Preference-Guided Prompt Optimization for Text-to-Image Generation

TL;DR

APPO is a preference-guided prompt optimization algorithm that efficiently improves text-to-image generation by using binary user feedback, reducing effort and iterations needed for satisfactory results.

Contribution

This paper introduces APPO, a novel method that leverages binary preferences for prompt optimization, enabling faster and more user-friendly control of generative models.

Findings

APPO achieves satisfactory image generation results in fewer iterations.

APPO reduces cognitive load compared to manual prompt editing.

APPO effectively balances exploration and exploitation based on user feedback.

Abstract

Generative models are increasingly powerful, yet users struggle to guide them through prompts. The generative process is difficult to control and unpredictable, and user instructions may be ambiguous or under-specified. Prior prompt refinement tools heavily rely on human effort, while prompt optimization methods focus on numerical functions and are not designed for human-centered generative tasks, where feedback is better expressed as binary preferences and demands convergence within few iterations. We present APPO, a preference-guided prompt optimization algorithm. Instead of iterating prompts, users only provide binary preferential feedback. APPO adaptively balances its strategies between exploiting user feedback and exploring new directions, yielding effective and efficient optimization. We evaluate APPO on image generation, and the results show APPO enables achieving satisfactory outcomes in fewer iterations with lower cognitive load than manual prompt editing. We anticipate APPO will advance human-AI collaboration in generative tasks by leveraging user preferences to guide complex content creation.