1991 Commits

Author	SHA1	Message	Date
rattus	ef73070ea4	mp: Fix checkpoint saving (#12268) ... Fix regression in the recent model saving refactor. Pass the non unet pieces down the layers so that checkpoints are complete.	2026-02-04 02:08:45 -05:00
rattus	d30c609f5a	utils: safetensors: dont slice data on torch level (#12266) ... Torch has alignment enforcement when viewing with data type changes but only relative to itself. Do all tensor constructions straight off the memory-view individually so pytorch doesnt see an alignment problem. The is needed for handling misaligned safetensors weights, which are reasonably common in third party models. This limits usage of this safetensors loader to GPU compute only as CPUs kernnel are very likely to bus error. But it works for dynamic_vram, where we really dont want to take a deep copy and we always use GPU copy_ which disentangles the misalignment.	2026-02-04 01:48:47 -05:00
comfyanonymous	a31681564d	Fix crash with ace step 1.5 (#12264)	2026-02-04 00:03:21 -05:00
rattus	855849c658	mm: Remove Aimdo exemption for empty_cache (#12260) ... Its more important to get the torch caching allocator GC up and running than supporting the pyt2.7 bug. Switch it on. Defeature dynamic_vram + pyt2.7.	2026-02-03 21:39:19 -05:00
comfyanonymous	fe2511468d	Support the 4B ace step 1.5 lm model. (#12257) ... Can be used as an alternative to the 1.7B	2026-02-03 19:01:38 -05:00
comfyanonymous	b8315e66cb	Fix tiled vae for ace step 1.5 (#12253)	2026-02-03 14:40:45 -05:00
comfyanonymous	ab1050bec3	Support ace step 1.5 base model loras. (#12252)	2026-02-03 13:54:23 -05:00
comfyanonymous	85fc35e8fa	Fix mac issue. (#12250)	2026-02-03 12:19:39 -05:00
comfyanonymous	223364743c	llama: cast logits as a comfy-weight (#12248) ... This is using a different layers weight with .to(). Change it to use the ops caster if the original layer is a comfy weight so that it picks up dynamic_vram and async_offload functionality in full. Co-authored-by: Rattus <rattus128@gmail.com>	2026-02-03 11:31:36 -05:00
comfyanonymous	affe881354	Fix some issues with mac. (#12247)	2026-02-03 11:07:04 -05:00
comfyanonymous	f5030e26fd	Add progress bar to ace step. (#12242)	2026-02-03 04:09:30 -05:00
comfyanonymous	3c1a1a2df8	Basic support for the ace step 1.5 model. (#12237)	2026-02-03 00:06:18 -05:00
comfyanonymous	c05a08ae66	Add back function. (#12234)	2026-02-02 19:52:07 -05:00
rattus	de9ada6a41	Dynamic VRAM unloading fix (#12227) ... * mp: fix full dynamic unloading This was not unloading dynamic models when requesting a full unload via the unpatch() code path. This was ok, i your workflow was all dynamic models but fails with big VRAM leaks if you need to fully unload something for a regular ModelPatcher It also fices the "unload models" button. * mm: load models outside of Aimdo Mempool In dynamic_vram mode, escape the Aimdo mempool and load into the regular mempool. Use a dummy thread to do it.	2026-02-02 17:35:20 -05:00
rattus	37f711d4a1	mm: Fix cast buffers with intel offloading (#12229) ... Intel has offloading support but there were some nvidia calls in the new cast buffer stuff.	2026-02-02 17:34:46 -05:00
comfyanonymous	dd86b15521	Enable embeddings for some qwen 3 models. (#12218)	2026-02-02 03:51:09 -05:00
comfyanonymous	021ba20719	Fix issue with parameters on root model object. (#12216)	2026-02-01 20:12:52 -05:00
rattus	2b5da3b72e	dynamic_vram: silence pytorch buffer warning (#12210) ... This is log clutter and concerning to users. Its a false alarm.	2026-02-01 20:09:55 -05:00
rattus	794d05bdb1	dynamic_vram: respect argument cast dtypes in non-comfy weights (#12209) ... This function has a dtype argument that allows the caller to set the dtype in the cast. TIL Some models override this on weight casts, which means its the highest priority. Priority scheme is: argument > model dtype > state dict dtype	2026-02-01 20:09:21 -05:00
