← sources

Custom Nodes for LongCat Video Avatar 1.6

source: reddit (reddit:StableDiffusion)published 2026-11-19 09:16ingested 2026-11-20 02:37

retained text

I ported LongCat Video Avatar 1.6 to ComfyUI. Audio-driven talking video from a start image, a prompt and speech.  
  
It's quite slow depending on the resolution you choose. About 15 s per second of generated video at 832×464 and 59 s at 1280×720 on a 4090.  
  
For many use cases a small talking-head frame does the job.

Nodes and workflow: [https://github.com/PedroMarinhoDev/ComfyUI-LongCat-Avatar](https://github.com/PedroMarinhoDev/ComfyUI-LongCat-Avatar)  
Weights: [https://huggingface.co/PedroMarinhoDev/LongCat-Video-Avatar-1.6-ComfyUI](https://huggingface.co/PedroMarinhoDev/LongCat-Video-Avatar-1.6-ComfyUI)



document fetched 2026-11-20 00:30 · canonical https://reddit.com/r/StableDiffusion/comments/1y051xn/custom_nodes_for_longcat_video_avatar_16