shape[-2], x.shape[-1])
+            else:
+                # Per-view local attention.
+                t = x.reshape(B * S, x.shape[-2], x.shape[-1])
+                p = l_pos.reshape(B * S, l_pos.shape[-2], l_pos.shape[-1]) if l_pos is not None else None
+                t = blk(t, optimized_attention=optimized_attention, pos=p, rope=block_rope)
+                x = t.reshape(B, S, x.shape[-2], x.shape[-1])
+                local_x = x
+
+            if i in out_set:
+                if self.cat_token:
+                    out_x = torch.cat([local_x, x], dim=-1)
+                else:
+                    out_x = x
+                # Restore original view order on the way out so heads see views
+                # in the user's expected order.
+                if b_idx is not None and self.alt_start != -1:
+                    out_x = restore_original_order(out_x, b_idx)
+                outputs.append(out_x)
+
+            if i in export_set:
+                aux = x
+                if b_idx is not None and self.alt_start != -1:
+                    aux = restore_original_order(aux, b_idx)
+                aux_outputs.append(aux)
+
+        # Apply final norm. When cat_token is set, only the right half
+        # ("global" features) is normalised; the left half is left as-is to
+        # match the upstream DA3 head signature.
+        normed: list[torch.Tensor] = []
+        cls_tokens: list[torch.Tensor] = []
+        for out_x in outputs:
+            cls_tokens.append(out_x[:, :, 0])
+            if out_x.shape[-1] == self.embed_dim:
+                normed.append(self.layernorm(out_x))
+            elif out_x.shape[-1] == self.embed_dim * 2:
+                left = out_x[..., :self.embed_dim]
+                right = self.layernorm(out_x[..., self.embed_dim:])
+                normed.append(torch.cat([left, right], dim=-1))
+            else:
+                raise ValueError(f"Unexpected token width: {out_x.shape[-1]}")
+
+        # Drop cls/cam token from the patch sequence.
+        normed = [o[..., 1 + self.num_register_tokens:, :] for o in normed]
+
+        # Final layernorm + drop cls token from auxiliary features too.
+        aux_normed = [self.layernorm(o)[..., 1 + self.num_register_tokens:, :]
+                      for o in aux_outputs]
+        return list(zip(normed, cls_tokens)), aux_normed

