Research map / Video / Video generation
Flexible Diffusion Modeling of Long Videos
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| FDM with hierarchical sampling vs VDM | long video diffusion · GQN-Mazes | FVD: 66.7 -> 53.1 | Helped |
| hierarchical vs autoregressive sampling scheme | FDM · GQN-Mazes | FVD: 86.4 -> 53.1 | Helped |
| hierarchical vs autoregressive sampling scheme | FDM · MineRL | FVD: 267 -> 311 | Hurt |
| hierarchical vs autoregressive sampling scheme | FDM · CARLA Town01 | OP: 0.47 -> 3.28 | Hurt |
| optimize observed indices offline | FDM Hierarchy-2 · GQN-Mazes | FVD: 53.1 -> 51.1 | Helped |
| optimize observed indices offline | FDM Autoreg · MineRL | FVD: 281 -> 257 | Helped |
| train on single Autoreg task instead of broad task distribution (ablation) | FDM · MineRL | FVD: 281 -> 234 | Helped |
| use structured training task distribution instead of uniform (ablation) | FDM · GQN-Mazes | FVD: 296 -> 226 | Helped |
| use structured training task distribution instead of uniform (ablation) | FDM · MineRL | FVD: 786 -> 518 | Helped |
| diffusion models (FDM/VDM) vs TATS and CWVAE | video completion · MineRL | FVD: 807 -> 267 | Helped |
Closest papers by meaning
- FreeLong: Training-Free Long Video Generation with SpectralBlend Temporal Attention (NeurIPS 2024 · similarity 0.91)
- FreeNoise: Tuning-Free Longer Video Diffusion via Noise Rescheduling (ICLR 2024 · similarity 0.90)
- AR-Diffusion: Asynchronous Video Generation with Auto-Regressive Diffusion (CVPR 2025 · similarity 0.90)
- Efficient Video Diffusion Models via Content-Frame Motion-Latent Decomposition (ICLR 2024 · similarity 0.89)
- Generative Inbetweening: Adapting Image-to-Video Models for Keyframe Interpolation (ICLR 2025 · similarity 0.89)
- TRIP: Temporal Residual Learning with Image Noise Prior for Image-to-Video Diffusion Models (CVPR 2024 · similarity 0.89)
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.