InfinityStory: Unlimited Video Generation with World Consistency and Character-Aware Shot Transitions

TL;DR

InfinityStory introduces a comprehensive framework for generating long-form storytelling videos with consistent backgrounds, seamless multi-subject transitions, and scalability, significantly advancing video synthesis quality and coherence.

Contribution

The paper presents a novel background-consistent generation pipeline, a transition-aware synthesis module, and a synthetic dataset to improve long-form video storytelling with multiple subjects.

Findings

Achieves 88.94 background consistency on VBench.

Attains 82.11 subject consistency, outperforming prior methods.

Demonstrates improved stability and temporal coherence in generated videos.

Abstract

Generating long-form storytelling videos with consistent visual narratives remains a significant challenge in video synthesis. We present a novel framework, dataset, and a model that address three critical limitations: background consistency across shots, seamless multi-subject shot-to-shot transitions, and scalability to hour-long narratives. Our approach introduces a background-consistent generation pipeline that maintains visual coherence across scenes while preserving character identity and spatial relationships. We further propose a transition-aware video synthesis module that generates smooth shot transitions for complex scenarios involving multiple subjects entering or exiting frames, going beyond the single-subject limitations of prior work. To support this, we contribute with a synthetic dataset of 10,000 multi-subject transition sequences covering underrepresented dynamic scene compositions. On VBench, InfinityStory achieves the highest Background Consistency (88.94), highest Subject Consistency (82.11), and the best overall average rank (2.80), showing improved stability, smoother transitions, and better temporal coherence.