Taming the Tri-Space Tension: ARC-Guided Hallucination Modeling and Control for Text-to-Image Generation Article Swipe
YOU?
·
· 2025
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2507.04946
Despite remarkable progress in image quality and prompt fidelity, text-to-image (T2I) diffusion models continue to exhibit persistent "hallucinations", where generated content subtly or significantly diverges from the intended prompt semantics. While often regarded as unpredictable artifacts, we argue that these failures reflect deeper, structured misalignments within the generative process. In this work, we propose a cognitively inspired perspective that reinterprets hallucinations as trajectory drift within a latent alignment space. Empirical observations reveal that generation unfolds within a multiaxial cognitive tension field, where the model must continuously negotiate competing demands across three key critical axes: semantic coherence, structural alignment, and knowledge grounding. We then formalize this three-axis space as the Hallucination Tri-Space and introduce the Alignment Risk Code (ARC): a dynamic vector representation that quantifies real-time alignment tension during generation. The magnitude of ARC captures overall misalignment, its direction identifies the dominant failure axis, and its imbalance reflects tension asymmetry. Based on this formulation, we develop the TensionModulator (TM-ARC): a lightweight controller that operates entirely in latent space. TM-ARC monitors ARC signals and applies targeted, axis-specific interventions during the sampling process. Extensive experiments on standard T2I benchmarks demonstrate that our approach significantly reduces hallucination without compromising image quality or diversity. This framework offers a unified and interpretable approach for understanding and mitigating generative failures in diffusion-based T2I systems.
Related Topics
- Type
- preprint
- Language
- en
- Landing Page
- http://arxiv.org/abs/2507.04946
- https://arxiv.org/pdf/2507.04946
- OA Status
- green
- OpenAlex ID
- https://openalex.org/W4415163575
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W4415163575Canonical identifier for this work in OpenAlex
- DOI
-
https://doi.org/10.48550/arxiv.2507.04946Digital Object Identifier
- Title
-
Taming the Tri-Space Tension: ARC-Guided Hallucination Modeling and Control for Text-to-Image GenerationWork title
- Type
-
preprintOpenAlex work type
- Language
-
enPrimary language
- Publication year
-
2025Year of publication
- Publication date
-
2025-07-07Full publication date if available
- Authors
-
Jianjiang Yang, Ziyan Huang, Yanrong Li, Da Peng, Hong YaoList of authors in order
- Landing page
-
https://arxiv.org/abs/2507.04946Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2507.04946Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2507.04946Direct OA link when available
- Cited by
-
0Total citation count in OpenAlex
Full payload
| id | https://openalex.org/W4415163575 |
|---|---|
| doi | https://doi.org/10.48550/arxiv.2507.04946 |
| ids.doi | https://doi.org/10.48550/arxiv.2507.04946 |
| ids.openalex | https://openalex.org/W4415163575 |
| fwci | |
| type | preprint |
| title | Taming the Tri-Space Tension: ARC-Guided Hallucination Modeling and Control for Text-to-Image Generation |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| topics[0].id | https://openalex.org/T10799 |
| topics[0].field.id | https://openalex.org/fields/17 |
| topics[0].field.display_name | Computer Science |
| topics[0].score | 0.8454999923706055 |
| topics[0].domain.id | https://openalex.org/domains/3 |
| topics[0].domain.display_name | Physical Sciences |
| topics[0].subfield.id | https://openalex.org/subfields/1707 |
| topics[0].subfield.display_name | Computer Vision and Pattern Recognition |
| topics[0].display_name | Data Visualization and Analytics |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| language | en |
| locations[0].id | pmh:oai:arXiv.org:2507.04946 |
| locations[0].is_oa | True |
| locations[0].source.id | https://openalex.org/S4306400194 |
| locations[0].source.issn | |
| locations[0].source.type | repository |
| locations[0].source.is_oa | True |
| locations[0].source.issn_l | |
| locations[0].source.is_core | False |
| locations[0].source.is_in_doaj | False |
| locations[0].source.display_name | arXiv (Cornell University) |
| locations[0].source.host_organization | https://openalex.org/I205783295 |
