Stable diffusion

seen the post from @tonfilm and wanted to point out that a major possible node to connect to the #stablediffusion scene might be comfyUI

1 Like

I think the big news is that Tebjan has created a powerful tool to interact with Python. So it will probably be possible to connect ComfyUI not only at API level, which can be done without much difficulty with knowledge of JSON, but also via custom Python ComfyUI nodes or even inside a patch.

I’m very interested, will there be Beta testing of the Python-Gamma plugin?

1 Like

Yes, @yar is correct. you can probably run anything that’s written in Python with it. Marshaling the objects directly is much faster than using the web API that needs to serialize/deserialize everything.

However, you can only have one Python interpreter per process, so that means only one venv per vvvv.exe. If you want to run StreamDiffusion, LLaMA, and ComfyUI workflows together, you need to unify their dependencies in one venv or start multiple vvvv instances.


custom Python ComfyUI nodes

there is already a lot of control in ComfyUI in regard to sd

could these advancements apply in a pipeline that makes use of existing platforms and their funcitonality (such as ComfyUI and its many addons), or is it required to go at a lower level where tools have to be implemented from scratch ?

There is now an official thread about the project: