Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

DeepCache with image2image and img2img ControlNet for SDXL/SD1.5 #1181

Open
WyattAutomation opened this issue Jan 10, 2025 · 0 comments
Open

Comments

@WyattAutomation
Copy link

WyattAutomation commented Jan 10, 2025

🚀 The feature, motivation and pitch

Requested Feature: enable DeepCache acceleration in onedfiff for use with img2img and img2imgControlNet diffusers pipelines.

Alternatives

There are many other optimizations I use in my pipeline already, I just need DeepCache to work with OneDiff

Additional context

The performace boost according to your documentation from DeepCache+OneDiff is great enough that it would be a huge benefit to have it fully functional for img2img pipelines (at least). ControlNet and img2img ControlNet pipelines would also be extremely useful as a lot of realtime applications make heavy use of these for controlling the output of realtime SDXL and SD 1.5 rendering.

The below video demonstrates one such usecase: Multiple ControlNet outputs from NDI from Unity3D into OneDiff accelerated diffusers pipelines for interactive/realtime WASD/mouse control

Without DeepCache the speed is somewhat OK at 2 steps on a single rtx3090, but to be able to add just 1 or 2 extra denoising steps and retain this same speed per the performance boost added by DeepCache, would result in far more viable rendering quality for use in a realtime ControlNet+Agent based WASD-controlled videogame.

367620057-03de1e1f-5d9f-49c4-ade1-58f9743135a1.mp4
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant