InfiniCube: Unbounded and Controllable Dynamic 3D Driving Scene Generation with World-Guided Video Models

TL;DR

InfiniCube is a scalable, controllable method for generating high-fidelity, unbounded 3D driving scenes by integrating voxel-based 3D models with video models, enabling flexible scene control and realistic synthesis.

Contribution

The paper introduces a novel approach combining voxel-based 3D generation with video models for unbounded, controllable dynamic scene synthesis in driving environments.

Findings

Generates high-fidelity, unbounded 3D driving scenes

Enables flexible control via HD maps, bounding boxes, and text

Outperforms previous methods in realism and scalability

Abstract

We present InfiniCube, a scalable method for generating unbounded dynamic 3D driving scenes with high fidelity and controllability. Previous methods for scene generation either suffer from limited scales or lack geometric and appearance consistency along generated sequences. In contrast, we leverage the recent advancements in scalable 3D representation and video models to achieve large dynamic scene generation that allows flexible controls through HD maps, vehicle bounding boxes, and text descriptions. First, we construct a map-conditioned sparse-voxel-based 3D generative model to unleash its power for unbounded voxel world generation. Then, we re-purpose a video model and ground it on the voxel world through a set of carefully designed pixel-aligned guidance buffers, synthesizing a consistent appearance. Finally, we propose a fast feed-forward approach that employs both voxel and pixel branches to lift the dynamic videos to dynamic 3D Gaussians with controllable objects. Our method can generate controllable and realistic 3D driving scenes, and extensive experiments validate the effectiveness and superiority of our model.