Prompt Optimizer of Text-to-Image Diffusion Models for Abstract Concept Understanding

arXiv:2404.11589v1 Announce Type: new
Abstract: The rapid evolution of text-to-image diffusion models has opened the door of generative AI, enabling the translation of textual descriptions into visually compelling images with remarkable quality. However, a persistent challenge within this domain is the optimization of prompts to effectively convey abstract concepts into concrete objects. For example, text encoders can hardly express "peace", while can easily illustrate olive branches and white doves. This paper introduces a novel approach named Prompt Optimizer for Abstract …

abstract arxiv challenge concept concepts cs.ai cs.cv cs.lg diffusion diffusion models domain enabling evolution generative however image image diffusion images optimization prompt prompts quality text text-to-image textual translation type understanding