| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
Update sheetsage2 abc creation from upstream yue code. (#16569) | 12 天前 | |
Make a context manager for cast_bias_weight and use it. (#14750) | 1 个月前 | |
Add better error message for common error. (#10846) | 10 个月前 | |
Remove useless annotations imports. (#14105) | 4 个月前 | |
Uni pc sampler now works with audio and video models. | 1 年前 | |
feat: Support Sam3d-body (CORE-35) (#14370) | 1 个月前 | |
Add cfgpp_ud10_ab sampler. (#15951) This is the result of an experiment where I asked a LLM to create a better sampler for the base Anima model and gave it a lot of different things to try. I wouldn't call this revolutionary but from my basic testing it seems to be a bit better than the other samplers at lower steps so I have decided to include it in ComfyUI. For best result use it at CFG 2.0 on base Anima. It should also work on other models but I'm not sure about the optimal parameters. | 1 个月前 | |
Change model code to use function wrappers for Linear calls (#16861) Amp-Thread-ID: https://ampcode.com/threads/T-01a1185e-e6c2-7558-b534-dd3992e43bcd Co-authored-by: Amp <amp@ampcode.com> | 11 小时前 | |
Silence clip tokenizer warning. (#8934) | 1 年前 | |
Controlnet refactor. | 2 年前 | |
Support tiny VAE for Qwen-Image 2.1 (#16552) | 11 天前 | |
Change model code to use function wrappers for Linear calls (#16861) Amp-Thread-ID: https://ampcode.com/threads/T-01a1185e-e6c2-7558-b534-dd3992e43bcd Co-authored-by: Amp <amp@ampcode.com> | 11 小时前 | |
Fix issue with bypass loras and fused paths. (#16862) | 10 小时前 | |
Remove torchaudio dependency. (#16457) | 16 天前 | |
Fast disk detection to all model loaders. (#16425) | 16 天前 | |
Add governance enforcement for custom nodes (#16167) | 1 天前 | |
Fix potential issue with non clip text embeddings. | 2 年前 | |
Fix some clip vision regression. (#15506) | 1 个月前 | |
Fast disk detection to all model loaders. (#16425) | 16 天前 | |
Add support for clip g vision model to CLIPVisionLoader. | 3 年前 | |
Add support for unCLIP SD2.x models. See _for_testing/unclip in the UI for the new nodes. unCLIPCheckpointLoader is used to load them. unCLIPConditioning is used to add the image cond and takes as input a CLIPVisionEncode output which has been moved to the conditioning section. | 3 年前 | |
Add support for unCLIP SD2.x models. See _for_testing/unclip in the UI for the new nodes. unCLIPCheckpointLoader is used to load them. unCLIPConditioning is used to add the image cond and takes as input a CLIPVisionEncode output which has been moved to the conditioning section. | 3 年前 | |
support clip-vit-large-patch14-336 (#4042) * support clip-vit-large-patch14-336 * support clip-vit-large-patch14-336 | 2 年前 | |
Support llava clip vision model. | 1 年前 | |
Support the siglip 2 naflex model as a clip vision model. (#11831) Not useful yet. | 8 个月前 | |
Support new flux model variants. | 1 年前 | |
Support 512 siglip model. | 1 年前 | |
feat: allow --comfy-api-base target ephemeral testenvs (#14569) * feat: allow --comfy-api-base target ephemeral testenvs Signed-off-by: bigcat88 <bigcat88@icloud.com> * refactor: name /features data as backend flags, not frontend --------- Signed-off-by: bigcat88 <bigcat88@icloud.com> Co-authored-by: guill <jacob.e.segal@gmail.com> | 2 个月前 | |
Cleanups to the last PR. (#12646) | 7 个月前 | |
feat: Context Windows sampling with LTX2 models and IC-LoRa guides (CORE-3) (#13325) | 3 个月前 | |
Support Qwen-Image 2.1 union fun controlnet (#16519) | 9 天前 | |
Add deploy environment header (Comfy-Env) to partner node API calls (#13425) | 5 个月前 | |
Remove useless code. | 1 年前 | |
