AI art in architecture
Open accessThe DOI matches its Crossref record; 24 of the 24 references it cites were found in a registry or an index; 1 of the 2 sources its claims cite were checked for support.
Abstract
Recent diffusion-based AI art platforms can create impressive images from simple text descriptions. This makes them powerful tools for concept design in any discipline that requires creativity in visual design tasks. This is also true for early stages of architectural design with multiple stages of ideation, sketching and modelling. In this paper, we investigate how applicable diffusion-based models already are to these tasks. We research the applicability of the platforms Midjourney, DALL $$\cdot$$ · E 2 and Stable Diffusion to a series of common use cases in architectural design to determine which are already solvable or might soon be. Our novel contributions are: (i) a comparison of the capabilities of public AI art platforms; (ii) a specification of the requirements for AI art platforms in supporting common use cases in civil engineering and architecture; (iii) an analysis of 85 million Midjourney queries with Natural Language Processing (NLP) methods to extract common usage patterns. From this we derived (iv) a workflow for creating images for interior designs and (v) a workflow for creating views for exterior design that combines the strengths of the individual platforms.
Overview
CoChat Summary
TL;DR
This paper evaluates diffusion-based AI art platforms (Midjourney, DALL·E 2, Stable Diffusion) for architectural design use cases and analyzes how users already query them. It finds strong support for ideation, style variants, and interior/exterior concept imagery, but weak semantic understanding for construction plans and complex spatial tasks.
Key findings
- From about 85 million Midjourney queries collected over one year since Jan. 30, 2022, 5.7 million (6.7%) had potential architectural intent, and 2.2 million (2.6%) explicitly contained “architect,” “interior,” or “exterior” design.
- “Architecture” was the most popular keyword, and “interior” was six times more popular than “exterior” as a keyword; only 13.1% of all unique queries were upscaled or remastered.
- Users rarely succeeded in one shot: 54% of unique queries were run only once, while upscaled queries required about 6.75 steps and remastered queries about 5.1 iterations.
- In the interior-design case study, Stable Diffusion performed best overall, while Midjourney offered similar final quality but weaker initial selections; the paper proposes a combined exterior-design workflow using Midjourney for ideation, DALL·E for in-/outpainting, and Stable Diffusion for further inpainting.
Methods
- The authors qualitatively compared Midjourney, DALL·E 2, and Stable Diffusion v2.1 by their supported features (txt2img, img2img, in-/outpainting, editing, upscaling, semantics) and mapped these to architectural use cases.
- They analyzed approximately 85 million public Midjourney queries using NLP: filtering for architectural keywords and 941 famous architects, then building a Word2Vec model to examine term frequency, co-occurrence, query length, and iteration workflows.
- They conducted case studies for interior design across the three platforms and for exterior design using a combined workflow, with prompts and refinement steps shown in figures.
Limitations
- The paper states that current AI models are trial-and-error tools, often fail on complex prompts, lack semantic understanding of image content (e.g., floor plans are not just lines), and are bad at counting and complex spatial arrangements beyond foreground/background.
- The paper states that Midjourney datasets are not publicly available for data privacy reasons, and that little information exists on training data for DALL·E and Midjourney.
- Evident from the text: the query analysis covers only Midjourney, the case studies use Midjourney v3 (with remastering noted as usually skippable in v4), and the evaluation of output quality is largely qualitative and subjective rather than a systematic quantitative benchmark.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
1 Valid · 19 Not valid · 1 Not checked · 68 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Nonetheless, many variants that users explore fail entirely.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“these queries are usually refined in more than 4 iterations in mean;”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Queries that remain in draft mode require about 3.7 steps.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The low red bar shows that they are usually not used in architectural context but for their art contributions.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Midjourney starts out with several results that are stylized or of strange perspective, visible in Fig. 3(a, b).”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
28 more claims
Create a free account to see all 89 claims and every source verdict.
- Probably not a substantial claim. 44
- It may state two separate claims as one. 25
Citations
Cites
Distributed Representations of Words and Phrases and their Compositionality
Mikolov, T., Sutskever, I., Chen, K., Corrado, G., & Dean, J. (2013) · arXiv · doi:10.48550/arxiv.1310.4546 · Cited by 24,533 · cited 1×
““a”, “and”, “the”) and build a Word2Vec model (Mikolov et al., 2013) from these queries to get a model of the occurrence and co-occurrence of terms in these queries.”
BackgroundLanguage Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372 · cited 1×
“GPT-3 is trained to generate text that completes a textual input query, more precisely it predicts the next possible combinations of words to an input text.”
BackgroundHigh-Resolution Image Synthesis with Latent Diffusion Models
Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01042 · Cited by 20,302 · cited 1×
“However, Stable Diffusion was trained on the LAION-5B dataset (Rombach et al., 2022), which is based on image and text data scraped from the web.”
