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

Yongkang Zhang; Zonglin Zhao; Yuechen Zhang; Fei Ding; Pei Li; Wenxuan Wang

cs.CLMar 4, 2026

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

Authors: Yongkang Zhang, Zonglin Zhao, Yuechen Zhang, Fei Ding, Pei Li, Wenxuan Wang

Organizations: Renmin University of China, China · Huazhong Agricultural University, China · Huazhong University of Science and Technology, China · Jiangnan University, China ·

Abstract

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.

Figures & tables

Figure 1: Order-to-Space Bias makes models treat mention order as a cue for spatial layout and role binding. (a) With neutral prompts, the first-mentioned entity tends to appear on the left, and under-specified edits are often applied to a default side. (b) When order conflicts with grounded cues, the model follows order instead, producing incorrect layouts or role inversions.

Figure 2: Overview of OTS-Bench . We evaluate OTS in four settings: T2I and I2I, each with homogenization and correctness.

Component	Count	Source
Subjects
Humans	64	SOC 2018
Animals	25	COCO, ImageNet
Objects	49	COCO, OpenImages
Subjects Total	138
Actions & States

Table 1: OTS-Bench taxonomy overview. Subject and action/state inventories are curated with reference to COCO ( Lin et al., 2014 ) , ImageNet ( Deng et al., 2009 ) , OpenImages ( Kuznetsova et al., 2020 ) , AVA ( Gu et al., 2018 ) , VerbNet ( Schuler, 2005 ) , Animal Kingdom ( Ng et al., 2022 ) , and Something-Something ( Goyal et al., 2017 ) .

Modality	Homogenization	Correctness	Total
T2I	1,700	400	2,100
I2I	1,100	1,100	2,200
Total	2,800	1,500	4,300

Table 2: OTS-Bench summary.

Figure 3: Multi-agent pipeline for generating constrained test cases. Task Explorer proposes candidates, Insight Synthesizer validates real-world constraints, and Diversity Architect produces structured data.

Model	Homogenization		T2I Correctness (%)		I2I Correctness (%)		Cohen’s $\kappa$ $\uparrow$
Model	T2I	I2I	Ali	Rev	Ali	Rev
llama-3.2-11b-vision	97.1	73.3	97.0	2.4	89.7	87.6	0.62
Qwen3-VL-8B-Instruct	94.2	63.9	85.3	12.5	90.4	83.7	0.81
InternVL3-8B	90.2	59.8	84.8	24.3	94.9	92.9	0.72
llava-critic-r1-7b	54.9	60.2	91.4	23.8	96.9	95.2	0.47
Human	90.3	68.7	82.0	9.2	89.8	82.5	–

Table 3: Alignment results between VL judges and human annotations. Homogenization : deviation score defined as

\mathrm{Hom}=|2p-100|

, where

p

is the percentage of valid samples assigned to label 1 (valid samples exclude label 0). Correctness : accuracy on valid samples under Ali and Rev . Cohen’s

\kappa

: computed on the full 2,400-sample 3-class confusion matrix (labels 0/1/2).

Model	Homogenization $\downarrow$		T2I Correctness $\uparrow$		$\Delta\downarrow$	I2I Correctness $\uparrow$		$\Delta\downarrow$
Model	T2I	I2I	Ali	Rev	T2I	Ali	Rev	I2I
SDXL	52.6	55.8	83.3	23.6	59.7	73.5	62.2	11.3
SD3.5	84.2	38.4	84.8	21.1	63.7	79.5	71.8	7.7
FLUX-dev	88.8	42.6	79.8	24.7	55.1	84.3	76.8	7.5
Qwen-Image	91.6	83.4	81.8	28.2	53.6	88.1	78.8	9.3
DALL-E 3	70.4	–	87.7	18.2	69.5	–	–	–

Table 4: OTS-Bench main results. Homogenization measures deviation from a balanced 50/50 split in left–right outcomes:

\text{Hom}=|2p-100|

where

p

is the percentage of valid samples assigned to the left (lower is better). Correctness reports accuracy when prompt order aligns ( Ali ) or contradicts ( Rev ) the grounding constraint. We report degradation

\Delta=\text{Ali}-\text{Rev}

as a diagnostic of order sensitivity.

Dataset	Samples	OTS-align (%)
LAION-2B-en-aesthetic	23,020	89.2
DataComp-Large	10,150	87.2

Table 5: Order-to-space alignment in web-scale corpora. Samples is the number of valid caption–image pairs after filtering. OTS-align is the proportion of cases where the caption order (first entity, second entity) matches the image layout (left entity, right entity).

