---
id: 20261030-T0-05
title: "StruPO：让文生图提示词改写不改变用户要求的场景"
title_en: "StruPO: Faithful Text-to-Image Prompt Editing Without Changing the Scene"
url: https://ai.daily.yangsir.net/daily/20261030-T0-05
issue_date: 2026-10-30
publish_date: 2026-10-29T04:00:00.000Z
source_name: "arXiv cs.AI"
source_url: https://arxiv.org/abs/2610.36146
---

# StruPO：让文生图提示词改写不改变用户要求的场景

arXiv论文提出StruPO，用于训练文生图提示词编辑器。问题在于：提示词优化本应改善出图效果而不改变用户要求的场景，但风格化改写常常削弱物体属性和关系，即使整体图文相似度仍然很高。StruPO利用结构反馈来约束改写，保留原始提示中的物体与关系信息。对文生图用户而言，这意味着自动优化提示词时，生成的图像不会偏离最初描述的场景。

## English Version

**StruPO: Faithful Text-to-Image Prompt Editing Without Changing the Scene**

An arXiv paper proposes StruPO for training text-to-image prompt editors. The problem: prompt optimization should improve the resulting image without changing the scene the user requested, but stylistic rewriting often weakens object attributes and relations even when global text-image similarity remains high. StruPO uses structural feedback to constrain rewriting and preserve object and relation information from the original prompt. For text-to-image users, this means automatic prompt optimization won't drift the generated image away from the originally described scene.

---

**来源**：[arXiv cs.AI](https://arxiv.org/abs/2610.36146)

**详情页**：https://ai.daily.yangsir.net/daily/20261030-T0-05

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*