RenderBench: Benchmarking Render-to-Real Video Transfer with Reconstructed Digital Twins
We introduce RenderBench, a benchmark of 12 reconstructed real-world scenes spanning large-scale indoor environments and egocentric viewpoints, with both static and dynamic settings.
ProofPaper ↗
Key points
- Modern video models can generate realistic videos from real appearance references and proxy renders that specify scene structure, viewpoint changes, and motion.
- Evaluating this render-to-real capability requires a real target video depicting the same scene evolution, paired with an editable, geometrically registered 3D replica.
- We evaluate transfer models against paired real target videos, retain PAI-Bench-C-compatible structural projections, and use scene annotations to localize failures by object, visibility, articulation, and motion.
- RenderBench provides paired real observations and editable scene state for assessing both appearance fidelity and preservation of geometry and dynamics.
Sources (1)
- [1]RenderBench: Benchmarking Render-to-Real Video Transfer with Reconstructed Digital TwinsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 05:03 PM
We introduce RenderBench, a benchmark of 12 reconstructed real-world scenes spanning large-scale indoor environments and egocentric viewpoints, with both static and dynamic settings.
Modern video models can generate realistic videos from real appearance references and proxy renders that specify scene structure, viewpoint changes, and motion.
Extractive summary: sentences quoted from the sources.