Jedrzej Kosinski
ca39552954
Merge branch 'master' into asset-management
2025-09-24 23:44:57 -07:00
Guy Niv
c8d2117f02
Fix memory leak by properly detaching model finalizer ( #9979 )
...
When unloading models in load_models_gpu(), the model finalizer was not
being explicitly detached, leading to a memory leak. This caused
linear memory consumption increase over time as models are repeatedly
loaded and unloaded.
This change prevents orphaned finalizer references from accumulating in
memory during model switching operations.
2025-09-24 22:35:12 -04:00
comfyanonymous
fccab99ec0
Fix issue with .view() in HuMo. ( #10014 )
2025-09-24 20:09:42 -04:00
Jukka Seppänen
fd79d32f38
Add new audio nodes ( #9908 )
...
* Add new audio nodes
- TrimAudioDuration
- SplitAudioChannels
- AudioConcat
- AudioMerge
- AudioAdjustVolume
* Update nodes_audio.py
* Add EmptyAudio -node
* Change duration to Float (allows sub seconds)
2025-09-24 18:59:29 -04:00
Changrz
341b4adefd
Rodin3D - add [Rodin3D Gen-2 generate] api-node ( #9994 )
...
* update Rodin api node
* update rodin3d gen2 api node
* fix images limited bug
2025-09-24 14:05:37 -04:00
comfyanonymous
b8730510db
ComfyUI version 0.3.60
2025-09-23 11:50:33 -04:00
Alexander Piskun
e808790799
feat(api-nodes): add wan t2i, t2v, i2v nodes ( #9996 )
2025-09-23 11:36:47 -04:00
ComfyUI Wiki
145b0e4f79
update template to 0.1.86 ( #9998 )
...
* update template to 0.1.84
* update template to 0.1.85
* Update template to 0.1.86
2025-09-23 11:22:35 -04:00
comfyanonymous
707b2638ec
Fix bug with WanAnimateToVideo. ( #9990 )
2025-09-22 17:34:33 -04:00
comfyanonymous
8a5ac527e6
Fix bug with WanAnimateToVideo node. ( #9988 )
2025-09-22 17:26:58 -04:00
Christian Byrne
e3206351b0
add offset param ( #9977 )
2025-09-22 17:12:32 -04:00
comfyanonymous
1fee8827cb
Support for qwen edit plus model. Use the new TextEncodeQwenImageEditPlus. ( #9986 )
2025-09-22 16:49:48 -04:00
comfyanonymous
27bc181c49
Set some wan nodes as no longer experimental. ( #9976 )
2025-09-21 19:48:31 -04:00
comfyanonymous
d1d9eb94b1
Lower wan memory estimation value a bit. ( #9964 )
...
Previous pr reduced the peak memory requirement.
2025-09-20 22:09:35 -04:00
Kohaku-Blueleaf
7be2b49b6b
Fix LoRA Trainer bugs with FP8 models. ( #9854 )
...
* Fix adapter weight init
* Fix fp8 model training
* Avoid inference tensor
2025-09-20 21:24:48 -04:00
Jedrzej Kosinski
9ed3c5cc09
[Reviving #5709 ] Add strength input to Differential Diffusion ( #9957 )
...
* Update nodes_differential_diffusion.py
* Update nodes_differential_diffusion.py
* Make strength optional to avoid validation errors when loading old workflows, adjust step
---------
Co-authored-by: ThereforeGames <eric@sparknight.io>
2025-09-20 21:10:39 -04:00
comfyanonymous
66241cef31
Add inputs for character replacement to the WanAnimateToVideo node. ( #9960 )
2025-09-20 02:24:10 -04:00
comfyanonymous
e8df53b764
Update WanAnimateToVideo to more easily extend videos. ( #9959 )
2025-09-19 18:48:56 -04:00
Alexander Piskun
852704c81a
fix(seedream4): add flag to ignore error on partial success ( #9952 )
2025-09-19 16:04:51 -04:00
Alexander Piskun
9fdf8c25ab
api_nodes: reduce default timeout from 7 days to 2 hours ( #9918 )
2025-09-19 16:02:43 -04:00
comfyanonymous
dc95b6acc0
Basic WIP support for the wan animate model. ( #9939 )
2025-09-19 03:07:17 -04:00
Christian Byrne
711bcf33ee
Bump frontend to 1.26.13 ( #9933 )
2025-09-19 03:03:30 -04:00
comfyanonymous
24b0fce099
Do padding of audio embed in model for humo for more flexibility. ( #9935 )
2025-09-18 19:54:16 -04:00
Jodh Singh
1ea8c54064
make kernel of same type as image to avoid mismatch issues ( #9932 )
2025-09-18 19:51:16 -04:00
DELUXA
8d6653fca6
Enable fp8 ops by default on gfx1200 ( #9926 )
2025-09-18 19:50:37 -04:00
Jedrzej Kosinski
4dd843d36f
Merge branch 'master' into asset-management
2025-09-18 14:08:20 -07:00
Jedrzej Kosinski
46fdd636de
Merge pull request #9545 from bigcat88/asset-management
...
