Grounded Forcing: Bridging Time-Independent Semantics and Proximal Dynamics in Autoregressive Video Synthesis

TL;DR

Grounded Forcing introduces a comprehensive framework for long-term, coherent, and controllable autoregressive video synthesis by integrating semantic stability, positional consistency, and smooth prompt transitions.

Contribution

It presents three novel mechanisms—Dual Memory KV Cache, Dual-Reference RoPE Injection, and Asymmetric Proximity Recache—that collectively improve long-range coherence and controllability in video generation.

Findings

Enhanced long-term semantic coherence and identity stability.

Reduced visual drift and improved visual stability.

Facilitated smooth semantic inheritance during prompt transitions.

Abstract

Autoregressive video synthesis offers a promising pathway for infinite-horizon generation but is fundamentally hindered by three intertwined challenges: semantic forgetting from context limitations, visual drift due to positional extrapolation, and controllability loss during interactive instruction switching. Current methods often tackle these issues in isolation, limiting long-term coherence. We introduce Grounded Forcing, a novel framework that bridges time-independent semantics and proximal dynamics through three interlocking mechanisms. First, to address semantic forgetting, we propose a Dual Memory KV Cache that decouples local temporal dynamics from global semantic anchors, ensuring long-term semantic coherence and identity stability. Second, to suppress visual drift, we design Dual-Reference RoPE Injection, which confines positional embeddings within the training manifold while rendering global semantics time-invariant. Third, to resolve controllability issues, we develop Asymmetric Proximity Recache, which facilitates smooth semantic inheritance during prompt transitions via proximity-weighted cache updates. These components operate synergistically to tether the generative process to stable semantic cores while accommodating flexible local dynamics. Extensive experiments demonstrate that Grounded Forcing significantly enhances long-range consistency and visual stability, establishing a robust foundation for interactive long-form video synthesis.