Thread Reader
Tweet

ComfyUI workflow for consistent characters in Seedance 2.0 is finally cracked Train once โ†’ generate infinitely Breaking down the exact workflow in the thread ๐Ÿ‘‡

First generate a character sheet image from an input image Make sure a single face in character sheet doesn't take more than 20% of the face, here is a reference attached
Next you need to use this reference as input instead of the direct character image in any of the Seedance 2.0 models. Works with both omni and non-omni models and reference the image in the prompt directly
Here are the generated videos
Anil Chandra Naidu Matcha
CTO @VadooAI โšก๏ธ Youtube : https://t.co/MRA3d9h6pv ๐ŸŒ AI apps : https://t.co/Qpxd3qhdOk
Follow on ๐•
Missing some tweets in this thread? Or failed to load images or videos? You can try to .