rattus	361b9a82a3	fix pinning with model defined dtype (#12208) ... pinned memory was converted back to pinning the CPU side weight without any changes. Fix the pinner to use the CPU weight and not the model defined geometry. This will either save RAM or stop buffer overruns when the types mismatch. Fix the model defined weight caster to use the [ s.weight, s.bias ] interpretation, as xfer_dest might be the flattened pin now. Fix the detection of needing to cast to not be conditional on !pin.	2026-02-01 08:42:32 -08:00
comfyanonymous	667a1b8878	Fix some custom nodes breaking. (#12203)	2026-02-01 01:55:18 -05:00
rattus	f8acd9c402	Reduce RAM usage, fix VRAM OOMs, and fix Windows shared memory spilling with adaptive model loading (#11845)	2026-02-01 01:01:11 -05:00
comfyanonymous	873de5f37a	KV cache implementation for using llama models for text generation. (#12195)	2026-01-31 21:11:11 -05:00
comfyanonymous	b8f848bfe3	Fix model not working with any res. (#12186)	2026-01-31 00:12:48 -05:00
comfyanonymous	c9b633d84f	Add missing spacial downscale ratios. (#12146)	2026-01-28 20:52:51 -05:00
guill	dcff27fe3f	Add support for dev-only nodes. (#12106) ... When a node is declared as dev-only, it doesn't show in the default UI unless the dev mode is enabled in the settings. The intention is to allow nodes related to unit testing to be included in ComfyUI distributions without confusing the average user.	2026-01-27 13:03:29 -08:00
rattus	6516ab335d	wan-vae: Switch off feature cache for single frame (#12090) ... The code throughout is None safe to just skip the feature cache saving step if none. Set it none in single frame use so qwen doesn't burn VRAM on the unused cache.	2026-01-26 19:40:19 -05:00
comfyanonymous	2129e7d278	Fix mistral 3 tokenizer code failing on latest transformers version and other breakage. (#12095) ... * Fix mistral 3 tokenizer code failing on latest transformers version. * Add requests to the requirements	2026-01-26 11:39:00 -05:00
Kohaku-Blueleaf	a97c98068f	[Weight-adapter/Trainer] Bypass forward mode in Weight adapter system (#11958) ... * Add API of bypass forward module * bypass implementation * add bypass fwd into nodes list/trainer	2026-01-24 22:56:22 -05:00
comfyanonymous	635406e283	Only enable fp16 on z image models that actually support it. (#12065)	2026-01-24 22:32:28 -05:00
comfyanonymous	aef4e13588	Make empty latent node work with other models. (#12062)	2026-01-24 19:23:20 -05:00
rattus	4e6a1b66a9	speed up and reduce VRAM of QWEN VAE and WAN (less so) (#12036) ... * ops: introduce autopad for conv3d This works around pytorch missing ability to causal pad as part of the kernel and avoids massive weight duplications for padding. * wan-vae: rework causal padding This currently uses F.pad which takes a full deep copy and is liable to be the VRAM peak. Instead, kick spatial padding back to the op and consolidate the temporal padding with the cat for the cache. * wan-vae: implement zero pad fast path The WAN VAE is also QWEN where it is used single-image. These convolutions are however zero padded 3d convolutions, which means the VAE is actually just 2D down the last element of the conv weight in the temporal dimension. Fast path this, to avoid adding zeros that then just evaporate in convoluton math but cost computation.	2026-01-23 19:56:14 -05:00
comfyanonymous	9cf299a9f9	Make regular empty latent node work properly on flux 2 variants. (#12050)	2026-01-23 19:50:48 -05:00
ComfyUI Wiki	e89b22993a	Support ModelScope-Trainer/DiffSynth LoRA format for Flux.2 Klein models (#12042)	2026-01-23 15:27:49 -05:00
Jukka Seppänen	55bd606e92	LTX2: Refactor forward function for better VRAM efficiency and fix spatial inpainting (#12046) ... * Disable timestep embed compression when inpainting Spatial inpainting not compatible with the compression * Reduce crossattn peak VRAM * LTX2: Refactor forward function for better VRAM efficiency	2026-01-23 15:26:38 -05:00
Omri Marom	d7f3241bf6	qwen_image: propagate attention mask. (#11966)	2026-01-22 20:02:31 -05:00
comfyanonymous	09a2e67151	Support loading flux 2 klein checkpoints saved with SaveCheckpoint. (#12033)	2026-01-22 18:20:48 -05:00