| locations[0].source.host_organization_name | Cornell University |
| locations[0].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[0].license | |
| locations[0].pdf_url | https://arxiv.org/pdf/2507.04946 |
| locations[0].version | submittedVersion |
| locations[0].raw_type | text |
| locations[0].license_id | |
| locations[0].is_accepted | False |
| locations[0].is_published | False |
| locations[0].raw_source_name | |
| locations[0].landing_page_url | http://arxiv.org/abs/2507.04946 |
| locations[1].id | doi:10.48550/arxiv.2507.04946 |
| locations[1].is_oa | True |
| locations[1].source.id | https://openalex.org/S4306400194 |
| locations[1].source.issn | |
| locations[1].source.type | repository |
| locations[1].source.is_oa | True |
| locations[1].source.issn_l | |
| locations[1].source.is_core | False |
| locations[1].source.is_in_doaj | False |
| locations[1].source.display_name | arXiv (Cornell University) |
| locations[1].source.host_organization | https://openalex.org/I205783295 |
| locations[1].source.host_organization_name | Cornell University |
| locations[1].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[1].license | |
| locations[1].pdf_url | |
| locations[1].version | |
| locations[1].raw_type | article |
| locations[1].license_id | |
| locations[1].is_accepted | False |
| locations[1].is_published | |
| locations[1].raw_source_name | |
| locations[1].landing_page_url | https://doi.org/10.48550/arxiv.2507.04946 |
| indexed_in | arxiv, datacite |
| authorships[0].author.id | https://openalex.org/A5102954986 |
| authorships[0].author.orcid | https://orcid.org/0000-0002-6061-4552 |
| authorships[0].author.display_name | Jianjiang Yang |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Yang, Jianjiang |
| authorships[0].is_corresponding | False |
| authorships[1].author.id | https://openalex.org/A5107946738 |
| authorships[1].author.orcid | |
| authorships[1].author.display_name | Ziyan Huang |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Huang, Ziyan |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A5108053777 |
| authorships[2].author.orcid | https://orcid.org/0000-0002-2015-028X |
| authorships[2].author.display_name | Yanrong Li |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | li, Yanshu |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A5078754843 |
| authorships[3].author.orcid | |
| authorships[3].author.display_name | Da Peng |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Peng, Da |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A5101151232 |
| authorships[4].author.orcid | https://orcid.org/0009-0000-7967-8962 |
| authorships[4].author.display_name | Hong Yao |
| authorships[4].author_position | last |
| authorships[4].raw_author_name | Yao, Huaiyuan |
| authorships[4].is_corresponding | False |
| has_content.pdf | False |
| has_content.grobid_xml | False |
| is_paratext | False |
| open_access.is_oa | True |
| open_access.oa_url | https://arxiv.org/pdf/2507.04946 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-10-14T00:00:00 |
| display_name | Taming the Tri-Space Tension: ARC-Guided Hallucination Modeling and Control for Text-to-Image Generation |
| has_fulltext | False |
| is_retracted | False |
| updated_date | 2025-11-06T06:51:31.235846 |
| primary_topic.id | https://openalex.org/T10799 |
| primary_topic.field.id | https://openalex.org/fields/17 |
| primary_topic.field.display_name | Computer Science |
| primary_topic.score | 0.8454999923706055 |
| primary_topic.domain.id | https://openalex.org/domains/3 |
| primary_topic.domain.display_name | Physical Sciences |
| primary_topic.subfield.id | https://openalex.org/subfields/1707 |
| primary_topic.subfield.display_name | Computer Vision and Pattern Recognition |
| primary_topic.display_name | Data Visualization and Analytics |
| cited_by_count | 0 |
| locations_count | 2 |
| best_oa_location.id | pmh:oai:arXiv.org:2507.04946 |
| best_oa_location.is_oa | True |
| best_oa_location.source.id | https://openalex.org/S4306400194 |
| best_oa_location.source.issn | |
| best_oa_location.source.type | repository |
| best_oa_location.source.is_oa | True |
| best_oa_location.source.issn_l | |
| best_oa_location.source.is_core | False |
| best_oa_location.source.is_in_doaj | False |
| best_oa_location.source.display_name | arXiv (Cornell University) |
| best_oa_location.source.host_organization | https://openalex.org/I205783295 |
| best_oa_location.source.host_organization_name | Cornell University |
