Exact ComfiUI workflow for provided SD3M demo images?

#116
by fabulo - opened

I'm wondering why the demo images provided here https://huggingface.co/stabilityai/stable-diffusion-3-medium/tree/main/demo_images don't have EXIF data with the workflow as when using ComfyUI. I would like to reproduce the exact results, but when I create images with SD3M, I have the feeling that each result is far from the demo images in terms of quality. Is there a ComfyUI workflow to get the same ouput as a starting point?

I remember reading on Reddit that someone involved was claiming that those demos were made during development and wouldn't work now. It's a poor excuse and that's all that's been offered so far, poor excuses.

You're asking for a sane and rational explanation of how to make this work very well, and it's not being admitted to being a beta product rushed out the door by a company with poor communication skills.

Sign up or log in to comment