# Accelerating Diffusion-based Video Editing via Heterogeneous Caching: Beyond Full Computing at Sampled Denoising Timestep

arXiv: https://arxiv.org/abs/2603.24260 · HTML: https://mlsanity.com/paper/2603.24260/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Video generation, one of 14 directions in Diffusion models.
- Builds on 34 of its 36 references that are in the map.
- First posted on arXiv 25 Mar 2026.

## Placed in

- Diffusion models / [Video generation](https://mlsanity.com/diffusion/video-generation/)

## Builds on (most cited first, 30 of 34)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Scalable Diffusion Models with Transformers](https://mlsanity.com/paper/2212.09748/) (2022)
- [Improved Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2102.09672/) (2021)
- [Wan: Open and Advanced Large-Scale Video Generative Models](https://mlsanity.com/paper/2503.20314/) (2025)
- [Video Diffusion Models](https://mlsanity.com/paper/2204.03458/) (2022)
- [VBench: Comprehensive Benchmark Suite for Video Generative Models](https://mlsanity.com/paper/2311.17982/) (2023)
- [Exploring CLIP for Assessing the Look and Feel of Images](https://mlsanity.com/paper/2207.12396/) (2022)
- [The 2017 DAVIS Challenge on Video Object Segmentation](https://mlsanity.com/paper/1704.00675/) (2017)
- [On Distillation of Guided Diffusion Models](https://mlsanity.com/paper/2210.03142/) (2022)
- [LTX-Video: Realtime Video Latent Diffusion](https://mlsanity.com/paper/2501.00103/) (2024)
- [Latte: Latent Diffusion Transformer for Video Generation](https://mlsanity.com/paper/2401.03048/) (2024)
- [VACE: All-in-One Video Creation and Editing](https://mlsanity.com/paper/2503.07598/) (2025)
- [SmartBrush: Text and Shape Guided Object Inpainting with Diffusion Model](https://mlsanity.com/paper/2212.05034/) (2022)
- [Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model](https://mlsanity.com/paper/2411.19108/) (2024)
- [DepGraph: Towards Any Structural Pruning](https://mlsanity.com/paper/2301.12900/) (2023)
- [Structural Pruning for Diffusion Models](https://mlsanity.com/paper/2305.10924/) (2023)
- [Accelerating Diffusion Transformers with Token-wise Feature Caching](https://mlsanity.com/paper/2410.05317/) (2024)
- [Cache Me if You Can: Accelerating Diffusion Models through Block Caching](https://mlsanity.com/paper/2312.03209/) (2023)
- [Real-Time Video Generation with Pyramid Attention Broadcast](https://mlsanity.com/paper/2408.12588/) (2024)
- [Token Fusion: Bridging the Gap between Token Pruning and Token Merging](https://mlsanity.com/paper/2312.01026/) (2023)
- [Towards An End-to-End Framework for Flow-Guided Video Inpainting](https://mlsanity.com/paper/2204.02663/) (2022)
- [VideoPainter: Any-length Video Inpainting and Editing with Plug-and-Play Context Control](https://mlsanity.com/paper/2503.05639/) (2025)
- [GenTron: Diffusion Transformers for Image and Video Generation](https://mlsanity.com/paper/2312.04557/) (2023)
- [FlowVid: Taming Imperfect Optical Flows for Consistent Video-to-Video Synthesis](https://mlsanity.com/paper/2312.17681/) (2023)
- [AutoDiffusion: Training-Free Optimization of Time Steps and Architectures for Automated Diffusion Model Acceleration](https://mlsanity.com/paper/2309.10438/) (2023)
- [InstructEdit: Improving Automatic Masks for Diffusion-based Image Editing With User Instructions](https://mlsanity.com/paper/2305.18047/) (2023)
- [Towards Accurate Post-training Quantization for Diffusion Models](https://mlsanity.com/paper/2305.18723/) (2023)
- [Unified Discrete Diffusion for Simultaneous Vision-Language Generation](https://mlsanity.com/paper/2211.14842/) (2022)
- [FastCache: Fast Caching for Diffusion Transformer Through Learnable Linear Approximation](https://mlsanity.com/paper/2505.20353/) (2025)
- [Lazy Diffusion Transformer for Interactive Image Editing](https://mlsanity.com/paper/2404.12382/) (2024)

## Built on by (most cited first, 3 of 3)

- [Mixture-of-Modality-Experts with Holistic Token Learning for Fine-Grained Multimodal Visual Analytics in Driver Action Recognition](https://mlsanity.com/paper/2604.05947/) (2026)
- [Not All Tokens Need 40 Steps: Heterogeneous Step Allocation in Diffusion Transformers for Efficient Video Generation](https://mlsanity.com/paper/2605.06892/) (2026)
- [Reference-Efficient Multimodal Bitstream-Corrupted Video Recovery via Depth-Conditioned Adaptation](https://mlsanity.com/paper/2610.15806/) (2026)

## Abstract (arXiv)

Diffusion-based video editing has emerged as an important paradigm for high-quality and flexible content generation. However, despite their generality and strong modeling capacity, Diffusion Transformers (DiT) remain computationally expensive due to the iterative denoising process, posing challenges for practical deployment. Existing video diffusion acceleration methods primarily exploit denoising timestep-level feature reuse, which mitigates the redundancy in denoising process, but overlooks the architectural redundancy within the DiT that many attention operations over spatio-temporal tokens are redundantly executed, offering little to no incremental contribution to the model output. This work introduces HetCache, a training-free diffusion acceleration framework designed to exploit the inherent heterogeneity in diffusion-based masked video-to-video (MV2V) generation and editing. Instead of uniformly reuse or randomly sampling tokens, HetCache assesses the contextual relevance and interaction strength among various types of tokens in designated computing steps. Guided by spatial priors, it divides the spatial-temporal tokens in DiT model into context and generative tokens, and selectively caches the context tokens that exhibit the strongest correlation and most representative semantics with generative ones. This strategy reduces redundant attention operations while maintaining editing consistency and fidelity. Experiments show that HetCache achieves a noticeable acceleration, including a 2.67× latency speedup and FLOPs reduction over commonly used foundation models, with negligible degradation in editing quality.