rattus	0fd1b78736	Reduce LTX2 VAE VRAM consumption (#12028) ... * causal_video_ae: Remove attention ResNet This attention_head_dim argument does not exist on this constructor so this is dead code. Remove as generic attention mid VAE conflicts with temporal roll. * ltx-vae: consoldate causal/non-causal code paths * ltx-vae: add cache rolling adder * ltx-vae: use cached adder for resnet * ltx-vae: Implement rolling VAE Implement a temporal rolling VAE for the LTX2 VAE. Usually when doing temporal rolling VAEs you can just chunk on time relying on causality and cache behind you as you go. The LTX VAE is however non-causal. So go whole hog and implement per layer run ahead and backpressure between the decoder layers using recursive state beween the layers. Operations are ammended with temporal_cache_state{} which they can use to hold any state then need for partial execution. Convolutions cache their inputs behind the up to N-1 frames, and skip connections need to cache the mismatch between convolution input and output that happens due to missing future (non-causal) input. Each call to run_up() processes a layer accross a range on input that may or may not be complete. It goes depth first to process as much as possible to try and digest frames to the final output ASAP. If layers run out of input due to convolution losses, they simply return without action effectively applying back-pressure to the earlier layers. As the earlier layers do more work and caller deeper, the partial states are reconciled and output continues to digest depth first as much as possible. Chunking is done using a size quota rather than a fixed frame length and any layer can initiate chunking, and multiple layers can chunk at different granulatiries. This remove the old limitation of always having to process 1 latent frame to entirety and having to hold 8 full decoded frames as the VRAM peak.	2026-01-22 16:54:18 -05:00
Jukka Seppänen	16b9aabd52	Support Multi/InfiniteTalk (#10179) ... * re-init * Update model_multitalk.py * whitespace... * Update model_multitalk.py * remove print * this is redundant * remove import * Restore preview functionality * Move block_idx to transformer_options * Remove LoopingSamplerCustomAdvanced * Remove looping functionality, keep extension functionality * Update model_multitalk.py * Handle ref_attn_mask with separate patch to avoid having to always return q and k from self_attn * Chunk attention map calculation for multiple speakers to reduce peak VRAM usage * Update model_multitalk.py * Add ModelPatch type back * Fix for latest upstream * Use DynamicCombo for cleaner node Basically just so that single_speaker mode hides mask inputs and 2nd audio input * Update nodes_wan.py	2026-01-21 23:09:48 -05:00
Jukka Seppänen	245f6139b6	More targeted embedding_connector loading for LTX2 text encoder (#11992) ... Reduces errors	2026-01-21 23:05:06 -05:00
Jukka Seppänen	3365ad18a5	Support LTX2 tiny vae (taeltx_2) (#11929)	2026-01-21 23:03:51 -05:00
comfyanonymous	abe2ec26a6	Support the Anima model. (#12012)	2026-01-21 19:44:28 -05:00
Markury	0fc15700be	Add LyCoris LoKr MLP layer support for Flux2 (#11997)	2026-01-20 23:18:33 -05:00
comfyanonymous	e755268e7b	Config for Qwen 3 0.6B model. (#11998)	2026-01-20 23:08:31 -05:00
Mylo	c4a14df9a3	Dynamically detect chroma radiance patch size (#11991)	2026-01-20 18:46:11 -05:00
Ivan Zorin	965d0ed509	fix: remove normalization of audio in LTX Mel spectrogram creation (#11990) ... For LTX Audio VAE, remove normalization of audio during MEL spectrogram creation. This aligs inference with training and prevents loud audio from being attenuated.	2026-01-20 18:44:28 -05:00
comfyanonymous	8ccc0c94fa	Make omni stuff work on regular z image for easier testing. (#11985)	2026-01-20 00:32:00 -05:00
comfyanonymous	2108167f9f	Support zimage omni base model. (#11979)	2026-01-19 23:17:38 -05:00
comfyanonymous	70c91b8248	Fix #11963 (#11982)	2026-01-19 22:32:40 -05:00
rkfg	0da5a0fe58	Convert mono audio to fake stereo for LTXV VAE encoding (#11965)	2026-01-19 22:12:02 -05:00