[Assets] Initial implementation
2025-09-18 14:07:18 -07:00
bigcat88
283cd27bdc
final adjustments
2025-09-18 10:05:32 +03:00
comfyanonymous
dd611a7700
Support the HuMo 17B model. ( #9912 )
2025-09-17 18:39:24 -04:00
bigcat88
1a37d1476d
refactor(6): fully batched initial scan
2025-09-17 20:29:29 +03:00
bigcat88
f9602457d6
optimization: initial scan speed(batching metadata[filename])
2025-09-17 16:47:27 +03:00
bigcat88
85ef08449d
optimization: initial scan speed(batching tags)
2025-09-17 14:08:57 +03:00
bigcat88
5b6810a2c6
fixed hash calculation during model loading in ComfyUI
2025-09-17 13:25:56 +03:00
bigcat88
621faaa195
refactor(5): use less DB queries to create seed asset
2025-09-17 10:46:21 +03:00
comfyanonymous
9288c78fc5
Support the HuMo model. ( #9903 )
2025-09-17 00:12:48 -04:00
rattus128
e42682b24e
Reduce Peak WAN inference VRAM usage ( #9898 )
...
* flux: Do the xq and xk ropes one at a time
This was doing independendent interleaved tensor math on the q and k
tensors, leading to the holding of more than the minimum intermediates
in VRAM. On a bad day, it would VRAM OOM on xk intermediates.
Do everything q and then everything k, so torch can garbage collect
all of qs intermediates before k allocates its intermediates.
This reduces peak VRAM usage for some WAN2.2 inferences (at least).
* wan: Optimize qkv intermediates on attention
As commented. The former logic computed independent pieces of QKV in
parallel which help more inference intermediates in VRAM spiking
VRAM usage. Fully roping Q and garbage collecting the intermediates
before touching K reduces the peak inference VRAM usage.
2025-09-16 19:21:14 -04:00
bigcat88
d0aa64d57b
refactor(4): use one query to init DB with all tags for assets
2025-09-16 21:18:18 +03:00
bigcat88
677a0e2508
refactor(3): unite logic for Asset fast check
2025-09-16 20:29:50 +03:00
bigcat88
31ec744317
refactor(2)/fix: skip double checking the existing files during fast check
2025-09-16 19:50:21 +03:00
bigcat88
a336c7c165
refactor(1): use general fast_asset_file_check helper for fast check
2025-09-16 19:19:18 +03:00
bigcat88
77332d3054
optimization: fast scan: commit to the DB in chunks
2025-09-16 14:21:40 +03:00
bigcat88
24a95f5ca4
removed default scanning of "input" and "output" folders; added separate endpoint for test suite.
2025-09-16 11:28:29 +03:00
comfyanonymous
a39ac59c3e
Add encoder part of whisper large v3 as an audio encoder model. ( #9894 )
...
Not useful yet but some models use it.
2025-09-16 01:19:50 -04:00
blepping
1a85483da1
Fix depending on asserts to raise an exception in BatchedBrownianTree and Flash attn module ( #9884 )
...
Correctly handle the case where w0 is passed by kwargs in BatchedBrownianTree
2025-09-15 20:05:03 -04:00
comfyanonymous
47a9cde5d3
Support the omnigen2 umo lora. ( #9886 )
2025-09-15 18:10:55 -04:00
bigcat88
0be513b213
fix: escape "_" symbol in all other places
2025-09-15 20:26:48 +03:00
bigcat88
f1fb7432a0
fix+test: escape "_" symbol in assets filtering
2025-09-15 19:19:47 +03:00
bigcat88
f3cf99d10c
fix+test: escape "_" symbol in tags filtering
2025-09-15 17:29:27 +03:00
bigcat88
5f187fe6fb
optimization: make list_unhashed_candidates_under_prefixes single-query instead of N+1
2025-09-15 12:46:35 +03:00
bigcat88
025fc49b4e
optimization: DB Queries (Tags)
2025-09-15 10:26:13 +03:00
