omerbt/TokenFlow
Official Pytorch Implementation for "TokenFlow: Consistent Diffusion Features for Consistent Video Editing" presenting "TokenFlow" (ICLR 2024)
Enforces temporal consistency by propagating diffusion feature tokens across frames using inter-frame correspondences, eliminating the need for video-specific training. Works composably with existing image editing methods (Plug-and-Play, ControlNet, SDEdit) through a two-stage pipeline: DDIM inversion for reconstruction, then feature-guided editing that preserves spatial layout and dynamics.
1,707 stars. No commits in the last 6 months.
Stars
1,707
Forks
143
Language
Python
License
MIT
Category
Last pushed
Feb 03, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/omerbt/TokenFlow"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
hao-ai-lab/FastVideo
A unified inference and post-training framework for accelerated video generation.
thu-ml/TurboDiffusion
TurboDiffusion: 100–200× Acceleration for Video Diffusion Models
ModelTC/LightX2V
Light Image Video Generation Inference Framework
PKU-YuanGroup/Helios
Helios: Real Real-Time Long Video Generation Model
PKU-YuanGroup/MagicTime
[TPAMI 2025🔥] MagicTime: Time-lapse Video Generation Models as Metamorphic Simulators