support_VACE

This commit is contained in:
zehongma
2025-06-17 18:55:46 +08:00
parent 4bbd3f0484
commit dd0bb334ad
+8 -5
View File
@@ -7,8 +7,10 @@ MagCache has now been integrated into ComfyUI and is compatible with the ComfyUI
## 🔥 Latest News
- **If you like our project, please give us a star ⭐ on GitHub for the latest update.**
- [2025/6/16] 🔥 Support Chroma. Thanks @[kabachuha](https://github.com/kabachuha) for the codebase.
- [2025/6/10] 🔥 Support Wan2.1 T2V&I2V, HunyuanVideo T2V, FLUX-dev T2I
- [2025/6/17] 🔥 Support [Wan2.1-VACE-1.3B](https://github.com/ali-vilab/VACE), achieving a 1.7× acceleration.
- [2025/6/17] 🔥 MagCache is supported by [ComfyUI-WanVideoWrapper](https://github.com/kijai/ComfyUI-WanVideoWrapper). Thanks @[kijai](https://github.com/kijai).
- [2025/6/16] 🔥 Support [Chroma](https://huggingface.co/lodestones/Chroma). Thanks @[kabachuha](https://github.com/kabachuha) for the codebase.
- [2025/6/10] 🔥 Support [Wan2.1] T2V&I2V, [HunyuanVideo](https://github.com/Tencent/HunyuanVideo) T2V, [FLUX-dev]((https://github.com/black-forest-labs/flux)) T2I
## Installation
<!-- Installation via ComfyUI-Manager is preferred. Simply search for ComfyUI-MagCache in the list of nodes and click install.
@@ -41,12 +43,13 @@ To use MagCache node, simply add `MagCache` node to your workflow after `Load Di
| Wan2.1-T2V-14B | 0.24 | 0.2 | 6 |
| Wan2.1-I2V-480P-14B | 0.24 | 0.2 | 6 |
| Wan2.1-I2V-720P-14B | 0.24 | 0.2 | 6 |
| Wan2.1-VACE-1.3B | 0.02 | 0.2 | 3 |
</div>
If the image/video after applying MagCache is of low quality, please reduce **magcache_thresh** and **magcache_K**.
**If the image/video after applying MagCache is of low quality, please reduce `magcache_thresh` and `magcache_K`**. The default parameters are configured for extremely fast inference(2x-3x), which may lead to failures in some cases.
The demo workflows ([flux](./examples/flux.json), ([chroma](./examples/chroma.json)), [hunyuanvideo](./examples/hunyuanvideo.json), [wan2.1_t2v](./examples/wan2.1_t2v.json) and [wan2.1_i2v](./examples/wan2.1_i2v.json)) are placed in examples folder. The workflow [chroma_calibration](./examples/chroma_calibration.json) is used to calibrate the `mag_ratios` for `Chroma` when the number of inference steps differs from the pre-defined value.
The demo workflows ([flux](./examples/flux.json), ([chroma](./examples/chroma.json)), [hunyuanvideo](./examples/hunyuanvideo.json), [wan2.1_t2v](./examples/wan2.1_t2v.json), [wan2.1_i2v](./examples/wan2.1_i2v.json)), and [wan2.1_vace](./examples/wan2.1_vace.json) are placed in examples folder. The workflow [chroma_calibration](./examples/chroma_calibration.json) is used to calibrate the `mag_ratios` for `Chroma` when the number of inference steps differs from the pre-defined value.
**In our experiments, the videos generated by Wan2.1 are not as high-quality as those produced by the [original unquantized version](https://github.com/Wan-Video/Wan2.1).**
@@ -56,4 +59,4 @@ To use Compile Model node, simply add `Compile Model` node to your workflow afte
<!-- ![](./assets/compile.png) -->
## Acknowledgments
Thanks to [ComfyUI-TeaCache](https://github.com/welltop-cn/ComfyUI-TeaCache), [ComfyUI](https://github.com/comfyanonymous/ComfyUI), [ComfyUI-MagCache](https://github.com/wildminder/ComfyUI-MagCache), [MagCache](https://github.com/Zehong-Ma/MagCache/), [TeaCache](https://github.com/ali-vilab/TeaCache), [HunyuanVideo](https://github.com/Tencent/HunyuanVideo), [FLUX](https://github.com/black-forest-labs/flux), and [Wan2.1](https://github.com/Wan-Video/Wan2.1).
Thanks to [ComfyUI-TeaCache](https://github.com/welltop-cn/ComfyUI-TeaCache), [ComfyUI](https://github.com/comfyanonymous/ComfyUI), [ComfyUI-MagCache](https://github.com/wildminder/ComfyUI-MagCache), [MagCache](https://github.com/Zehong-Ma/MagCache/), [TeaCache](https://github.com/ali-vilab/TeaCache), [HunyuanVideo](https://github.com/Tencent/HunyuanVideo), [FLUX](https://github.com/black-forest-labs/flux), [Chroma](https://huggingface.co/lodestones/Chroma), and [Wan2.1](https://github.com/Wan-Video/Wan2.1).