I suggest you generate image and use inpaint to make different face expressions for each character.
Resolution for txt2img: 512x768/768x512
Preprocessor: none Model: segmentation.
Control mode: Test which one works better. Usually "My prompt is more important" works best.
Description
FAQ
Comments (4)
how i can use this please
You need to install ControlNet (https://www.youtube.com/watch?v=WZg3e6B2yPQ&t=16s&ab_channel=SebastianKamph), enable it in webui, set preprocessor to none, set model to segmentation and drop image with seg into it (the one with red silhouettes). Then you just need to write prompt and use sd as you normally do. For better results use img2img and inpainting.
¿Cómo instalar en fooocus?
thanks for uploading this but how do I apply this to my comfyui? is it work something work with controlnet?