ExplainableML / ReNO

ReNO: Enhancing One-step Text-to-Image Models through Reward-based Noise Optimization
MIT License
64 stars 3 forks source link

Does ReNO work with SDXL inpainting model? #2

Open khurramkhalil opened 3 months ago

khurramkhalil commented 3 months ago

Hi there, Thank you for open sourcing your wonderful work. I am just curious, if ReNO would work with SDXL Inpainting pipeline? Thanks

lucaeyring commented 3 months ago

Hey, thanks for your interest! Reno is currently designed for one-step models and, as such, would work for any one-step inpainting model. Unfortunately, as far as we are aware, there currently does not exist a one-step inpainting pipeline based on e.g. SD-Turbo/SDXL-Turbo. For this, a distilled version of the SDXL inpainting model would need to be trained.