comfy/image_encoders/dino3.py

@@ -0,0 +1,259 @@
+import math
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+import comfy.ops
+from comfy.ldm.modules.attention import optimized_attention_for_device
+from comfy.image_encoders.dino2 import LayerScale as DINOv3ViTLayerScale
+
+
+# DINOv3 ViT-H/16+ (SwiGLU)
+DINOV3_VITH_CONFIG = {
+    "model_type": "dinov3",
+    "num_hidden_layers": 32,
+    "hidden_size": 1280,
+    "num_attention_heads": 20,
+    "num_register_tokens": 4,
+    "intermediate_size": 5120,
+    "layer_norm_eps": 1e-5,
+    "num_channels": 3,
+    "patch_size": 16,
+    "rope_theta": 100.0,
+    "use_gated_mlp": True,
+    "gated_mlp_act": "silu",
+    "image_size": 1024,
+    "image_mean": [0.485, 0.456, 0.406],
+    "image_std": [0.229, 0.224, 0.225],
+}
+
+
+class DINOv3ViTMLP(nn.Module):
+    def __init__(self, hidden_size, intermediate_size, mlp_bias, device, dtype, operations):
+        super().__init__()
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.up_proj = operations.Linear(self.hidden_size, self.intermediate_size, bias=mlp_bias, device=device, dtype=dtype)
+        self.down_proj = operations.Linear(self.intermediate_size, self.hidden_size, bias=mlp_bias, device=device, dtype=dtype)
+        self.act_fn = torch.nn.GELU()
+
+    def forward(self, x):
+        return self.down_proj(self.act_fn(self.up_proj(x)))
+
+
+def rotate_half(x):
+    x1 = x[..., : x.shape[-1] // 2]
+    x2 = x[..., x.shape[-1] // 2 :]
+    return torch.cat((-x2, x1), dim=-1)
+
+
+def apply_rotary_pos_emb(q, k, cos, sin, **kwargs):
+    num_tokens = q.shape[-2]
+    num_patches = sin.shape[-2]
+    num_prefix_tokens = num_tokens - num_patches
+
+    q_prefix_tokens, q_patches = q.split((num_prefix_tokens, num_patches), dim=-2)
+    k_prefix_tokens, k_patches = k.split((num_prefix_tokens, num_patches), dim=-2)
+
+    q_patches = (q_patches * cos) + (rotate_half(q_patches) * sin)
+    k_patches = (k_patches * cos) + (rotate_half(k_patches) * sin)
+
+    q = torch.cat((q_prefix_tokens, q_patches), dim=-2)
+    k = torch.cat((k_prefix_tokens, k_patches), dim=-2)
+
+    return q, k
+
+
+class DINOv3ViTAttention(nn.Module):
+    def __init__(self, hidden_size, num_attention_heads, device, dtype, operations):
+        super().__init__()
+        self.embed_dim = hidden_size
+        self.num_heads = num_attention_heads
+        self.head_dim = self.embed_dim // self.num_heads
+
+        self.k_proj = operations.Linear(self.embed_dim, self.embed_dim, bias=False, device=device, dtype=dtype)  # key_bias = False
+        self.v_proj = operations.Linear(self.embed_dim, self.embed_dim, bias=True, device=device, dtype=dtype)
+        self.q_proj = operations.Linear(self.embed_dim, self.embed_dim, bias=True, device=device, dtype=dtype)
+        self.o_proj = operations.Linear(self.embed_dim, self.embed_dim, bias=True, device=device, dtype=dtype)
+
+    def forward(self, hidden_states, attention_mask=None, position_embeddings=None, **kwargs):
+        batch_size, patches, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        query_states = query_states.view(batch_size, patches, self.num_heads, self.head_dim).transpose(1, 2)
+        key_states = key_states.view(batch_size, patches, self.num_heads, self.head_dim).transpose(1, 2)
+        value_states = value_states.view(batch_size, patches, self.num_heads, self.head_dim).transpose(1, 2)
+
+        if position_embeddings is not None:
+            cos, sin = position_embeddings
+            query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin)
+
+        attn = optimized_attention_for_device(query_states.device, mask=False)
+        attn_output = attn(
+            query_states, key_states, value_states, self.num_heads, attention_mask,
+            skip_reshape=True, skip_output_reshape=True, low_precision_attention=False,
+        )
+
+        attn_output = attn_output.transpose(1, 2)
+        attn_output = attn_output.reshape(batch_size, patches, -1).contiguous()
+        attn_output = self.o_proj(attn_output)
+        return attn_output
+
+
+class DINOv3ViTGatedMLP(nn.Module):
+    def __init__(self, hidden_size, intermediate_size, mlp_bias, device, dtype, operations, act="silu"):
+        super().__init__()
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.gate_proj = operations.Linear(self.hidden_size, self.intermediate_size, bias=mlp_bias, device=device, dtype=dtype)
+        self.up_proj = operations.Linear(self.hidden_size, self.intermediate_size, bias=mlp_bias, device=device, dtype=dtype)
+        self.down_proj = operations.Linear(self.intermediate_size, self.hidden_size, bias=mlp_bias, device=device, dtype=dtype)
+        self.act_fn = torch.nn.SiLU() if act == "silu" else torch.nn.GELU()
+
+    def forward(self, x):
+        return self.down_proj(self.act_fn(self.gate_proj(x)) * self.up_proj(x))
+
+
+def get_patches_center_coordinates(num_patches_h, num_patches_w, dtype, device):
+    coords_h = torch.arange(0.5, num_patches_h, dtype=dtype, device=device)
+    coords_w = torch.arange(0.5, num_patches_w, dtype=dtype, device=device)
+    coords_h = coords_h / num_patches_h
+    coords_w = coords_w / num_patches_w
+    coords = torch.stack(torch.meshgrid(coords_h, coords_w, indexing="ij"), dim=-1)
+    coords = coords.flatten(0, 1)
+    coords = 2.0 * coords - 1.0
+    return coords
+
+
+class DINOv3ViTRopePositionEmbedding(nn.Module):
+    inv_freq: torch.Tensor
+
+    def __init__(self, rope_theta, hidden_size, num_attention_heads, patch_size, device, dtype):
+        super().__init__()
+        self.base = rope_theta
+        self.head_dim = hidden_size // num_attention_heads
+        self.patch_size = patch_size
+
+        inv_freq = 1 / self.base ** torch.arange(0, 1, 4 / self.head_dim, dtype=torch.float32, device=device)
+        self.register_buffer("inv_freq", inv_freq, persistent=False)
+
+    def forward(self, pixel_values):
+        _, _, height, width = pixel_values.shape
+        num_patches_h = height // self.patch_size
+        num_patches_w = width // self.patch_size
+
+        patch_coords = get_patches_center_coordinates(num_patches_h, num_patches_w, dtype=torch.float32, device=pixel_values.device)
