A$^2$RD: Agentic Autoregressive Diffusion for Long Video Consistency

2026-05-07

AI Summary

This work addresses the challenges of semantic drift and narrative collapse that undermine consistency and coherence in long-form video generation. To this end, the authors propose a closed-loop, segment-wise generation and self-optimization framework that decouples creative synthesis from consistency constraints through an iterative “retrieve–synthesize–refine–update” cycle. The approach integrates multimodal video memory, adaptive segmentation, hierarchical test-time self-optimization, and multi-generation mode switching. Evaluated on established benchmarks and a newly introduced LVBench-C dataset, the method achieves up to a 30% improvement in visual-temporal consistency and a 20% gain in narrative coherence, with human evaluations further confirming significant enhancements in motion smoothness and scene transition quality.

Abstract

Synthesizing consistent and coherent long video remains a fundamental challenge. Existing methods suffer from semantic drift and narrative collapse over long horizons. We present A$^2$RD, an Agentic Auto-Regressive Diffusion architecture that decouples creative synthesis from consistency enforcement. A$^2$RD formulates long video synthesis as a closed-loop process that synthesizes and self-improves video segment-by-segment through a Retrieve--Synthesize--Refine--Update cycle. It comprises three core components: (i) Multimodal Video Memory that tracks video progression across modalities; (ii) Adaptive Segment Generation that switches among generation modes for natural progression and visual consistency; and (iii) Hierarchical Test-Time Self-Improvement that self-improves each segment at frame and video levels to prevent error propagation. We further introduce LVBench-C, a challenging benchmark with non-linear entity and environment transitions to stress-test long-horizon consistency. Across public and LVBench-C benchmarks spanning one- to ten-minute videos, A$^2$RD outperforms state-of-the-art baselines by up to 30% in consistency and 20% in narrative coherence. Human evaluations corroborate these gains while also highlighting notable improvements in motion and transition smoothness.

Problem

• long video synthesis
• semantic drift
• narrative coherence
• video consistency
• temporal consistency

Innovation

• Agentic Autoregressive Diffusion
• Long Video Synthesis
• Multimodal Video Memory
• Self-Improvement
• Consistency Enforcement