Echo-Forcing: A Scene Memory Framework for Interactive Long Video Generation

2026-05-15

AI Summary

This work addresses the limitations of current training-free long video generation methods in interactive scenarios, where prompt switching, catastrophic forgetting of prior scenes, and failure to recall historical content remain challenging—issues rooted in the functional coupling of historical key-value (KV) cache representations. To overcome this, we propose Echo-Forcing, a novel framework that decouples the KV cache into three distinct components: stable anchors, compressed history, and recent dynamics. Our approach integrates hierarchical temporal memory with relative RoPE-based positional encoding, spatially structured scene recall frames, and a difference-aware memory decay mechanism. Within a fixed cache budget, Echo-Forcing enables smooth transitions, hard cuts, and long-range memory retrieval. Evaluated on the VBench-Long benchmark, our method significantly outperforms existing approaches, achieving state-of-the-art performance across both long-form and interactive video generation tasks.

Abstract

Autoregressive video diffusion models enable open-ended generation through local attention and KV caching. However, existing training-free long-video optimization methods mainly focus on stable extension under a single prompt, making them difficult to handle interactive scenarios involving prompt switching, old scene forgetting, and historical scene recall. We identify the core bottleneck as the functional entanglement of historical KV states: stable anchors and recent dynamics are handled by the same cache policy, leading to outdated background contamination, delayed response to new prompts, and loss of long-range memory. To address this issue, we propose Echo-Forcing, a training-free scene memory framework specifically designed for interactive long video generation with three core mechanisms: (1) Hierarchical Temporal Memory, which decouples stable anchors, compressed history, and recent windows under relative RoPE; (2) Scene Recall Frames, which compresses historical scenes into spatially structured KV representations to support long-term recall; and (3) Difference-aware Memory Decay, which adaptively forgets conflicting tokens according to the discrepancy between old and new scenes. Based on these designs, Echo-Forcing uniformly supports smooth transitions, hard cuts, and long-range scene recall under a bounded cache budget. Extensive evaluations on VBench-Long further demonstrate that Echo-Forcing achieves the best overall performance in both long-video generation and interactive video generation settings. Our code is released in https://github.com/mingqiangWu/Echo-Forcing

Problem

• interactive video generation
• long video generation
• scene memory
• prompt switching
• KV caching

Innovation

• Echo-Forcing
• scene memory
• interactive video generation
• KV caching
• long video diffusion