Research map / Video / Video generation
Physical Simulator In-the-Loop Video Generation
Lin Geng Foo, Mark He Huang, Alexandros Lattas, Stylianos Moschoglou, Thabo Beeler, Christian Theobalt
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 test-time texture-consistency optimization (TTCO) (ablation) | PSIVG · Physics text-to-video prompts (LLM-generated) | Corr. Pixel MSE: 0.009 -> 0.007 | Improved |
| add TTCO (ablation) | PSIVG · Physics text-to-video prompts (LLM-generated) | SAM mIoU: 0.82 -> 0.84 | Improved |
| add TTCO (ablation) | PSIVG · Physics text-to-video prompts (LLM-generated) | Subject Consistency: 0.93 -> 0.95 | Improved |
| optimize text-token residuals and feature modulation instead of test-time LoRA in TTCO (ablation) | PSIVG · qualitative (Fig. 6) | video quality (background) | Improved |
| optimize object-specific spatio-temporal tokens instead of text tokens (ablation) | PSIVG TTCO · qualitative | visual artifacts | Hurt |
| guide video diffusion with simulator-derived optical flow instead of text-only generation (vs CogVideoX) | video generator · Physics text-to-video prompts (LLM-generated) | SAM mIoU: 0.47 -> 0.84 | Improved |
| guide video diffusion with simulator flow (vs HunyuanVideo) | video generator · Physics text-to-video prompts (LLM-generated) | Corr. Pixel MSE: 0.017 -> 0.007 | Improved |
| fine-tune for physics (PISA variants) instead of simulator guidance | video generator · Physics text-to-video prompts (LLM-generated) | SAM mIoU: 0.46 -> 0.50 (PISA-Seg vs HunyuanVideo) | No clear effect |
| use PSIVG instead of baseline generators on general video quality | PSIVG · Physics text-to-video prompts (LLM-generated) | Motion Smoothness: 0.96 -> 0.96 (vs HunyuanVideo, similar to PISA) | No clear effect |
| use PSIVG over strongest baselines in user preference | PSIVG · User study | preference rate (%): 7.2 -> 82.3 (vs CogVideoX) | Improved |
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.
- SG-I2V — SG-I2V: Self-Guided Trajectory Control in Image-to-Video Generation (ICLR 2025) · better on 7 of 8 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
- PhysCtrl: Generative Physics for Controllable and Physics-Grounded Video Generation (NeurIPS 2025 · similarity 0.89)
- MotionCraft: Physics-Based Zero-Shot Video Generation (NeurIPS 2024 · similarity 0.88)
- How Far Is Video Generation from World Model: A Physical Law Perspective (ICML 2025 · similarity 0.88)
- DreamPhysics: Learning Physics-Based 3D Dynamics with Video Diffusion Priors (AAAI 2025 · similarity 0.87)
- LaMo: Self-Supervised Latent Motion Priors for Physical Realism in Video Generation (NeurIPS 2026 · similarity 0.86)
- Force Prompting: Video Generation Models Can Learn And Generalize Physics-based Control Signals (NeurIPS 2025 · similarity 0.86)
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.