+        self.inv_freq = self.inv_freq.to(pixel_values.device)
+        angles = 2 * math.pi * patch_coords[:, :, None] * self.inv_freq[None, None, :]
+        angles = angles.flatten(1, 2)
+        angles = angles.tile(2)
+        cos = torch.cos(angles).to(dtype=pixel_values.dtype)
+        sin = torch.sin(angles).to(dtype=pixel_values.dtype)
+        return cos, sin
+
+
+class DINOv3ViTEmbeddings(nn.Module):
+    def __init__(self, hidden_size, num_register_tokens, num_channels, patch_size, dtype, device, operations):
+        super().__init__()
+        self.cls_token = nn.Parameter(torch.empty(1, 1, hidden_size, device=device, dtype=dtype))
+        self.mask_token = nn.Parameter(torch.empty(1, 1, hidden_size, device=device, dtype=dtype))
+        self.register_tokens = nn.Parameter(torch.empty(1, num_register_tokens, hidden_size, device=device, dtype=dtype))
+        self.patch_embeddings = operations.Conv2d(
+            num_channels, hidden_size, kernel_size=patch_size, stride=patch_size, device=device, dtype=dtype
+        )
+
+    def forward(self, pixel_values, bool_masked_pos=None):
+        batch_size = pixel_values.shape[0]
+
+        patch_embeddings = self.patch_embeddings(pixel_values)
+        patch_embeddings = patch_embeddings.flatten(2).transpose(1, 2)
+
+        if bool_masked_pos is not None:
+            mask_token = comfy.ops.cast_to_input(self.mask_token, patch_embeddings)
+            patch_embeddings = torch.where(bool_masked_pos.unsqueeze(-1), mask_token, patch_embeddings)
+
+        cls_token = comfy.ops.cast_to_input(self.cls_token.expand(batch_size, -1, -1), patch_embeddings)
+        register_tokens = comfy.ops.cast_to_input(self.register_tokens.expand(batch_size, -1, -1), patch_embeddings)
+        embeddings = torch.cat([cls_token, register_tokens, patch_embeddings], dim=1)
+        return embeddings
+
+
+class DINOv3ViTLayer(nn.Module):
+    def __init__(self, hidden_size, layer_norm_eps, use_gated_mlp, mlp_bias, intermediate_size,
+                 num_attention_heads, device, dtype, operations, gated_mlp_act="silu"):
+        super().__init__()
+        self.norm1 = operations.LayerNorm(hidden_size, eps=layer_norm_eps, device=device, dtype=dtype)
+        self.attention = DINOv3ViTAttention(hidden_size, num_attention_heads, device=device, dtype=dtype, operations=operations)
+        self.layer_scale1 = DINOv3ViTLayerScale(hidden_size, device=device, dtype=dtype, operations=None)
+
+        self.norm2 = operations.LayerNorm(hidden_size, eps=layer_norm_eps, device=device, dtype=dtype)
+        if use_gated_mlp:
+            self.mlp = DINOv3ViTGatedMLP(hidden_size, intermediate_size, mlp_bias, device=device, dtype=dtype, operations=operations, act=gated_mlp_act)
+        else:
+            self.mlp = DINOv3ViTMLP(hidden_size, intermediate_size=intermediate_size, mlp_bias=mlp_bias, device=device, dtype=dtype, operations=operations)
+        self.layer_scale2 = DINOv3ViTLayerScale(hidden_size, device=device, dtype=dtype, operations=None)
+
+    def forward(self, hidden_states, attention_mask=None, position_embeddings=None):
+        residual = hidden_states
+        hidden_states = self.norm1(hidden_states)
+        hidden_states = self.attention(hidden_states, attention_mask=attention_mask, position_embeddings=position_embeddings)
+        hidden_states = self.layer_scale1(hidden_states)
+        hidden_states = hidden_states + residual
+
+        residual = hidden_states
+        hidden_states = self.norm2(hidden_states)
+        hidden_states = self.mlp(hidden_states)
+        hidden_states = self.layer_scale2(hidden_states)
+        hidden_states = hidden_states + residual
+        return hidden_states
+
+
+class DINOv3ViTModel(nn.Module):
+    def __init__(self, config, dtype, device, operations):
+        super().__init__()
+        num_hidden_layers = config["num_hidden_layers"]
+        hidden_size = config["hidden_size"]
+        num_attention_heads = config["num_attention_heads"]
+        num_register_tokens = config["num_register_tokens"]
+        intermediate_size = config["intermediate_size"]
+        layer_norm_eps = config["layer_norm_eps"]
+        num_channels = config["num_channels"]
+        patch_size = config["patch_size"]
+        rope_theta = config["rope_theta"]
+        use_gated_mlp = config.get("use_gated_mlp", False)
+        gated_mlp_act = config.get("gated_mlp_act", "silu")
+
+        self.embeddings = DINOv3ViTEmbeddings(
+            hidden_size, num_register_tokens, num_channels=num_channels, patch_size=patch_size,
+            dtype=dtype, device=device, operations=operations
+        )
+        self.rope_embeddings = DINOv3ViTRopePositionEmbedding(
+            rope_theta, hidden_size, num_attention_heads, patch_size=patch_size, dtype=dtype, device=device
+        )
+        self.layer = nn.ModuleList([
+            DINOv3ViTLayer(hidden_size, layer_norm_eps, use_gated_mlp=use_gated_mlp, mlp_bias=True,
+                           intermediate_size=intermediate_size, num_attention_heads=num_attention_heads,
+                           dtype=dtype, device=device, operations=operations, gated_mlp_act=gated_mlp_act)
+            for _ in range(num_hidden_layers)])
+        self.norm = operations.LayerNorm(hidden_size, eps=layer_norm_eps, dtype=dtype, device=device)
+
+    def get_input_embeddings(self):
+        return self.embeddings.patch_embeddings
+
+    def forward(self, pixel_values, bool_masked_pos=None, **kwargs):
+        hidden_states = self.embeddings(pixel_values, bool_masked_pos=bool_masked_pos)
+        position_embeddings = self.rope_embeddings(pixel_values)
+
+        for layer_module in self.layer:
+            hidden_states = layer_module(hidden_states, position_embeddings=position_embeddings)
+
+        if kwargs.get("skip_norm_elementwise", False):
+            sequence_output = F.layer_norm(hidden_states, hidden_states.shape[-1:])
+        else:
+            norm = self.norm.to(hidden_states.device)
+            sequence_output = norm(hidden_states)
+        pooled_output = sequence_output[:, 0, :]
+        return sequence_output, None, pooled_output, None