Model	Homogenization $\downarrow$		T2I Correctness $\uparrow$		I2I Correctness $\uparrow$		Image Quality $\uparrow$
Model	T2I	I2I	Ali	Rev	Ali	Rev	ImageReward
FLUX-dev	88.8	42.6	79.8	24.7	84.3	76.8	$0.186_{\scriptsize\pm 0.672}$
FLUX-dev (SFT)	47.4	25.0	83.2	49.3	87.0	79.5	$0.217_{\scriptsize\pm 0.687}$
Qwen-Image	91.6	83.4	81.8	28.2	88.1	78.8	$0.204_{\scriptsize\pm 0.732}$
Qwen-Image (SFT)	50.8	32.8	84.7	57.7	90.5	80.1	$0.223_{\scriptsize\pm 0.717}$

Table 6: Merged comparison of homogenization, correctness, and image quality for FLUX-dev and Qwen-Image (with LoRA variants). Homogenization is lower-is-better (

\downarrow

); correctness and image quality are higher-is-better (

\uparrow

). Image quality is reported with ImageReward (mean

\pm

std).

(a) Switch direction: Successful Rate (%)
	Switch step
Setting	base	1	2	3	4	5
A2B	100.0	25.0	14.8	11.1	3.2	2.2
B2A	100.0	23.0	17.8	5.5	2.4	1.2
(b) Prompt transition: Homogenization (%)
N2A	89.9	28.8	21.8	27.2	34.1	39.1

Table 7: Temporal effects of order cues. (a) A2B/B2A: switching direction. (b) N2A/N2B: switching from neutral to rich prompts with different entity orders. Values are in %.

Appendix figures & tables21 assets

Supplementary material from the paper’s appendix.

Appendix

Type	Category	Count	Examples	Source
Humans	Healthcare	8	physician, nurse, dentist, pharmacist	SOC 2018
	Education	8	teacher, professor, librarian, counselor	SOC 2018
	Engineering	8	civil engineer, architect, electrician, mechanical engineer	SOC 2018
	Food Service	8	chef, waiter, barista, bartender	SOC 2018
	Management	8	manager, financial manager, construction manager, sales manager	SOC 2018
	Arts & Media	8	designer, photographer, musician, artist	SOC 2018

Appendix

Table 8: OTS-Bench entity taxonomy with category breakdown and representative examples.

Type	Category	Count	Examples
Human Actions	Upper Limbs	6	raising hand, waving, pointing, clapping, saluting, crossing arms
	Torso & Lower Limbs	9	leaning forward, stretching up, stepping forward, kneeling, squatting, lifting leg, jumping, dancing, running pose
	Interactive Actions	10	pointing at, scolding, patting shoulder, examining, photographing, instructing
	Parallel Actions	10	standing vs sitting, raising hand vs saluting, waving vs clapping, looking up vs down, leaning forward vs backward
	Subtotal	35	Individual + interactive + parallel actions
Animal Actions	Static Postures	4	sitting, standing, lying down, crouching

Appendix

Table 9: Action library with comprehensive counts (unique actions; no double-counting).

Modality	Dimension	H–H	A–A	O–O	H–A	H–O	A–O	Total
T2I	Homogenization	500	300	300	200	200	200	1700
T2I	Correctness	–	–	–	–	–	–	400
I2I	Homogenization	500	300	300	–	–	–	1100
I2I	Correctness	500	300	300	–	–	–	1100
Total								4300

Appendix

Table 10: Full OTS-Bench breakdown by modality, evaluation dimension, and entity-pair type.

Agent Role	System Instruction
Prompt Scout	Objective: Generate prompts that expose sequence-to-spatial bias in text-to-image models. Critical Understanding: Sequence-to-spatial bias occurs when models incorrectly map text order to spatial positions (first→left, second→right). Valid test cases must have real-world layouts that contradict text order. Example: “gas pedal and brake pedal in a car”—reality has gas(right) and brake(left), but biased models generate gas(left) and brake(right) by following text order. Requirements: (1) Each prompt must test a verifiable spatial constraint (keyboard keys, anatomy, vehicle controls, etc.); (2) Elements A and B must have fixed positions in the given context; (3) Only left/right semantics allowed (no top/bottom, front/back); (4) Domain diversity: aim for 10+ distinct domains; (5) Total target: 400 test prompts. Deliverable: (1) Mission restatement confirming bias detection objective; (2) Draft prompt list with clear rationale for each domain; (3) Handoff notes for Constraint Critic highlighting coverage gaps. Archive requirement: label each round clearly for version control.
Constraint Critic	Objective: Review Prompt Scout’s draft and provide strict, objective feedback. Evaluation Criteria: (1) Spatial validity—does each A-B pair have a verifiable real-world left/right constraint? (2) Bias exposure—will the prompt actually trigger sequence-to-spatial errors? (3) Domain coverage—progress toward 10+ domains and 400 prompts; (4) Constraint adherence—only left/right axis used. Required Output: (1) Quantitative metrics (domain count, constraint coverage, total prompts, archive status); (2) Concrete strengths and weaknesses of current draft; (3) Numbered directives for Benchmark Curator; (4) Strict scores (0–5) for diversity, coverage, failure-exposure, and archive completeness. Critical: Flag any prompts where left/right positions are arbitrary or undefined.
Benchmark Curator	Objective: Rewrite benchmark based on Constraint Critic directives. Output Format: JSON per domain containing: domain name, round number, seed pairs with elements A/B, context, constraint type (must be “left_right”), semantic basis, difficulty level (obvious/subtle/tricky), spatial reality (A/B positions), rationale citing verifiable standards, sequence bias failure mode, test prompts with expected errors, and status (new/revised/retired). Critical Rules: (1) spatial_reality must cite verifiable standards (QWERTY layout, anatomy atlases, traffic codes, etc.); (2) Each test prompt must demonstrate how sequence bias causes spatial error; (3) Mark prompts as new/revised/retired vs previous round; (4) Concise format: “A and B in/on/at Context” (single sentence). Validation: Reject any pair where left/right positions are arbitrary or context-independent.

Appendix

Table 11: System instructions for the three-agent benchmark construction pipeline. Each agent receives role-specific prompts that enforce left-right constraint validation and bias detection objectives.

Setting	Value
Models	FLUX.1-dev, Qwen-Image
Trainable modules	UNet only (text encoder frozen)
LoRA (linear)	rank=32, alpha=32
LoRA (conv)	rank=16, alpha=16
Training steps	10,000
Batch size	1

Appendix

Table 12: Key hyperparameters for anti-OTS LoRA-SFT.

	Qwen3-VL-8B-Instruct ( $\kappa=0.81$ )				llama-3.2-11b-vision ( $\kappa=0.62$ )				InternVL3-8B ( $\kappa=0.72$ )				llava-critic-r1-7b ( $\kappa=0.47$ )
Human	0	1	2	Total	0	1	2	Total	0	1	2	Total	0	1	2	Total
0	224	11	4	239	207	19	13	239	209	19	11	239	183	37	19	239
1	94	1475	30	1599	103	1377	119	1599	96	1471	32	1599	92	1342	165	1599
2	35	56	471	562	31	178	353	562	18	154	390	562	19	281	262	562
Total	353	1542	505	2400	341	1574	485	2400	323	1644	433	2400	294	1660	446	2400

Appendix

Table 13: Full 3-class confusion matrices (labels 0/1/2) between human annotations (rows) and VL judges (columns) on the 2,400-sample validation set. Each judge occupies a 4

\times

4 block (including marginal totals). Cohen’s

\kappa

is computed on the 3

\times

3 core (excluding marginals) and reported in the header.

Dimension	Objective	Output 1	Output 2	Output 3
T2I Homogenization	Determine left-right positioning	A left of B	A right of B	Invalid
T2I Correctness	Verify semantic constraint adherence	Respects	Violates	Invalid
I2I Homogenization	Identify action performer	A performs	B performs	Invalid
I2I Correctness	Validate action attribution	Reversed	Aligned	Invalid

