Order Is Not Layout: Order-to-Space Bias in Image Generation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Yongkang, Zhao, Zonglin, Zhang, Yuechen, Ding, Fei, Li, Pei, Wang, Wenxuan
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917312387350528
author Zhang, Yongkang
Zhao, Zonglin
Zhang, Yuechen
Ding, Fei
Li, Pei
Wang, Wenxuan
author_facet Zhang, Yongkang
Zhao, Zonglin
Zhang, Yuechen
Ding, Fei
Li, Pei
Wang, Wenxuan
contents We study a systematic bias in modern image generation models: the mention order of entities in text spuriously determines spatial layout and entity--role binding. We term this phenomenon Order-to-Space Bias (OTS) and show that it arises in both text-to-image and image-to-image generation, often overriding grounded cues and causing incorrect layouts or swapped assignments. To quantify OTS, we introduce OTS-Bench, which isolates order effects with paired prompts differing only in entity order and evaluates models along two dimensions: homogenization and correctness. Experiments show that Order-to-Space Bias (OTS) is widespread in modern image generation models, and provide evidence that it is primarily data-driven and manifests during the early stages of layout formation. Motivated by this insight, we show that both targeted fine-tuning and early-stage intervention strategies can substantially reduce OTS, while preserving generation quality.
format Preprint
id arxiv_https___arxiv_org_abs_2603_03714
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Order Is Not Layout: Order-to-Space Bias in Image Generation
Zhang, Yongkang
Zhao, Zonglin
Zhang, Yuechen
Ding, Fei
Li, Pei
Wang, Wenxuan
Computation and Language
Artificial Intelligence
Computer Vision and Pattern Recognition
Multimedia
We study a systematic bias in modern image generation models: the mention order of entities in text spuriously determines spatial layout and entity--role binding. We term this phenomenon Order-to-Space Bias (OTS) and show that it arises in both text-to-image and image-to-image generation, often overriding grounded cues and causing incorrect layouts or swapped assignments. To quantify OTS, we introduce OTS-Bench, which isolates order effects with paired prompts differing only in entity order and evaluates models along two dimensions: homogenization and correctness. Experiments show that Order-to-Space Bias (OTS) is widespread in modern image generation models, and provide evidence that it is primarily data-driven and manifests during the early stages of layout formation. Motivated by this insight, we show that both targeted fine-tuning and early-stage intervention strategies can substantially reduce OTS, while preserving generation quality.
title Order Is Not Layout: Order-to-Space Bias in Image Generation
topic Computation and Language
Artificial Intelligence
Computer Vision and Pattern Recognition
Multimedia
url https://arxiv.org/abs/2603.03714