Problem
Research questions and friction points this paper is trying to address.
long-context generation
image consistency
multi-image generation
visual narratives
character consistency
LCG: Long-Context Consistent Image Generation with Sparse Relational Attention
🤖 AI Summary
Current image generation models excel at single-image synthesis but often suffer from semantic and visual inconsistencies when generating long sequences of images, such as comics or storyboards. This work proposes a Long-context Generation framework (LCG) that introduces a Sparse Relational Attention (SRA) mechanism to selectively focus on salient features across extended visual contexts, along with a Routing Consistency Constraint (RCC) to align structural patterns across multiple generative branches. The study also constructs LCCD, the first large-scale dataset dedicated to long-context consistency, and incorporates an identity-aware masking strategy. Experimental results demonstrate that LCG significantly outperforms existing baselines in generating sequences of 6 to 20 images, achieving state-of-the-art performance in character consistency and prompt alignment.
📝 Abstract
Recent image generation models achieve impressive quality in single-image synthesis, but often fail to maintain consistency across sequential outputs, as required in comics, storyboards, and visual narratives. We propose Long-Context Generation (LCG), a framework for long-context multi-image text-to-image generation, to improve consistency and scalability in long-context multi-image generation. LCG employs the Sparse Relational Attention (SRA) mechanism to selectively attend to core features across extended visual contexts, ensuring that the propagation of semantic and layout information remains computationally tractable. To enforce semantic alignment, we introduce the Routing Consistency Constraint (RCC), which leverages identity-aware masks to align structural patterns across generation branches, effectively mitigating drift in appearance even in complex multi-character scenes. To support training and evaluation in this setting, we construct the Long-Context Consistency Dataset (LCCD), a large-scale synthetic dataset comprising character-centric multi-image sequences spanning varied situational contexts. LCCD contains 600K training sequences and a separate 1K test set, with each sequence containing 6 to 20 images. The experiments demonstrate that LCG outperforms the compared baselines in prompt alignment and character consistency for long-context image generation, including multi-character scenes.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Long-Context Generation
Sparse Relational Attention
Routing Consistency Constraint
Multi-image Consistency
Text-to-Image Synthesis
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
