TensorX
返回文献探索

Paper · arXiv 2403.12706

AnimateDiff-Lightning: Cross-Model Diffusion Distillation

Shanchuan Lin, Xiao Yang

18 upvotesMarch 19, 2024arXiv 预印本
AI 摘要

AnimateDiff-Lightning uses progressive adversarial diffusion distillation for efficient video generation, enhancing style compatibility through joint distillation of multiple diffusion models.

progressive adversarial diffusiondistillationfew-step video generationmotion module

Abstract

We present AnimateDiff-Lightning for lightning-fast video generation. Our model uses progressive adversarial diffusion distillation to achieve new state-of-the-art in few-step video generation. We discuss our modifications to adapt it for the video modality. Furthermore, we propose to simultaneously distill the probability flow of multiple base diffusion models, resulting in a single distilled motion module with broader style compatibility. We are pleased to release our distilled AnimateDiff-Lightning model for the community's use.

北京市昌平区探索星信息技术及软件开发工作室

京ICP备2026059466号
AnimateDiff-Lightning: Cross-Model Diffusion Distillation | TensorX