LSD-3D: Large-Scale 3D Driving Scene Generation with Geometry Grounding

TL;DR

LSD-3D introduces a novel method for generating large-scale, realistic 3D driving scenes with accurate geometry and high controllability, bridging the gap between static reconstruction and diffusion-based scene generation.

Contribution

The paper presents a new approach combining proxy geometry and score distillation to generate controllable, geometrically accurate 3D driving scenes from prompts.

Findings

Enables prompt-guided, geometry-aware 3D scene generation.

Produces realistic, consistent 3D driving scenes with object permanence.

Allows explicit 3D geometry estimation and causal view synthesis.

Abstract

Large-scale scene data is essential for training and testing in robot learning. Neural reconstruction methods have promised the capability of reconstructing large physically-grounded outdoor scenes from captured sensor data. However, these methods have baked-in static environments and only allow for limited scene control -- they are functionally constrained in scene and trajectory diversity by the captures from which they are reconstructed. In contrast, generating driving data with recent image or video diffusion models offers control, however, at the cost of geometry grounding and causality. In this work, we aim to bridge this gap and present a method that directly generates large-scale 3D driving scenes with accurate geometry, allowing for causal novel view synthesis with object permanence and explicit 3D geometry estimation. The proposed method combines the generation of a proxy geometry and environment representation with score distillation from learned 2D image priors. We find that this approach allows for high controllability, enabling the prompt-guided geometry and high-fidelity texture and structure that can be conditioned on map layouts -- producing realistic and geometrically consistent 3D generations of complex driving scenes.