SPIRAL: Self-Evolving Action-Conditioned Video Generation via Reflective Planning Agents

TL;DR

SPIRAL is a novel closed-loop framework for long-horizon, action-conditioned video generation that employs planning, reflection, and self-evolution to improve temporal coherence and action accuracy.

Contribution

It introduces a think-act-reflect process with a planning agent, a video generator, and a critic agent, along with new datasets and benchmarks for evaluation.

Findings

SPIRAL improves long-horizon video coherence across multiple backbones.

The self-evolving strategy enhances action quality and temporal consistency.

Experiments show consistent gains on ActVideoGen-Bench and VBench.

Abstract

Long-horizon action-conditioned video generation aims to synthesize temporally coherent videos that follow complex action instructions over extended horizons, requiring procedural ordering, persistent action execution, and scene consistency beyond conventional TI2V's short-term fidelity. Existing single-shot video generation models typically operate in an open-loop manner, leading to incomplete action execution, hallucinated motions, and temporal drift. To address this, we propose SPIRAL, a closed-loop framework that performs sequential planning and iterative reflection for action-conditioned long-horizon video generation. Specifically, SPIRAL instantiates a think-act-reflect process: a PlanAgent decomposes high-level goals into sub-actions, which condition a VideoGenerator to synthesize each segment alongside a memory context, while a CriticAgent evaluates intermediate video segments to provide corrective feedback for iterative refinement. This closed-loop design further supports self-evolution by utilizing PlanAgent-proposed actions and CriticAgent-derived rewards for GRPO-based post-training to enhance the video generator's long-horizon consistency. Moreover, we introduce ActVideoGen-Dataset for task-specific training, and establish ActVideoGen-Bench as a dedicated evaluation suite for measuring action quality and temporal coherence. Experiments across multiple TI2V backbones alongside the self-evolving strategy show consistent gains on ActVideoGen-Bench and VBench, demonstrating the effectiveness of SPIRAL.