Research map / Video / Video generation
DynaMem: Consistent Long Video Generation via Hierarchical Memory and Motion Priors
Jingyu Lin, Xinyi Shang, Peng Sun, Cunjian Chen, Zhiqiang Shen
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 curated prompt-guided memory retrieval instead of last frame only (ablation) | DynaMem · LV-Bench | VDE Subject: 1.2190 -> 0.0753 | Improved |
| add motion-driven references instead of last frame only (ablation) | DynaMem memory · LV-Bench | VDE Subject: 1.2190 -> 0.6547 | Improved |
| curate memory instead of retrieving from all frames (ablation) | DynaMem · LV-Bench | VDE Subject: 0.1845 -> 0.0753 | Improved |
| curate memory instead of retrieving from all frames (ablation) | DynaMem · LV-Bench VBench | Aesthetic Quality: 0.6534 -> 0.6881 | Improved |
| curate memory instead of retrieving from all frames (ablation) | DynaMem · LV-Bench VBench | Motion Smoothness: 0.9961 -> 0.9956 | No clear effect |
| add motion-prioritized optimization over Ldiff only (ablation) | DynaMem · LV-Bench | VDE Motion: 0.0537 -> 0.0096 | Improved |
| add motion-aware SFT to Ldiff only (ablation) | DynaMem training · LV-Bench | VDE Motion: 0.0537 -> 0.0243 | Improved |
| full MPO with gradient-based data selection over motion SFT only (ablation) | DynaMem training · LV-Bench | VDE Subject: 0.3429 -> 0.0753 | Improved |
| add RAPA perceptual alignment (ablation) | DynaMem · 1-minute generations | IQA at 60s: 1.67 -> 1.97 | Improved |
| add RAPA perceptual alignment (ablation) | DynaMem · 1-minute generations | IQA at 10s: 1.98 -> 1.99 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- LongLive — LongLive: Real-time Interactive Long Video Generation (ICLR 2026) · better on 3 of 3 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
- StreamingT2V: Consistent, Dynamic, and Extendable Long Video Generation from Text (CVPR 2025 · similarity 0.91)
- VSTAR: Generative Temporal Nursing for Longer Dynamic Video Synthesis (ICLR 2025 · similarity 0.89)
- Advancing Narrative Long Video Generation via Training-Free Identity-Aware Memory (NeurIPS 2026 · similarity 0.89)
- StoryDiffusion: Consistent Self-Attention for Long-Range Image and Video Generation (NeurIPS 2024 · similarity 0.89)
- Long Video Diffusion Generation with Segmented Cross-Attention and Content-Rich Video Data Curation (CVPR 2025 · similarity 0.88)
- LVD-2M: A Long-take Video Dataset with Temporally Dense Captions (NeurIPS 2024 · similarity 0.88)
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.
