idea:Can this project work with the current stable diffusion WebUI program as an extension like inpaint tools?

[ApprenticeF]

Diffusion models cannot assign specific tags to a certain concept. If SceneComposor supports the custom prompt tokens that are trained from textual inversion or Dreambooth, that must be fantastic.

[zengxianyu]

the model needs to be trained so we probably cannot work with the current stable diffusion WebUI. However, we are planning to release our own WebUI after the paper gets published