rese1f / StableVideo

[ICCV 2023] StableVideo: Text-driven Consistency-aware Diffusion Video Editing

Home Page:https://rese1f.github.io/StableVideo/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Thanks for sharing, Would it run on Colab?

amrrs opened this issue · comments

I've not tried it yet but jut would love to hear if it runs on Free Colab (T4 - 16GB VRAM)

yes, if you additionally use xformer package, it will works

please see the updated README.md @amrrs :)

VRAM (MiB)
float32 29145
amp 23005
amp + cpu 17639
amp + cpu + xformers 14185
  • cpu: use cpu cache as ControlNet, args: save_memory

under default setting (e.g. resolution, etc.) in app.py

Thank you @rese1f Let me try if I can manage to run this on Colab :)

Sorry for reopening, I was trying to replicate this on a colab. I didn't understand what all changes to be made on the app.py for this configuration

amp + cpu + xformers

Do you have any suggestion?

Never mind, I faced some other dependency issues as well, trying to run on Colab. I'll drop the effort now. If you happen to create a colab, please update me here, I'll try to use that.

@amrrs I guess there is someone create a colab demo in the "pull request", but I haven't check that. (cause I've never use colab before

Sorry for reopening, I was trying to replicate this on a colab. I didn't understand what all changes to be made on the app.py for this configuration

amp + cpu + xformers

Do you have any suggestion?

and for this one, I have set this optimization as default. The only thing that need to do is to addtional install xformers by pip install xformers.

I had installed the xformers. I'll check the PR.