Appendix

Table 14: Output semantics for evaluation prompts

T2I Homogenization Evaluation Prompt

Role. You are a professional image analysis assistant for spatial relationship verification. Task. Determine the left-right positional relationship between two subjects in a generated image, strictly from the observer’s perspective. Subject Definitions. • Subject A: {subject_A} • Subject B: {subject_B} Evaluation Protocol. 1. Verify that the image contains exactly two subjects corresponding to Subject A and Subject B, and that their visual characteristics match the provided definitions. 2. Determine the horizontal (left-right) spatial relationship between Subject A and Subject B from the observer’s viewpoint. 3. Focus exclusively on left-right positioning. Ignore depth (front-back), vertical (up-down), diagonal arrangements, or relative size differences. Output Specification. • Output “1” if Subject A is positioned to the LEFT of Subject B. • Output “2” if Subject A is positioned to the RIGHT of Subject B. • Output “3” if any of the following conditions apply: – The image does not contain exactly two subjects. – Subject identities or visual characteristics do not match the provided definitions. – One or both subjects are unclear, occluded, or cannot be reliably identified. – The subjects primarily exhibit depth-based or vertical relationships rather than a clear horizontal separation. – The left-right positional relationship cannot be confidently determined. Critical Instruction. You must output only a single digit : 1, 2, or 3. Do not include any explanations, reasoning steps, or additional text.

Appendix

Table 15: T2I Homogenization Evaluation Prompt

T2I Correctness Evaluation Prompt
You are a professional image analysis assistant for semantic constraint verification.
TASK
Verify whether the spatial arrangement in a generated image respects real-world semantic constraints rather than textual order.
SUBJECT DEFINITIONS
• Subject A: {subject_A} (expected position: LEFT according to semantic constraints) • Subject B: {subject_B} (expected position: RIGHT according to semantic constraints) EVALUATION PROTOCOL
1. Verify that both subjects appear clearly and are recognizable in the image. 2. Determine the horizontal (left-right) spatial relationship between the subjects from the observer’s perspective. 3. Compare the observed spatial arrangement against the expected semantic constraints. OUTPUT SPECIFICATION

Appendix

Table 16: T2I Correctness Evaluation Prompt

I2I Homogenization Evaluation Prompt
You are a professional image analysis assistant for action attribution verification.
TASK
Determine which subject performed the specified action in an image-to-image edited result.
REFERENCE SUBJECTS (from Seed Image)
• Subject A: {subject_A} ({description_A}) • Subject B: {subject_B} ({description_B}) ACTION SPECIFICATION
{action}

Appendix

Table 17: I2I Homogenization Evaluation Prompt

I2I Correctness Evaluation Prompt
You are a professional image analysis assistant for action attribution correctness verification.
TASK
Verify whether action attribution in an image-to-image edited result aligns with the prompt specification rather than the spatial order in the seed image.
REFERENCE SUBJECTS (from Seed Image)
• Subject A: {subject_A} ({description_A}) • Subject B: {subject_B} ({description_B}) ACTION SPECIFICATION
{action}

Appendix

Table 18: I2I Correctness Evaluation Prompt

Panel of Figure 4 — Figure 4: Representative outputs for T2I homogenization labeling. Output 1/2 correspond to valid left–right layouts, while Output 3 illustrates invalid generations.

Panel of Figure 5 — Figure 5: Representative T2I examples illustrating order-to-space effects. Row 1 shows correct spatial grounding. Row 2 demonstrates persistent left/right reversals even when spatial constraints are explicitly specified in the prompt. Row 3 contains invalid or unscorable cases excluded from evaluation.

Panel of Figure 6 — Figure 6: Representative outputs for I2I homogenization labeling. In each case, the left image is the Seed and the right image is the Edited result. Output 1/2 correspond to valid cases where the under-specified edit is assigned to Subject A vs. Subject B, while Output 3 illustrates invalid generations excluded from scoring.

Panel of Figure 7 — Figure 7: Representative examples for I2I Harm (Aligned) evaluation. In each case, the left image is the Seed and the right image is the Generated/Edited result under aligned prompt order. Label 1: correct attribution; Label 2: order-induced role/action swapping; Label 3: invalid generations excluded from scoring.

Panel of Figure 8 — Figure 8: Representative examples for I2I Harm (Reverse) evaluation. In each case, the left image is the Seed and the right image is the Generated/Edited result under reverse prompt order. Label 1: correct attribution; Label 2: order-induced role/action swapping; Label 3: invalid generations excluded from scoring.

Model	Overall	Prompt Align.	Fidelity
FLUX-dev	$4.37_{\scriptsize\pm 1.36}$	$3.69_{\scriptsize\pm 1.34}$	$4.93_{\scriptsize\pm 1.48}$
FLUX-dev (SFT)	$4.65_{\scriptsize\pm 1.28}$	$4.53_{\scriptsize\pm 1.31}$	$4.86_{\scriptsize\pm 1.24}$
Qwen-Image	$4.41_{\scriptsize\pm 1.19}$	$3.32_{\scriptsize\pm 0.96}$	$5.08_{\scriptsize\pm 1.42}$
Qwen-Image (SFT)	$4.79_{\scriptsize\pm 1.41}$	$4.03_{\scriptsize\pm 1.45}$	$5.24_{\scriptsize\pm 1.39}$

Appendix

Table 19: Human evaluation of image quality on a held-out set of 400 matched outputs per model variant (1–7 scale; higher is better). We report mean with standard deviation in subscript.

