Search papers, labs, and topics across Lattice.
The World Embedding Benchmark, a dataset of 8,000 controlled simulation cases with paired videos and physical annotations across mechanics and electromagnetism, is introduced to evaluate how video representations encode physical information. Experiments reveal that pre-trained omnimodal models exhibit weak physical alignment and near-chance classification, though lightweight probes can extract quantitative physics from frozen embeddings. Contrastive training improves alignment but sacrifices quantitative recoverability, underscoring a trade-off, and retrieval-augmented generation using these embeddings demonstrably enhances physical fidelity in generated videos.
Current omnimodal video embeddings struggle with physical reasoning, yet specific physics can be recovered and leveraged to significantly improve the physical fidelity of generated videos.
Physical fidelity has received increasing attention in world models and video generation, yet how video representations encode physical information remains less understood. We introduce the World Embedding Benchmark, comprising 8,000 controlled simulation cases from 80 families spanning fluid mechanics, solid mechanics, dynamics, and optics&electromagnetism. Each case pairs a rendered video with simulation-derived physical annotations, supporting three complementary tasks: text-video retrieval, physical-property regression, and multiple-choice video-description pair classification. We use these tasks to distinguish cross-modal physical alignment from the recoverability of quantitative physical information. Evaluated pre-trained omnimodal embedding models show weak retrieval and near-chance within-family pair classification, while lightweight probes recover useful physical information from frozen video embeddings. Continual contrastive training with physics-specific video-text pairs improves retrieval and pair classification but degrades physical-property regression, revealing a trade-off between alignment and quantitative information recoverability. Finally, we use the embeddings to retrieve reference videos for retrieval-augmented generation with MiniMax-H3. Retrieved references improve the physical fidelity of generated videos, with stronger retrieval models yielding larger gains in our experiments. Together, these findings highlight the need to evaluate physical alignment and property recoverability jointly, and demonstrate the utility of physical representations for improving video generation.
