TRAC: Trajectory-aware Reuse and Adaptive Correction for Efficient Autoregressive Video Generation

2026-10-02

AI Summary

This study addresses error accumulation-induced quality degradation and inference inefficiency in autoregressive video generation by proposing a training-free acceleration framework. The core innovation lies in the first trajectory-aware mechanism designed to model autoregressive chain dependencies, overcoming the limitations of conventional single-trajectory acceleration. Methodologically, the framework optimizes cache reuse strategies through Autoregressive Trajectory-aware Guided Scheduling (ATGS) and Robust Cumulative Scheduling (RCS), while introducing Spectral Structure Correction (SSC) to rectify low-frequency deviations. Experiments demonstrate that this framework simultaneously achieves state-of-the-art inference efficiency and generation quality on benchmarks such as SkyReels-V2, significantly outperforming existing methods.

Abstract

In this paper, we present trajectory-aware reuse and adaptive correction (TRAC), a training-free framework for efficient autoregressive (AR) video generation. Existing acceleration methods mainly target single-trajectory generation with bidirectional attention. AR video generation, by contrast, sequentially couples chunk-level denoising trajectories. Consequently, approximation errors accumulate and propagate through the generation process. TRAC addresses this challenge with three components, including robust cumulative scheduling (RCS), autoregressive trajectory-aware guidance scheduling (ATGS), and spectral structure correction (SSC). RCS selects cache reuse schedules by cumulative rollout error and cross-chunk/prompt variation. ATGS coordinates CFG refreshes along the global AR trajectory. SSC restores low-frequency structure of the first chunk to correct long-term structural loss. Experiments on SkyReels-V2 and FramePack-F1 show that, compared with existing methods, TRAC achieves both the highest inference efficiency and the best generation quality for AR video generation.

Problem

• Autoregressive Video Generation
• Inference Acceleration
• Error Accumulation
• Denoising Trajectory

Innovation

• Autoregressive Video Generation
• Training-free Acceleration
• Trajectory-aware Reuse
• Adaptive Correction
• Spectral Structure Correction