ComfyUI-EvTexture

ComfyUI-EvTexture
★ 18

视频超分EvTexture模型封装插帧兼容
将 EvTexture 视频超分模型封装为 ComfyUI 节点,便于加载放于 upscale_models 的预训练模型,对低分辨率帧进行细节放大并可与插帧模型联合使用。
💡 在 ComfyUI 中对低分辨率视频帧批量超分并配合插帧使用。
🍴 1 Forks💻 Python🔄 2025-01-05
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/2df45d172dc1
📦 requirements.txt
torch
torchvision>=0.9.0
workflow
📄 README

ComfyUI wrapper nodes for EvTexture

Get pre-trained models as described here and place them in upscale_models folder.

At the time the pre-trained models works better on low resolution images (64-256 px) packed with detail. The Vimeo90K version seems to work best (less artefacts) for larger images.

Example Workflow:

You can also combine with RIFE or other interpolation model to generate events at a higher FPS before feeding them into the upscaler.