StableStudio
StableStudio copied to clipboard
Stability-AI
Embed StableStudio within Stable Diffusion WebUI
With so many extensions working for so many people with SD, I think making this a part of that, perhaps a new tab, or just being able to select the new larger model, would help many designers keep their prompts, LORAs, settings, and then use the new product as well.
Thanks
Hmmm haven't thought of embedding the StableStudio UI within stablediffusion-webui, but we do have an open issue for powering StableStudio using stablediffusion-webui's REST API: https://github.com/Stability-AI/StableStudio/issues/5
Is that what you're imagining?
Well, of course, I don't know all the features of StableStudioUI, but what I mean is that many of us have a large number of extensions for altering images, correcting things, allowing poses, Deforum, everything under the sun, you know what I mean?
If this new model was available in the same Web-UI, even if it had all the features but was a tab (just like img2img, or extras, etc) I could send a picture from the new model to inpainting, or use ControlNet on the new model's pic (hopefully, etc), not switch between programs... and if integrated correctly, extension makers might not have to redo their extensions for StableStudio.
Anyhow, just my 2 cents. I use Stable Diffusion for work so....
OR!
If somehow the new StableStudioUI could run all SD models (checkpoints/safetensors), embeddings, LORAs, and extensions like Deforum and ControlNet... like being backwards compatible.... this would also rock the house.
@HarrisTerry This would be possible by using A1111-Webui as a backend ... in any case the StableStudio plugin to talk to Webui would be a pre-requisite. A good example of this is the openOutpaint extension which uses the webui API + it can run in a Gradio tab.
Another API integration in the wild can be found on the AutoGPT Repository which implements the basic image generation functionality from the API but also adds support for DALLE-2 and Huggingface's (working) Models.
Just another 2 cents ;)
The approach of using stablediffusion-webui as the main GUI would most likely be the option with the lowest level of impact for users while still being able to use all the features and plugins the community already developed. I am also supporting the idea to add a tab in stablediffusion-webui to use the StableStudio as part of an image creation/processing pipeline.
So the plan is to get the functionality of automatic1111 into Stable Studio, but use the simpler Stable Studio UI to interact with it?
Hey all, someone has started on a plugin for webui!
Mind if I close this issue in favor of this one to centralize discussion?...
https://github.com/Stability-AI/StableStudio/issues/25
Actually I realize you're considering something totally separate, embedding StableStudio INTO stablediffusion-webui, so this really is a different discussion we should keep up.