The project uses KJ wan1.3B workflow, project address: https://github.com/kijai/ComfyUI WanVideoWrapper

You can control whether to enable blur processing to adjust the degree of redrawing,

Enable blur: significantly alters details of the source video, suitable for repairing hands, face, and other detailed repair scenarios.

Disable blur: refines details based on the source video, suitable for redrawing landscapes, distant figures, and other scenes.

You can combine enabling and disabling blur to complete processing, for example: enable blur the first time to repair face and hands, use the output video from the first time as input, disable blur the second time to further refine the details.

It is recommended to use AI-assisted generation for prompts to achieve better results. The description should closely match the source video, for example: a 25-year-old girl (main subject description), tall and slender, wearing a red long dress, the hem of the dress fluttering in the sea breeze (main subject dynamics), barefoot dancing on the beach (motion description). Her dance movements are elegant and rhythmic, her arms stretched out like a swan spreading its wings, her waist twisting to make the dress hem spin (motion details).

According to personal testing, this solution works better for videos with smaller motion ranges. It does not perform well for videos with excessive motion amplitude...

Creating content is not easy. If this helps you, please remember to bookmark, like, execute the application or workflow online. Your support is my motivation for creation. Thank you for your support!

If you have any questions, please leave a comment in the comment section. I will reply promptly when I see it. Feel free to communicate, learn together, and make progress together!