| best_oa_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| best_oa_location.license | |
| best_oa_location.pdf_url | https://arxiv.org/pdf/2507.04946 |
| best_oa_location.version | submittedVersion |
| best_oa_location.raw_type | text |
| best_oa_location.license_id | |
| best_oa_location.is_accepted | False |
| best_oa_location.is_published | False |
| best_oa_location.raw_source_name | |
| best_oa_location.landing_page_url | http://arxiv.org/abs/2507.04946 |
| primary_location.id | pmh:oai:arXiv.org:2507.04946 |
| primary_location.is_oa | True |
| primary_location.source.id | https://openalex.org/S4306400194 |
| primary_location.source.issn | |
| primary_location.source.type | repository |
| primary_location.source.is_oa | True |
| primary_location.source.issn_l | |
| primary_location.source.is_core | False |
| primary_location.source.is_in_doaj | False |
| primary_location.source.display_name | arXiv (Cornell University) |
| primary_location.source.host_organization | https://openalex.org/I205783295 |
| primary_location.source.host_organization_name | Cornell University |
| primary_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| primary_location.license | |
| primary_location.pdf_url | https://arxiv.org/pdf/2507.04946 |
| primary_location.version | submittedVersion |
| primary_location.raw_type | text |
| primary_location.license_id | |
| primary_location.is_accepted | False |
| primary_location.is_published | False |
| primary_location.raw_source_name | |
| primary_location.landing_page_url | http://arxiv.org/abs/2507.04946 |
| publication_date | 2025-07-07 |
| publication_year | 2025 |
| referenced_works_count | 0 |
| abstract_inverted_index.a | 54, 65, 76, 118, 158, 202 |
| abstract_inverted_index.In | 49 |
| abstract_inverted_index.We | 101 |
| abstract_inverted_index.as | 33, 61, 107 |
| abstract_inverted_index.in | 3, 164, 213 |
| abstract_inverted_index.of | 131 |
| abstract_inverted_index.on | 150, 182 |
| abstract_inverted_index.or | 22, 197 |
| abstract_inverted_index.to | 14 |
| abstract_inverted_index.we | 36, 52, 153 |
| abstract_inverted_index.ARC | 132, 169 |
| abstract_inverted_index.T2I | 184, 215 |
| abstract_inverted_index.The | 129 |
| abstract_inverted_index.and | 6, 98, 111, 143, 171, 204, 209 |
| abstract_inverted_index.for | 207 |
| abstract_inverted_index.its | 136, 144 |
| abstract_inverted_index.key | 91 |
| abstract_inverted_index.our | 188 |
| abstract_inverted_index.the | 26, 46, 82, 108, 113, 139, 155, 177 |
| abstract_inverted_index.Code | 116 |
| abstract_inverted_index.Risk | 115 |
| abstract_inverted_index.This | 199 |
| abstract_inverted_index.from | 25 |
| abstract_inverted_index.must | 84 |
| abstract_inverted_index.that | 38, 58, 72, 122, 161, 187 |
| abstract_inverted_index.then | 102 |
| abstract_inverted_index.this | 50, 104, 151 |
| abstract_inverted_index.(T2I) | 10 |
| abstract_inverted_index.Based | 149 |
| abstract_inverted_index.While | 30 |
| abstract_inverted_index.argue | 37 |
| abstract_inverted_index.axes: | 93 |
| abstract_inverted_index.axis, | 142 |
| abstract_inverted_index.drift | 63 |
| abstract_inverted_index.image | 4, 195 |
| abstract_inverted_index.model | 83 |
| abstract_inverted_index.often | 31 |
| abstract_inverted_index.space | 106 |
| abstract_inverted_index.these | 39 |
| abstract_inverted_index.three | 90 |
| abstract_inverted_index.where | 18, 81 |
| abstract_inverted_index.work, | 51 |
| abstract_inverted_index.(ARC): | 117 |
| abstract_inverted_index.TM-ARC | 167 |
| abstract_inverted_index.across | 89 |
| abstract_inverted_index.during | 127, 176 |
| abstract_inverted_index.field, | 80 |
| abstract_inverted_index.latent | 66, 165 |
| abstract_inverted_index.models | 12 |
| abstract_inverted_index.offers | 201 |
| abstract_inverted_index.prompt | 7, 28 |
| abstract_inverted_index.reveal | 71 |
| abstract_inverted_index.space. | 68, 166 |
| abstract_inverted_index.subtly | 21 |
| abstract_inverted_index.vector | 120 |
| abstract_inverted_index.within | 45, 64, 75 |
| abstract_inverted_index.Despite | 0 |
| abstract_inverted_index.applies | 172 |
| abstract_inverted_index.content | 20 |
| abstract_inverted_index.deeper, | 42 |
| abstract_inverted_index.demands | 88 |
| abstract_inverted_index.develop | 154 |
| abstract_inverted_index.dynamic | 119 |
| abstract_inverted_index.exhibit | 15 |
| abstract_inverted_index.failure | 141 |
| abstract_inverted_index.overall | 134 |
| abstract_inverted_index.propose | 53 |