MethodAdvances in Neural Information Processing Systems 19
(2007) · The MIT Press eBooks · doi:10.52202/085713 · Cited by 15,539
Learning Transferable Visual Models From Natural Language Supervision
Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al. (2021) · arXiv · doi:10.48550/arxiv.2103.00020 · Cited by 15,293 · cited 1×
“Perhaps the currently most advanced incarnation of the technology is DALL-E 2 (from here on out simply referred to as DALL-E), based on the unClip method developed in [7].”
MethodDenoising Diffusion Probabilistic Models
Ho, J., Jain, A., & Abbeel, P. (2020) · arXiv · doi:10.48550/arxiv.2006.11239 · Cited by 8,712
Deep Unsupervised Learning using Nonequilibrium Thermodynamics
Sohl-Dickstein, J., Weiss, E. A., Maheswaranathan, N., & Ganguli, S. (2015) · arXiv · doi:10.48550/arxiv.1503.03585 · Cited by 2,637
Hierarchical Text-Conditional Image Generation with CLIP Latents
Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022) · arXiv · doi:10.48550/arxiv.2204.06125 · Cited by 2,511
Image Super-Resolution Via Iterative Refinement
Saharia, C., Ho, J., Chan, W., Salimans, T., Fleet, D. J., & Norouzi, M. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3204461 · Cited by 1,829 · cited 1×
“Is another diffusion step applied to add more details to the image in a higher resolution then it is up-scaled (also called super-resolution [11]).”
BackgroundImage Super-Resolution via Iterative Refinement
Saharia, C., Ho, J., Chan, W., Salimans, T., Fleet, D. J., & Norouzi, M. (2021) · arXiv · doi:10.48550/arxiv.2104.07636 · Cited by 1,829
Cited by
A Survey on Multimodal Large Language Models for Autonomous Driving
Cui, C., Ma, Y., Cao, X., Ye, W., Zhou, Y., Liang, K., Chen, J., Lu, J., Yang, Z., Liao, K.-D., et al. (2024) · 2024 IEEE/CVF Winter Conference on Applications of Computer Vision Workshops (WACVW) · doi:10.1109/wacvw60836.2024.00106 · Cited by 369
“They showcase the potential for us-ing multimodal models in many application scenarios such as healthcare [97], civil engineering [133], robotics [71] and, art [80].”
BackgroundJang, S., Lee, G., Oh, J., Lee, J., & Koo, B. (2024) · Advanced Engineering Informatics · doi:10.1016/j.aei.2024.102532 · Cited by 77
Fareed, M. W., Nassif, A. B., & Nofal, E. (2024) · Heritage · doi:10.3390/heritage7030081 · Cited by 75
“At the same time, Midjourney tends towards a more artistic style, especially in earlier versions [38].”
BackgroundFitriawijaya, A., & Jeng, T. (2024) · Buildings · doi:10.3390/buildings14082533 · Cited by 44
“In comparison, Ploenings and Berger [11] noted that generative AI has shown promise in automating aspects of design generation.”
BackgroundKomatina, D. F., Miletić, M., & Ružičić, M. M. (2024) · Buildings · doi:10.3390/buildings14082578 · Cited by 48
Metadata
- Authors
- Ploennigs, J., & Berger, M.
- Affiliations
- University of Rostock (DE)
- Corresponding author
- Ploennigs, Joern
- Venue
- AI in Civil Engineering
- Published
- 2023-08-17
- Volume
- 2
- Issue
- 1
- ISSN
- 2097-0943 (ISSN-L)
- 2730-5392
- OpenAlex
- W4385953069
- Semantic Scholar
- 951016af892f5ecde12b2acd9374b92b66afb796
- Cited by
- 153
- References
- 17
- Citations by year
- 2026: 35 · 2025: 45 · 2024: 53 · 2023: 20
- FWCI
- 13.91
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- AI in Civil Engineering (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University) (cc-by)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Generative Adversarial Networks and Image Synthesis
- Physical Sciences › Engineering › Architecture › Architecture and Computational Design
- Physical Sciences › Computer Science › Computer Graphics and Computer-Aided Design › Digital Media and Visual Art
- Keywords
- AI-generated art, diffusion models, architectural design, conceptual design, Midjourney, DALL-E, Stable Diffusion, text-to-image generation, interior design, exterior design, design ideation, query analysis
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: AI art in architecture. https://app.cochat.ai/papers/10.1007/s43503-023-00018-y
- The paper
- Ploennigs, J., & Berger, M. (2023). AI art in architecture. AI in Civil Engineering. https://doi.org/10.1007/s43503-023-00018-y
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.