search

radames / Real-Time-Latent-Consistency-Model

App showcasing multiple real-time diffusion models pipelines with Diffusers
https://huggingface.co/spaces/radames/Real-Time-Latent-Consistency-Model
Apache License 2.0
873 stars 102 forks source link

Model Speed Up #35

Closed acs107102 closed 10 months ago

acs107102 commented 11 months ago

Hello, I'm using controlnetSDXLTurbo to do some project. My machine is RTX 3090.

I'm wondering how can I speed up the result or do you have any recommend model can do that, thanks.

radames commented 11 months ago

hi @acs107102 , I'll add some performance improvements soon, meanwhile, could you try img2imgSD21Turbo.py which is the fastest

acs107102 commented 11 months ago

I'm sorry, I can't find the file you mentioned above.

Btw, I also want to have high-quality video in live stream. Can this model do that?

radames commented 10 months ago

hi @acs107102 , please check now, you have the option to use --onediff or --sfast, it speeds up the inference a bit! please make sure to pul and reinstall requirements.txt