VistaGEN: Consistent Driving Video Generation with Fine-Grained Control Using Multiview Visual-Language Reasoning

2026-03-30

AI Summary

Existing methods for driving video generation struggle to achieve fine-grained, object-level control while maintaining spatiotemporal consistency in long-duration videos. To address this challenge, this work proposes a closed-loop generation framework that enables precise manipulation of specific entities—such as 3D objects, images, and text—through multi-view visual-language reasoning. The framework incorporates a Multi-View Visual-Language Evaluator (MV-VLM) and an object-level optimization module, facilitating an iterative generate–evaluate–regenerate refinement process. This approach substantially enhances controllability and spatiotemporal coherence in long driving videos across diverse scenarios, particularly those involving long-tail objects, significantly outperforming current state-of-the-art methods.

Abstract

Driving video generation has achieved much progress in controllability, video resolution, and length, but fails to support fine-grained object-level controllability for diverse driving videos, while preserving the spatiotemporal consistency, especially in long video generation. In this paper, we present a new driving video generation technique, called VistaGEN, which enables fine-grained control of specific entities, including 3D objects, images, and text descriptions, while maintaining spatiotemporal consistency in long video sequences. Our key innovation is the incorporation of multiview visual-language reasoning into the long driving video generation. To this end, we inject visual-language features into a multiview video generator to enable fine-grained controllability. More importantly, we propose a multiview vision-language evaluator (MV-VLM) to intelligently and automatically evaluate spatiotemporal consistency of the generated content, thus formulating a novel generation-evaluation-regeneration closed-loop generation mechanism. This mechanism ensures high-quality, coherent outputs, facilitating the creation of complex and reliable driving scenarios. Besides, within the closed-loop generation, we introduce an object-level refinement module to refine the unsatisfied results evaluated from the MV-VLM and then feed them back to the video generator for regeneration. Extensive evaluation shows that our VistaGEN achieves diverse driving video generation results with fine-grained controllability, especially for long-tail objects, and much better spatiotemporal consistency than previous approaches.

Problem

• driving video generation
• fine-grained control
• spatiotemporal consistency
• object-level controllability
• long video generation

Innovation

• multiview visual-language reasoning
• fine-grained control
• spatiotemporal consistency
• closed-loop generation
• driving video generation