Research map / Video / Video generation
Anchor Forcing: Anchor Memory and Tri-Region RoPE for Interactive Streaming Video Diffusion
yang yang, Tianyi Zhang, Wei Huang, Jinwei Chen, Boxi Wu, Xiaofei He, Deng Cai, Bo Li, Peng-Tao Jiang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add tri-region RoPE (ablation) | LongLive baseline · 60s interactive ablation | dynamic degree: 26.37 -> 74.83 | Improved |
| add tri-region RoPE (ablation) | LongLive baseline · 60s interactive ablation | quality score: 78.92 -> 81.72 | Improved |
| add tri-region RoPE (ablation) | LongLive baseline · 60s interactive ablation | CLIP score: 23.58 -> 23.71 | Improved |
| add anchor-guided re-cache on top of tri-region RoPE (ablation) | Anchor Forcing · 60s interactive ablation | quality score: 81.72 -> 82.25 | Improved |
| add anchor-guided re-cache on top of tri-region RoPE (ablation) | Anchor Forcing · 60s interactive ablation | CLIP score: 23.71 -> 23.85 | Improved |
| add anchor-guided re-cache on top of tri-region RoPE (ablation) | Anchor Forcing · 60s interactive ablation | dynamic degree: 74.83 -> 73.00 | Hurt |
| use anchor-guided re-cache instead of KV re-cache (ablation) | LongLive baseline · 60s interactive ablation | dynamic degree: 26.37 -> 50.73 | Improved |
| use Flush re-cache instead of KV re-cache (ablation) | LongLive baseline · 60s interactive ablation | dynamic degree: 26.37 -> 27.60 | No clear effect |
| use bounded RoPE instead of unbounded RoPE (ablation) | LongLive baseline · 60s interactive ablation | dynamic degree: 26.37 -> 29.09 | No clear effect |
| use tri-region RoPE instead of bounded RoPE (ablation) | LongLive baseline · 60s interactive ablation | dynamic degree: 29.09 -> 74.83 | Improved |
| use Anchor Forcing instead of LongLive on 30s generation | LongLive · VBenchLong 30s single-prompt | dynamic score: 40.19 -> 79.17 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 4 of 4 on most shared results.
- LongLive — LongLive: Real-time Interactive Long Video Generation (ICLR 2026) · better on 7 of 7 shared results
- MemFlow — MemFlow: Optical Flow Estimation and Prediction with Memory (CVPR 2024) · better on 6 of 6 shared results
- Infinity-RoPE — Infinity-RoPE: Action-Controllable Infinite Video Generation Emerges From Autoregressive Self-Rollout (CVPR 2026) · better on 5 of 5 shared results
- DeepForcing — Deep Forcing: Training-Free Long Video Generation with Deep Sink and Participative Compression (ICML 2026) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AdaState: Self-Evolving Anchors for Streaming Video Generation (NeurIPS 2026 · similarity 0.91)
- VideoMLA: Low-Rank Latent KV Cache for Minute-Scale Autoregressive Video Diffusion (NeurIPS 2026 · similarity 0.88)
- Attend Locally, Remember Linearly: Linear Attention as Cross-Frame Memory for Autoregressive Video Diffusion (NeurIPS 2026 · similarity 0.88)
- MagCache: Fast Video Generation with Magnitude-Aware Cache (NeurIPS 2025 · similarity 0.87)
- From Slow Bidirectional to Fast Autoregressive Video Diffusion Models (CVPR 2025 · similarity 0.87)
- LeMiCa: Lexicographic Minimax Path Caching for Efficient Diffusion-Based Video Generation (NeurIPS 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
