COVE: Unleashing the Diffusion Feature Correspondence for Consistent Video Editing

Gao Huang, Xiu Li, Jiangshan Wang, Yue Ma, Jiayi Guo, Yicheng Xiao

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. If it links a repository, it is listed below.

Repositories linked to this paper

wangjiangshan0725/cove — marked official in Papers with Code; named in the paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Snow leopard A Barbie Doll with blonde hair The Lion King and his children Sand sculptures by the sea Cartoon style Figure 1: We propose COrrespondence-guided Video Editing (COVE), which leverages the correspondence information of the diffusion feature to achieve consistent and high-quality video editing. Our method is capable of generating high-quality edited videos with various kinds of prompts (style, category, background, etc.) while effectively preserving temporal consistency in generated videos.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2406.08850")
get_code_for_paper("2406.08850")
have("2406.08850")

Connect an agent — have() is free.