comfy/latent_formats.py

@@ -239,6 +239,16 @@ class Flux2(LatentFormat):
     def process_out(self, latent):
         return latent
 
+class TripoSplat(LatentFormat):
+    # Sequence latent (B, 8192, 16) the camera token rides alongside as a second nested latent
+    latent_channels = 16
+
+    def process_in(self, latent):
+        return latent
+
+    def process_out(self, latent):
+        return latent
+
 class Mochi(LatentFormat):
     latent_channels = 12
     latent_dimensions = 3
@@ -799,13 +809,15 @@ class ZImagePixelSpace(ChromaRadiance):
     """
     pass
 
-
 class HiDreamO1Pixel(ChromaRadiance):
     """Pixel-space latent format for HiDream-O1.
     No VAE — model patches/unpatches raw RGB internally with patch_size=32.
     """
     pass
 
+class PixelDiTPixel(ChromaRadiance):
+    pass
+
 class CogVideoX(LatentFormat):
     """Latent format for CogVideoX-2b (THUDM/CogVideoX-2b).
 

comfy/ldm/audio/dit.py

@@ -433,11 +433,11 @@ class Attention(nn.Module):
         if self.differential:
             q, q_diff = q.unbind(dim=1)
             k, k_diff = k.unbind(dim=1)
-            out      = optimized_attention(q,      k,      v, h, skip_reshape=True, transformer_options=transformer_options)
-            out_diff = optimized_attention(q_diff, k_diff, v, h, skip_reshape=True, transformer_options=transformer_options)
+            out      = optimized_attention(q,      k,      v, h, skip_reshape=True, low_precision_attention=False, transformer_options=transformer_options)
+            out_diff = optimized_attention(q_diff, k_diff, v, h, skip_reshape=True, low_precision_attention=False, transformer_options=transformer_options)
             out = out - out_diff
         else:
-            out = optimized_attention(q, k, v, h, skip_reshape=True, transformer_options=transformer_options)
+            out = optimized_attention(q, k, v, h, skip_reshape=True, low_precision_attention=False, transformer_options=transformer_options)
 
         out = self.to_out(out)
 

comfy/ldm/audio/vae_sa3.py

@@ -138,11 +138,11 @@ class Attention(nn.Module):
                 k_diff = _apply_rotary_pos_emb(k_diff.float(), freqs).to(k_dtype)
 
         if self.differential:
-            out = (optimized_attention(q, k, v, h, mask=mask, skip_reshape=True)
-                   - optimized_attention(q_diff, k_diff, v, h, mask=mask, skip_reshape=True))
+            out = (optimized_attention(q, k, v, h, mask=mask, skip_reshape=True, low_precision_attention=False)
+                   - optimized_attention(q_diff, k_diff, v, h, mask=mask, skip_reshape=True, low_precision_attention=False))
             del q, k, v, q_diff, k_diff
         else:
-            out = optimized_attention(q, k, v, h, mask=mask, skip_reshape=True)
+            out = optimized_attention(q, k, v, h, mask=mask, skip_reshape=True, low_precision_attention=False)
             del q, k, v
 
         return self.to_out(out)