load_unet -> load_diffusion_model with a model_options argument. | 2 年前 | |
float: use CK stochastic rounding cuda kernel (#13971) | 4 个月前 | |
Remove some useless code. (#8812) | 1 年前 | |
Fix typos (#10986) | 4 个月前 | |
rename comfy/logging.py to comfy/internal_logging.py (#15231) This avoids name collision (circular imports) for external custom nodes, for which the comfy path is pushed into sys.path so Python's own logging module is shadowed otherwise. fixes: #15229 | 2 个月前 | |
Support tiny VAE for Qwen-Image 2.1 (#16552) | 11 天前 | |
feat: Qwen-image 2.1 support (CORE-423) (#16400) | 18 天前 | |
Use torch RMSNorm for flux models and refactor hunyuan video code. (#12432) | 7 个月前 | |
Threaded Loader performance fixes / improvements (+ Aimdo 0.4.6) (#14116) * memory_management: Add direct to read GPU mode Make destination optional (or make it optionally GPU) and use aimdo to file_read direct to GPU. * ops: Remove stream pin buffers and use aimdo reads This consumed too much RAM and its better to just take the hit on the CPU syncing back the stream on a short ring buffer. Aimdo implements this so just rip the stream pin buffer from comfy. * model_management: all active pin registration movement Its better to just let the active model load past the pin limit as pins and let the pins move around. The saves the HDD and SATA people disk traffic while only costing a few GPU syncs. * utils: use aimdo file handle This opens on windows with more favourable flags * mp: only count the model proper for loaded_ram and vram Exclude live loras from the numbers to avoid the case where the reported loaded memory exceeds the size of the model. This causes me confusion in the Kijai visualizer when it looked fully loaded but was hitting disk due to this accounding disrepency. * utils: add bit reverse utility useful for max scattering something ordered. * pinned_memory: Implement offload balancing Use a max scatter alogorithm to prioritize pins of the same size such that when doing a little bit of offloading it gets scattered, allowing the prefetcher to more evenly swollow the offload. * comfy-aimdo 0.4.7 Aimdo 0.4.7 implement VRAM buffer exhaustion predection to avoid early speculative load of weights that definately wont fix once the inference gets further in. * model-prefetch: consolidate pin ensures on the sync point This could happen mid prefetch block, cause a sync of the entire block and lose overlap. Get ahead of the problem with a free down at the natural compute stream sync point. * mm: Put a 2GB min on the pin ceiling This is reasonably bad if it starts causing swap pressure, moreso than during normal ram-cache proceedings. Clamp it. * add --fast-disk | 4 个月前 | |
feat: ming-image support (#16482) | 14 天前 | |
Support ID-V2V (#15139) | 11 天前 | |
Disable pinned memory automatically on integrated GPUs (#16746) * Disable pinned memory automatically on AMD APUs APU VRAM is carved out of system RAM, so pinning host memory only takes RAM away from the GPU. Detect AMD integrated GPUs and treat them as if --disable-pinned-memory was passed. dGPUs and the flag are unchanged. * mm: generalize is_integrated pin disable check to all cuda --------- Co-authored-by: tvukovic-amd <tvukovic@amd.com> | 5 天前 | |
Fast disk detection to all model loaders. (#16425) | 16 天前 | |
Qwen3/3.5/3.8 cudagraphs and w4a8 gemv support (CORE-390) (#15623) | 20 天前 | |
feat: Marigold v2 support (CORE-431) (#16232) | 26 天前 | |
Comfy Aimdo 0.5.3 + Memory compiler fixes (#16180) | 29 天前 | |
Make it easier to debug nested tensors. (#15383) | 1 个月前 | |
Fix issue with bypass loras and fused paths. (#16862) | 10 小时前 | |
