Video models 'commit' to physics at a depth boundary, new paper finds
ZimingLiu11 · x · 2026-09-16
A new arXiv paper by Ziming Liu et al., "A Chosen Future Can Still Be Rewritten: Causal Writability in Video Models," shows that when video models generate physically wrong motion, the correct motion is still inside the model and can be recovered via low-dimensional edits derived from simple physical variables — a property the authors call causal writability. Crucially, at fixed strength there is a sharp depth boundary: edits change the video before the boundary but not after, marking where the model "commits" to its physics. Stronger downstream writes can restore correct motion, and early writability predicts which errors later training fixes. Findings replicate in a pretrained 1.3B video model.
More from Multimodal
- Pixio bundles 500+ frontier models for image, video, audio and 3D in one credit-based studio — tsi_org · 2026-09-16
- YuE2 music model: individual creators can use generations commercially, license only applies to companies — cocktailpeanut · 2026-09-16
- Open-Source LynnReal-Omni: One 32B Model for Video Gen, Editing and Restoration, 377ms on H100 — AgeNo5351 · 2026-09-16
- Viggle's Meridian Adds Camera Control, FOV and Bullet Time to Existing Videos — init-5 · 2026-09-16
- MiniMax Unveils H3 IP Edition at Tokyo IP×AI Summit, 60+ Companies Interested — MiniMax_AI · 2026-09-16
- Turning a static Art Nouveau poster into animation: first/last frame workflow walkthrough — dipSlope · 2026-09-16