Instructing Text-to-Image Diffusion Models via Classifier-Guided Semantic Optimization

TL;DR

This paper introduces a method to guide text-to-image diffusion models using classifier-optimized semantic embeddings, enabling precise, prompt-free image editing with high disentanglement and generalization.

Contribution

It proposes a novel approach to steer diffusion models via attribute classifiers, eliminating the need for manual prompts or model fine-tuning.

Findings

Achieves high disentanglement in image editing.

Demonstrates strong generalization across domains.

Does not require training or fine-tuning of diffusion models.

Abstract

Text-to-image diffusion models have emerged as powerful tools for high-quality image generation and editing. Many existing approaches rely on text prompts as editing guidance. However, these methods are constrained by the need for manual prompt crafting, which can be time-consuming, introduce irrelevant details, and significantly limit editing performance. In this work, we propose optimizing semantic embeddings guided by attribute classifiers to steer text-to-image models toward desired edits, without relying on text prompts or requiring any training or fine-tuning of the diffusion model. We utilize classifiers to learn precise semantic embeddings at the dataset level. The learned embeddings are theoretically justified as the optimal representation of attribute semantics, enabling disentangled and accurate edits. Experiments further demonstrate that our method achieves high levels of disentanglement and strong generalization across different domains of data.