New dataset and model advance physically-grounded video generation

Researchers have introduced PhyParam-Dataset, a collection of 130,000 physically simulated videos designed to improve controllable video generation. This dataset includes detailed physical parameters such as force vectors, material properties, and environmental constants. To leverage this data, they developed PhyParam, a diffusion model that conditions video generation on object-level physical attributes and scene-level gravity. Additionally, they established PhyParam-Bench, a benchmark to evaluate the physical consistency of generated videos. AI

IMPACT Advances the ability to generate physically realistic and controllable videos, potentially impacting simulations and content creation.

RANK_REASON The cluster contains an academic paper detailing a new dataset, model, and benchmark for video generation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• arXiv
• CatalyzeX
• DagsHub
• Gotit.pub
• Hugging Face
• PhyParam
• PhyParam-Bench
• PhyParam-Dataset
• ScienceCast

• paper
• model release
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →