CoCoNO: Attention Contrast-and-Complete for Initial Noise Optimization in Text-to-Image Synthesis

TL;DR

CoCoNO introduces a novel noise optimization method for text-to-image synthesis that enhances semantic accuracy by addressing attention neglect and interference through specialized loss functions, improving alignment without retraining models.

Contribution

The paper proposes CoCoNO, a new algorithm that leverages attention contrast and completeness losses to improve initial latent optimization in text-to-image diffusion models, without retraining base models.

Findings

Significantly improves text-image alignment.

Outperforms current state-of-the-art methods.

Effective across multiple benchmarks.

Abstract

Despite recent advancements in text-to-image models, achieving semantically accurate images in text-to-image diffusion models is a persistent challenge. While existing initial latent optimization methods have demonstrated impressive performance, we identify two key limitations: (a) attention neglect, where the synthesized image omits certain subjects from the input prompt because they do not have a designated segment in the self-attention map despite despite having a high-response cross-attention, and (b) attention interference, where the generated image has mixed-up properties of multiple subjects because of a conflicting overlap between cross- and self-attention maps of different subjects. To address these limitations, we introduce CoCoNO, a new algorithm that optimizes the initial latent by leveraging the complementary information within self-attention and cross-attention maps. Our method introduces two new loss functions: the attention contrast loss, which minimizes undesirable overlap by ensuring each self-attention segment is exclusively linked to a specific subject's cross attention map, and the attention complete loss, which maximizes the activation within these segments to guarantee that each subject is fully and distinctly represented. Our approach operates within a noise optimization framework, avoiding the need to retrain base models. Through extensive experiments on multiple benchmarks, we demonstrate that CoCoNO significantly improves text-image alignment and outperforms the current state of the art.