omerbt / TokenFlow

Official Pytorch Implementation for "TokenFlow: Consistent Diffusion Features for Consistent Video Editing" presenting "TokenFlow" (ICLR 2024)
https://diffusion-tokenflow.github.io
MIT License
1.58k stars 137 forks source link

Code Release #1

Open zcdliuwei opened 1 year ago

zcdliuwei commented 1 year ago

Hi, This work is very interesting! Are you planning to release the code? If yes, what is the approximate timeline for that? Additionally, what is the maximum allowed video length for this method? Thanks a lot!

omerbt commented 1 year ago

Thanks! There is no limit in video length. We're working on the release, stay tuned :)

Gitterman69 commented 1 year ago

Quick question @omerbt - how can i set the video lenght to more than 200 frames though? i created a snippet with 300 frames (10 sec) and cant seem to set it higher than 200.... any help would be greatly appreciated! :) THANKS FOR YOUR HARD WORK!!!!

image

Gitterman69 commented 1 year ago

Thanks! There is no limit in video length. We're working on the release, stay tuned :)

any way to set more than 200 frames? possible on 3090? whats the max lenght on 24gb of vram in general?

zhanghongyong123456 commented 1 year ago

Thanks! There is no limit in video length. We're working on the release, stay tuned :)

about long video,The video reconstruction Very poor, how should it be improved