PromptEvolver: Prompt Inversion through Evolutionary Optimization in Natural-Language Space

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Buchnick, Asaf, Shamsian, Aviv, Navon, Aviv, Fetaya, Ethan
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911629859356672
author Buchnick, Asaf
Shamsian, Aviv
Navon, Aviv
Fetaya, Ethan
author_facet Buchnick, Asaf
Shamsian, Aviv
Navon, Aviv
Fetaya, Ethan
contents Text-to-image generation has progressed rapidly, but faithfully generating complex scenes requires extensive trial-and-error to find the exact prompt. In the prompt inversion task, the goal is to recover a textual prompt that can faithfully reconstruct a given target image. Currently, existing methods frequently yield suboptimal reconstructions and produce unnatural, hard-to-interpret prompts that hinder transparency and controllability. In this work, we present PromptEvolver, a prompt inversion approach that generates natural-language prompts while achieving high-fidelity reconstructions of the target image. Our method uses a genetic algorithm to optimize the prompt, leveraging a strong vision-language model to guide the evolution process. Importantly, it works on black-box generation models by requiring only image outputs. Finally, we evaluate PromptEvolver across multiple prompt inversion benchmarks and show that it consistently outperforms competing methods.
format Preprint
id arxiv_https___arxiv_org_abs_2604_06061
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle PromptEvolver: Prompt Inversion through Evolutionary Optimization in Natural-Language Space
Buchnick, Asaf
Shamsian, Aviv
Navon, Aviv
Fetaya, Ethan
Machine Learning
Text-to-image generation has progressed rapidly, but faithfully generating complex scenes requires extensive trial-and-error to find the exact prompt. In the prompt inversion task, the goal is to recover a textual prompt that can faithfully reconstruct a given target image. Currently, existing methods frequently yield suboptimal reconstructions and produce unnatural, hard-to-interpret prompts that hinder transparency and controllability. In this work, we present PromptEvolver, a prompt inversion approach that generates natural-language prompts while achieving high-fidelity reconstructions of the target image. Our method uses a genetic algorithm to optimize the prompt, leveraging a strong vision-language model to guide the evolution process. Importantly, it works on black-box generation models by requiring only image outputs. Finally, we evaluate PromptEvolver across multiple prompt inversion benchmarks and show that it consistently outperforms competing methods.
title PromptEvolver: Prompt Inversion through Evolutionary Optimization in Natural-Language Space
topic Machine Learning
url https://arxiv.org/abs/2604.06061