Only parse command line args when main.py is called. | 3 年前 | |
Remove useless annotations imports. (#14105) | 4 个月前 | |
Disable pinned memory automatically on integrated GPUs (#16746) * Disable pinned memory automatically on AMD APUs APU VRAM is carved out of system RAM, so pinning host memory only takes RAM away from the GPU. Detect AMD integrated GPUs and treat them as if --disable-pinned-memory was passed. dGPUs and the flag are unchanged. * mm: generalize is_integrated pin disable check to all cuda --------- Co-authored-by: tvukovic-amd <tvukovic@amd.com> | 5 天前 | |
Changes to the previous radiance commit. (#9851) | 1 年前 | |
Support w6a8 quantization format (#16483) | 9 天前 | |
feat: Gemma4 text generation support (CORE-30) (#13376) * initial gemma4 support * parity with reference implementation outputs can 100% match transformers with same sdpa flags, checkpoint this and then optimize * Cleanup, video fixes * cleanup, enable fused rms norm by default * update comment * Cleanup * Update sd.py * Various fixes * Add fp8 scaled embedding support * small fixes * Translate think tokens * Fix image encoder attention mask type So it works with basic attention * Handle thinking tokens different only for Gemma4 * Code cleanup * Update nodes_textgen.py * Use embed scale class instead of buffer Slight difference to HF, but technically more accurate and simpler code * Default to fused rms_norm * Update gemma4.py | 5 个月前 | |
Allow regular single image Empty Latent Image node to be used with H3. (#15677) | 1 个月前 | |
MultiGPU Work Units For Accelerated Sampling (CORE-184) (#7063) | 4 个月前 | |
Add cfgpp_ud10_ab sampler. (#15951) This is the result of an experiment where I asked a LLM to create a better sampler for the base Anima model and gave it a lot of different things to try. I wouldn't call this revolutionary but from my basic testing it seems to be a bit better than the other samplers at lower steps so I have decided to include it in ComfyUI. For best result use it at CFG 2.0 on base Anima. It should also work on other models but I'm not sure about the optimal parameters. | 1 个月前 | |
Support LynnReal light Minimax-H3 vae (#16657) | 8 天前 | |
Improve low VRAM performance of qwen 3.5 LLM (#16820) | 1 天前 | |
Fix potential issue with non clip text embeddings. | 2 年前 | |
Add a T5TokenizerOptions node to set options for the T5 tokenizer. (#7803) | 1 年前 | |
Add --disable-fast-disk to disable fast disk. (#16424) | 18 天前 | |
Support ID-V2V (#15139) | 11 天前 | |
Add SeedVR2 support (CORE-6) (#14424) | 2 个月前 | |
fix(memory): respect the container cgroup memory limit instead of host RAM (#15927) | 1 个月前 | |
Aimdo 0.5.5 + Auto-detect and enable --fast-disk when the disk is fast (CORE-440) (#16333) | 22 天前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 12 天前 | ||
| 1 个月前 | ||
| 10 个月前 | ||
| 4 个月前 | ||
| 1 年前 | ||
| 1 个月前 | ||
| 1 个月前 | ||
| 11 小时前 | ||
| 1 年前 | ||
| 2 年前 | ||
| 11 天前 | ||
| 11 小时前 | ||
| 10 小时前 | ||
| 16 天前 | ||
| 16 天前 | ||
| 1 天前 | ||
| 2 年前 | ||
| 1 个月前 | ||
| 16 天前 | ||
| 3 年前 | ||
| 3 年前 | ||
| 3 年前 | ||
| 2 年前 | ||
| 1 年前 | ||
| 8 个月前 | ||
| 1 年前 | ||
| 1 年前 | ||
| 2 个月前 | ||
| 7 个月前 | ||
| 3 个月前 | ||
| 9 天前 | ||
| 5 个月前 | ||
| 1 年前 | ||
| 2 年前 | ||
| 4 个月前 | ||
| 1 年前 | ||
| 4 个月前 | ||
| 2 个月前 | ||
| 11 天前 | ||
| 18 天前 | ||
| 7 个月前 | ||
| 4 个月前 | ||
| 14 天前 | ||
| 11 天前 | ||
| 5 天前 | ||
| 16 天前 | ||
| 20 天前 | ||
| 26 天前 | ||
| 29 天前 | ||
| 1 个月前 | ||
| 10 小时前 | ||
| 3 年前 | ||
| 4 个月前 | ||
| 5 天前 | ||
| 1 年前 | ||
| 9 天前 | ||
| 5 个月前 | ||
| 1 个月前 | ||
| 4 个月前 | ||
| 1 个月前 | ||
| 8 天前 | ||
| 1 天前 | ||
| 2 年前 | ||
| 1 年前 | ||
| 18 天前 | ||
| 11 天前 | ||
| 2 个月前 | ||
| 1 个月前 | ||
| 22 天前 |