FreeStory: Training-Free Character Consistency for Free-Form Visual Storytelling

2026-06-23

AI Summary

Existing methods struggle to generate temporally consistent character appearances in image sequences from free-form narrative prompts due to their reliance on repeatedly specifying full character descriptions. This work proposes a training-free framework that models character consistency through entity-based feature reuse, leveraging dynamic character masking, correspondence-aware feature matching, and key-value injection with query fusion to preserve identity while maintaining generation diversity. For the first time, we enable training-free generation of coherent visual narratives driven by natural-language-style prompts. We also introduce FreeStoryBench, a new evaluation benchmark encompassing both single- and multi-character story scenarios. The proposed method achieves state-of-the-art performance among training-free approaches on structured benchmarks and significantly outperforms existing baselines under free-form prompting conditions.

Abstract

Visual storytelling aims to generate image sequences that are both aligned with narrative prompts and consistent in character appearance across images. Recent training-free methods improve character consistency by reusing attention features, but rely on structured prompts where full character descriptions are repeated in every prompt. This assumption simplifies the task but deviates from natural storytelling, where characters are typically introduced once and later referred to using pronouns or type-based expressions. We propose \textbf{FreeStory}, a training-free framework that reformulates character consistency under free-form prompts as entity-grounded feature reuse. Our method associates reference mentions with their corresponding character descriptions and combines dynamic character masks, correspondence-aware feature matching, key-value injection, and query blending to preserve identity while retaining generation diversity. We also introduce \textbf{FreeStoryBench}, a benchmark for this setting that includes both single- and multi-character stories. Experiments show that FreeStory achieves state-of-the-art performance among training-free methods on structured benchmarks and stronger overall consistency over baselines under free-form prompts.

Problem

• visual storytelling
• character consistency
• free-form prompts
• training-free
• entity grounding

Innovation

• training-free
• character consistency
• free-form prompts
• entity-grounded feature reuse
• visual storytelling