 YunjieYu
|
546145dca5
|
fix compatibility issues
|
2025-04-24 18:43:43 +08:00 |
|
|
YunjieYu
|
274c5a5bce
|
Compatible with skip layer guidance
|
2025-03-29 18:50:30 +08:00 |
|
|
YunjieYu
|
a6140dea8c
|
Support retention mode for Wan2.1 models
|
2025-03-26 18:07:58 +08:00 |
|
|
YunjieYu
|
f0c6c56778
|
fix hunyuanvideo i2v v2 bug
|
2025-03-21 19:13:34 +08:00 |
|
|
YunjieYu
|
4a70030ccb
|
fix bug for flux pulid
|
2025-03-18 19:24:31 +08:00 |
|
|
YunjieYu
|
e28a937aa2
|
optimize LTX-Video for low VRAM device; optimize VRAM for all models
|
2025-03-17 18:48:00 +08:00 |
|
|
YunjieYu
|
7d9fd8644c
|
optimize Wan2.1 for low VRAM device; optimize VRAM
|
2025-03-13 18:45:34 +08:00 |
|
|
YunjieYu
|
2e75a8b611
|
optimize VRAM
|
2025-03-12 19:39:57 +08:00 |
|
|
YunjieYu
|
4542f76ff9
|
optimize VARM
|
2025-03-12 19:30:33 +08:00 |
|
|
YunjieYu
|
1a11412dea
|
Add max_skip_steps option; Merge TeaCache nodes; Update TeaCache for HunyuanVideo and LTX-Video
|
2025-03-10 19:23:38 +08:00 |
|
|
YunjieYu
|
1a61fe3cec
|
fix TeaCacheForImgGen node bug
|
2025-03-07 11:34:14 +08:00 |
|
|
YunjieYu
|
71c535a082
|
support Wan2.1 series model
|
2025-03-06 21:17:28 +08:00 |
|
|
maedtb
|
3597bb87aa
|
Fix error when using SkyReels I2V Hunyuan Model
The number of out channels on the SkyReels Image-to-Video Hunyuan model is 32, compared to the normal 16. This change is the same fix applied to ComfyUI directly.
See: https://github.com/comfyanonymous/ComfyUI/commit/acc152b674fd1c983acc6efd8aedbeb380660c0c#diff-828ff060ac292e5b8eff646d924ade3720f1500a17b42ebdc3034c193ed4b597
|
2025-02-19 06:20:46 -05:00 |
|
|
YunjieYu
|
61636b6503
|
fix Compile Model node bug
|
2025-02-11 18:56:52 +08:00 |
|
|
TangYanxin
|
97d8ae3cb2
|
fix: 解决使用Compile Model 节点，LoRA失效的问题
|
2025-02-11 16:27:13 +08:00 |
|
|
YunjieYu
|
5b3d9f1b7a
|
Compile Model add eager and aot_eager backend.
|
2025-01-17 15:07:37 +08:00 |
|
|
YunjieYu
|
a40c99dc56
|
Support CogVideoX
|
2025-01-17 12:09:50 +08:00 |
|
|
YunjieYu
|
992a5b483c
|
Support PuLID-FLUX
|
2025-01-15 12:34:55 +08:00 |
|
|
TangYanxin
|
c4fa3e6460
|
feat: If rel_l1_thresh==0 doe‘t execute TeaCache logic
|
2025-01-15 09:56:34 +08:00 |
|
|
TangYanxin
|
be41e9ad9f
|
Fixed: When there are multiple TeaCache nodes in a workflow, the rel_l1_thresh value is always the value of the last TeaCache node
|
2025-01-15 09:22:28 +08:00 |
|
|
TangYanxin
|
a0750c4e6f
|
support flux pulid
|
2025-01-14 22:41:36 +08:00 |
|
|
唐焱鑫
|
34f0830162
|
Merge branch 'main' into main
|
2025-01-14 14:51:04 +08:00 |
|
|
YunjieYu
|
722c373a50
|
Support Compile Model
|
2025-01-14 13:05:06 +08:00 |
|
|
TangYanxin
|
296642b74c
|
refactor: Remove steps from the node settings，fix RuntimeError “The size of tensor a must match the size of tensor b”
- 移除节点中的steps配置项
- 移除了不必要的计数器变量 (cnt)
- 优化了累积相对 L1 距离的计算和重置逻辑
- 修复工作流中存在多个步数不一样的采样器时报错的问题
|
2025-01-13 20:12:27 +08:00 |
|
|
YunjieYu
|
bd459ec5d1
|
main
|
2025-01-10 19:35:00 +08:00 |
|
|
YunjieYu
|
f8783a9f79
|
main
|
2025-01-09 19:40:13 +08:00 |
|
|
YunjieYu
|
35231722cb
|
first commit
|
2025-01-07 17:06:20 +08:00 |
|