Innovate Futures @ Benji
patreon
[For Patreon Supporters] [Workflow] Wan Video V2V - InfiniteTalk - Long Content Talking Avatar Generator (Single) (Ver.20250823)
🕑 Added 2025-08-25 13:00:21 +0000 UTCRelated Post : https://www.patreon.com/posts/137286679
Tutorial Video : https://youtu.be/CA-CQo_Q198
If you missed part 1, how to get started, go here: https://youtu.be/6MS-KAnvTBg
Some InfiniteTalk Video2video examples here : https://www.patreon.com/posts/137187389
*** Remember, update the Wan Wrapper node to the latest version
For InfiniteTalk img2video,
Wan I2V - InfiniteTalk - Long Content Talking Avatar Generator (Single) (Ver.20250822): https://www.patreon.com/posts/137035392
MeiGen-AI/InfiniteTalk
https://github.com/MeiGen-AI/InfiniteTalk
InfiniteTalk for ComfyUI
https://huggingface.co/MeiGen-AI/InfiniteTalk/tree/main/comfyui
ComfyUI_ChatterBox_SRT_Voice
https://github.com/diodiogod/ComfyUI_ChatterBox_SRT_Voice
How to setup Ollama in Windows: https://youtu.be/3t_P0tDvRCE
*********** For Audio Input Config setting ***********
If you want to use your own audio file as input
Step 1 - Bypass group like this :
Step 2 - Set Group to Never for the TTS group
Then it will not having issue with the SET/GET node ID error.
Comments
Benjamin
have you try lower the LoRA strength? if going for 30+ sec to be safe, I will set it to 0.6 or 0.5. and also if the reference video moving another angles, it might cause that as well.
C Lindesay
I am using Ubuntu Studio 24.04 comfyui in an isolated environment using a rtx 5090 with sageattn all working. I’ve tried many of these longer generation workflows which work fine up to about 12 seconds. Then they all gradually distort with blotches. I’ve attempted many combinations of settings with no success yet. Just thought I’d ask if you have any suggestions?
Melting Province
Infinite fun ! So good ! Thanks !
Benjamin Law
Mentioned in the tutorial video, i added voice via audio file
Soonwon Baek
Instead of extracting content from audio to generate a video, is it possible to directly create a video from the audio itself? I tried using the default InfiniteTalk workflow by feeding it an angry-tone audio line, but the resulting video didn’t show the character expressing anger at all. What I actually want is to first create a video where the character clearly looks angry, then add the angry-tone audio on top, and finally run V2V to sync them together.
