GRASE: Geometry-native Real-to-sim Agentic Scene Modeling from Video
We present GRASE, an agentic framework that closes this loop by treating the captured multi-view video observations as a standing reference throughout scene modeling.
ProofPaper ↗
Key points
- Real-to-sim conversion aims to turn a captured video of a real environment into an editable 3D scene while preserving the geometry and layout of what was observed.
- Existing methods often follow a one-pass, loosely coupled paradigm: each stage receives only limited feedback from previous stages, allowing errors to propagate without reliable correction and leaving the final scene inconsistent.
- An agent then places generated assets in a shared coordinate frame while checking their spatial relationships.
- The resulting 3D scenes are faithful to the input, consistent across views, and suitable for simulations.
Sources (1)
- [1]GRASE: Geometry-native Real-to-sim Agentic Scene Modeling from VideoarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 15, 08:34 AM
We present GRASE, an agentic framework that closes this loop by treating the captured multi-view video observations as a standing reference throughout scene modeling.
Real-to-sim conversion aims to turn a captured video of a real environment into an editable 3D scene while preserving the geometry and layout of what was observed.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 14, 2026Grok Voice 3.0 from SpaceXSI outscores OpenAI and Google on speech-to-speech index
- Oct 14, 2026NVIDIA/TensorRT-LLM v1.4.0rc0
- Oct 14, 2026The Safety Culture at the Labs is Broken
- Oct 14, 2026Amp puts AI agents to work beyond code with OpenAI models
- Oct 14, 2026Discover AI: A Learning Guide - 51 hands-on labs for learning how AI actually works.
- Oct 13, 2026My local Qwen3.8FN agent is significantly cheaper than cloud