Search papers, labs, and topics across Lattice.
This paper introduces Cycle-World, a framework that mitigates error accumulation in long-term video generation by enforcing temporal reversibility during both training and inference. By integrating a reverse-prediction model, the approach ensures that generated sequences adhere closely to the natural video manifold, effectively controlling forward generative drift through a cycle-consistency objective. Experimental results on the VBench benchmark reveal that Cycle-World significantly enhances generation quality and temporal consistency over 60-second video synthesis compared to existing methods.
Error drift in long-horizon video generation can be effectively controlled, leading to state-of-the-art synthesis quality and temporal consistency.
Autoregressive diffusion models have enabled high-quality video generation, yet their sequential nature inherently suffers from error accumulation. In long-horizon video synthesis, minor prediction deviations compound over time, inevitably leading to unconstrained generative drift, structural collapse, and severe visual degradation. To address this, we propose Cycle-World, a novel framework designed for stable and temporally consistent long-video generation. Our approach tackles error drift by enforcing strict temporal reversibility across both the training and inference phases. Theoretically, we demonstrate that forward generative drift can be strictly bottlenecked by a cycle-consistency objective. During training, we integrate an efficient reverse-prediction model to implicitly embed causal constraints into the forward generator, compelling it to produce reversible sequences that tightly adhere to the natural video manifold. At inference time, we repurpose this frozen reverse model as a runtime corrector. Through gradient-based cycle guidance, it iteratively refines the generated latent representations, actively suppressing accumulated errors before they are committed to the historical context. Extensive experiments on the VBench benchmark demonstrate that Cycle-World's dual-phase synergy significantly mitigates error drift, achieving state-of-the-art overall generation quality and long-horizon temporal consistency in 60-second synthesis.
