LongVie 2: Multimodal Controllable Ultra-Long Video World Model

TL;DR

LongVie 2 is a novel multimodal autoregressive framework for controllable, long-term, high-quality video generation that advances the state-of-the-art in video world modeling.

Contribution

It introduces a three-stage training process incorporating multi-modal guidance, degradation-aware training, and history-context guidance for improved controllability and temporal consistency.

Findings

Achieves state-of-the-art long-range controllability and coherence.

Supports continuous video generation up to five minutes.

Introduces LongVGenBench benchmark for diverse high-resolution videos.

Abstract

Building video world models upon pretrained video generation systems represents an important yet challenging step toward general spatiotemporal intelligence. A world model should possess three essential properties: controllability, long-term visual quality, and temporal consistency. To this end, we take a progressive approach-first enhancing controllability and then extending toward long-term, high-quality generation. We present LongVie 2, an end-to-end autoregressive framework trained in three stages: (1) Multi-modal guidance, which integrates dense and sparse control signals to provide implicit world-level supervision and improve controllability; (2) Degradation-aware training on the input frame, bridging the gap between training and long-term inference to maintain high visual quality; and (3) History-context guidance, which aligns contextual information across adjacent clips to ensure temporal consistency. We further introduce LongVGenBench, a comprehensive benchmark comprising 100 high-resolution one-minute videos covering diverse real-world and synthetic environments. Extensive experiments demonstrate that LongVie 2 achieves state-of-the-art performance in long-range controllability, temporal coherence, and visual fidelity, and supports continuous video generation lasting up to five minutes, marking a significant step toward unified video world modeling.