| abstract_inverted_index.quality | 5, 196 |
| abstract_inverted_index.reduces | 191 |
| abstract_inverted_index.reflect | 41 |
| abstract_inverted_index.signals | 170 |
| abstract_inverted_index.tension | 79, 126, 147 |
| abstract_inverted_index.unfolds | 74 |
| abstract_inverted_index.unified | 203 |
| abstract_inverted_index.without | 193 |
| abstract_inverted_index.approach | 189, 206 |
| abstract_inverted_index.captures | 133 |
| abstract_inverted_index.continue | 13 |
| abstract_inverted_index.critical | 92 |
| abstract_inverted_index.diverges | 24 |
| abstract_inverted_index.dominant | 140 |
| abstract_inverted_index.entirely | 163 |
| abstract_inverted_index.failures | 40, 212 |
| abstract_inverted_index.inspired | 56 |
| abstract_inverted_index.intended | 27 |
| abstract_inverted_index.monitors | 168 |
| abstract_inverted_index.operates | 162 |
| abstract_inverted_index.process. | 48, 179 |
| abstract_inverted_index.progress | 2 |
| abstract_inverted_index.reflects | 146 |
| abstract_inverted_index.regarded | 32 |
| abstract_inverted_index.sampling | 178 |
| abstract_inverted_index.semantic | 94 |
| abstract_inverted_index.standard | 183 |
| abstract_inverted_index.systems. | 216 |
| abstract_inverted_index.(TM-ARC): | 157 |
| abstract_inverted_index.Alignment | 114 |
| abstract_inverted_index.Empirical | 69 |
| abstract_inverted_index.Extensive | 180 |
| abstract_inverted_index.Tri-Space | 110 |
| abstract_inverted_index.alignment | 67, 125 |
| abstract_inverted_index.cognitive | 78 |
| abstract_inverted_index.competing | 87 |
| abstract_inverted_index.diffusion | 11 |
| abstract_inverted_index.direction | 137 |
| abstract_inverted_index.fidelity, | 8 |
| abstract_inverted_index.formalize | 103 |
| abstract_inverted_index.framework | 200 |
| abstract_inverted_index.generated | 19 |
| abstract_inverted_index.imbalance | 145 |
| abstract_inverted_index.introduce | 112 |
| abstract_inverted_index.knowledge | 99 |
| abstract_inverted_index.magnitude | 130 |
| abstract_inverted_index.negotiate | 86 |
| abstract_inverted_index.real-time | 124 |
| abstract_inverted_index.targeted, | 173 |
| abstract_inverted_index.alignment, | 97 |
| abstract_inverted_index.artifacts, | 35 |
| abstract_inverted_index.asymmetry. | 148 |
| abstract_inverted_index.benchmarks | 185 |
| abstract_inverted_index.coherence, | 95 |
| abstract_inverted_index.controller | 160 |
| abstract_inverted_index.diversity. | 198 |
| abstract_inverted_index.generation | 73 |
| abstract_inverted_index.generative | 47, 211 |
| abstract_inverted_index.grounding. | 100 |
| abstract_inverted_index.identifies | 138 |
| abstract_inverted_index.mitigating | 210 |
| abstract_inverted_index.multiaxial | 77 |
| abstract_inverted_index.persistent | 16 |
| abstract_inverted_index.quantifies | 123 |
| abstract_inverted_index.remarkable | 1 |
| abstract_inverted_index.semantics. | 29 |
| abstract_inverted_index.structural | 96 |
| abstract_inverted_index.structured | 43 |
| abstract_inverted_index.three-axis | 105 |
| abstract_inverted_index.trajectory | 62 |
| abstract_inverted_index.cognitively | 55 |
| abstract_inverted_index.demonstrate | 186 |
| abstract_inverted_index.experiments | 181 |
| abstract_inverted_index.generation. | 128 |
| abstract_inverted_index.lightweight | 159 |
| abstract_inverted_index.perspective | 57 |
| abstract_inverted_index.compromising | 194 |
| abstract_inverted_index.continuously | 85 |
| abstract_inverted_index.formulation, | 152 |
| abstract_inverted_index.observations | 70 |
| abstract_inverted_index.reinterprets | 59 |
| abstract_inverted_index.Hallucination | 109 |
| abstract_inverted_index.axis-specific | 174 |
| abstract_inverted_index.hallucination | 192 |
| abstract_inverted_index.interpretable | 205 |
| abstract_inverted_index.interventions | 175 |
| abstract_inverted_index.misalignment, | 135 |
| abstract_inverted_index.misalignments | 44 |
| abstract_inverted_index.significantly | 23, 190 |
| abstract_inverted_index.text-to-image | 9 |
| abstract_inverted_index.understanding | 208 |
| abstract_inverted_index.unpredictable | 34 |
| abstract_inverted_index.hallucinations | 60 |
| abstract_inverted_index.representation | 121 |
| abstract_inverted_index.diffusion-based | 214 |
| abstract_inverted_index.TensionModulator | 156 |
| abstract_inverted_index."hallucinations", | 17 |
| cited_by_percentile_year | |
| countries_distinct_count | 0 |
| institutions_distinct_count | 5 |
| citation_normalized_percentile |