Research map / Video / Video generation
Explorative Inbetweening of Time and Space
Haiwen Feng, Zheng Ding, Zhihao Xia, Simon Niklaus, Victoria Fernandez Abrevaya, Michael J. Black, Xuaner Zhang
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 |
|---|---|---|---|
| use Time Reversal Fusion instead of frame interpolation (FILM) for dynamic bounds | Stable Video Diffusion · dynamic bounds | FVD25: 656.88 -> 431.16 | Improved |
| use TRF instead of Text2Cinemagraph for identical-bound looping | Stable Video Diffusion · identical bounds | FVD25: 911.67 -> 458.91 | Improved |
| use TRF instead of Du et al. for wide-baseline view synthesis without poses | Stable Video Diffusion · view bounds | FID192 / COLMAP: FID192 28.70 -> 10.31; COLMAP 379.61 -> 884.08 | Improved |
| condition on linear interpolation of start and end frame (condition manipulation) (ablation) | Stable Video Diffusion · qualitative | end frame adherence | Hurt |
| temporal inpainting by replacing last frame each denoising step (ablation) | Stable Video Diffusion · qualitative | temporal smoothness | Hurt |
| add noise re-injection to the time reversal fusion (ablation) | TRF · qualitative | smoothness and quality of transitions | Improved |
| increase SVD churn term instead of noise re-injection (ablation) | TRF · qualitative | frame quality | Hurt |
Closest papers by meaning
- Generative Inbetweening through Frame-wise Conditions-Driven Video Generation (CVPR 2025 · similarity 0.91)
- Generative Inbetweening: Adapting Image-to-Video Models for Keyframe Interpolation (ICLR 2025 · similarity 0.90)
- Extrapolating and Decoupling Image-to-Video Generation Models: Motion Modeling is Easier Than You Think (CVPR 2025 · similarity 0.89)
- Frame In-N-Out: Unbounded Controllable Image-to-Video Generation (NeurIPS 2025 · similarity 0.88)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.88)
- Training-free Camera Control for Video Generation (ICLR 2025 · 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.