Panel of Figure 9 — Figure 9: Representative outputs for T2V homogenization labeling. Output 1/2 correspond to valid generations under original/reversed prompt order, while Output 3 illustrates invalid generations excluded from scoring.

Panel of Figure 10 — Figure 10: Representative outputs for I2V homogenization labeling. Each example shows the Seed frame (top) and a video preview (bottom). Output 1/2 are valid cases where an under-specified edit/action is assigned to Subject A vs. Subject B, while Output 3 illustrates invalid generations excluded from scoring.

Panel of Figure 11 — Figure 11: Examples of vision-language understanding. Each case follows the same system prompt and reports the model’s one-sentence description.

Panel of Figure 12 — Figure 12: Temporal intervention on order-sensitive conditioning during diffusion sampling. We switch entity order from a neutral prompt to a rich prompt at different denoising steps. Case 1 and Case 2 illustrate that order cues injected during the early, layout-sensitive stage can reconfigure spatial assignment, while Case 3 shows unstable behavior when switching near the boundary of layout formation.

T2I Generation

Text-to-image generation creates images from natural-language prompts using generative models, with research focused on fidelity, control, efficiency, and safety.

See all 299 papers

Can Text-to-Image Models Draw from the Right Frame of Reference?
Aug 4, 2026Zheyuan Gu, Ruihang Li, Yong Huang +5Vision-Language Models T2I Generation
1ERNIE Team, Baidu Inc. · 2Peking University
Enhancing Spatial Understanding in Image Generation via Reward Modeling
Feb 27, 2026Zhenyu Tang, Chaoran Feng, Yufan Deng +5T2I Generation Reward Modeling
Peking University · ByteDance Seed
Think, Plan, Paint: Layout-Aware Reasoning for Controllable Image Generation in Unified Models
Jul 17, 2026Junhao Liu, Jian-Wei Zhang, Tao Huang +3Unified Multimodal Models Image Generation
Hunyuan, Tencent · School of Computer Science, Peking University

Aug 4, 2026J/K move · Enter open · S save

Can Text-to-Image Models Draw from the Right Frame of Reference?

Zheyuan Gu, Ruihang Li, Yong Huang, Yiqian Zhang+4

1ERNIE Team, Baidu Inc. · 2Peking University

Vision-Language Models T2I Generation Compositional T2I Generation T2I Generation Evaluation Spatial Reasoning

Open paper AskPDF arXiv

Spatial instruction following has become a crucial requirement for text-to-image (T2I) generation. A common challenge arises when directional expressions are interpreted under different frames of reference. For example, ``the left of'' may refer to the viewer's image coordinates or to the intrinsic orientation of an object, leading to different expected layouts. Existing T2I benchmarks reveal important layout failures, yet they rarely isolate whether models can follow a specified frame of reference when it differs from camera view. To mitigate this gap, we introduce FoR-T2I, a benchmark for evaluating this distinction with 1,200 prompt pairs built from controlled spatial layouts. In each pair, the camera-view (Cam) prompt states the target relation in camera view, while the frame-of-reference (FoR) prompt describes the same target placement through an oriented anchor object. Across 22 closed-source and open-source T2I models, mean final accuracy is 41.8% lower on FoR prompts than on matched Cam prompts; even the best-performing model achieves only 44.3% FoR accuracy. This suggests that current models struggle more when the same layout is described through an object's orientation rather than directly in image coordinates. We further analyze this gap by relation type and camera view, compare several training-free prompting and feedback-based mitigation strategies, and propose a VLM-gated rewriting approach that selects rewritten prompts using visual feedback, improving average FoR accuracy from 25.0% to 29.2% under the same generation budget.

Abstract

T2I Generation

Explore similar work

Can Text-to-Image Models Draw from the Right Frame of Reference?

Enhancing Spatial Understanding in Image Generation via Reward Modeling

Think, Plan, Paint: Layout-Aware Reasoning for Controllable Image Generation in Unified Models