The DOI matches its Crossref record; 40 of the 40 references it cites were found in a registry or an index; 3 of the 5 sources its claims cite were checked for support.
Design Guidelines for Prompt Engineering Text-to-Image Generative Models
Open accessAbstract
Text-to-image generative models are a new and powerful way to generate visual artwork. However, the open-ended nature of text as interaction is double-edged; while users can input anything and have access to an infinite range of generations, they also must engage in brute-force trial and error with the text prompt when the result quality is poor. We conduct a study exploring what prompt keywords and model hyperparameters can help produce coherent outputs. In particular, we study prompts structured to include subject and style keywords and investigate success and failure modes of these prompts. Our evaluation of 5493 generations over the course of five experiments spans 51 abstract and concrete subjects as well as 51 abstract and figurative styles. From this evaluation, we present design guidelines that can help people produce better outcomes from text-to-image generative models.
Overview
CoChat Summary
TL;DR
This CHI 2022 paper studies prompt engineering for text-to-image generation using VQGAN+CLIP, evaluating 5,493 generations across five experiments on prompt permutations, random seeds, optimization length, style breadth, and subject–style interaction. It derives design guidelines for users, including focusing on subject/style keywords, trying multiple seeds, using shorter optimization lengths, and warning about misinterpretation and offensive outputs.
Key findings
- Prompt rephrasings with the same subject and style keywords did not significantly differ: Experiment 1 tested 1,296 images across 9 permutations and found no significant difference (Chi-square statistic 0.354, p=0.55), despite 71.3% annotator agreement and a low Cohen’s kappa of 0.0013.
- Random seeds significantly affected generation quality: Experiment 2 tested 1,296 images across 9 seeds and found outlier generations significant at p<0.01, with inter-rater reliability 0.13; the authors recommend generating 3 to 9 seeds.
- Optimization length did not correlate simply with preference: Experiment 3 evaluated 1,000 iterations every 100 steps and found significant differences (p=0.01), with 200, 100, and 500 iterations most preferred; the authors suggest 300 iterations as a default and 100–500 as sufficient for fast iteration.
- Style and subject categories mattered: Experiment 4 found abstract styles averaged 2.63 (SE 0.06) versus figurative 3.16 (SE 0.06), p<0.01; Western 2.92 (SE 0.07) versus non-Western 2.95 (SE 0.06), p=0.377; digital 2.41, modern 2.83, premodern 3.11, p<0.001. Experiment 5 found concrete subjects generally rated higher, with top ten subjects averaging concreteness 4.47 and a Pearson correlation of 0.62 between concreteness and generation quality.
Methods
- Five experiments generated 5,493 images with VQGAN+CLIP pretrained on ImageNet with a 16,384 codebook; images were 256×256 and typically run for 300 steps on a local NVIDIA GeForce RTX 3080 GPU, except Experiment 3 varied iterations up to 1,000.
- Experiments manipulated prompt permutations (12 subjects × 12 styles × 9 permutations), random seeds (12 subjects × 12 styles × 9 seeds), iteration length (6 subjects × 12 styles, evaluated every 100 iterations), style breadth (51 styles), and subject–style interaction (51 subjects × 31 styles).
- Annotators with media arts/art practice backgrounds rated generations using grids or rubrics; analyses included Chi-square, Fisher’s exact, Mann-Whitney, Kruskal-Wallis, two-way ANOVA, Pearson correlation, and inter-rater reliability.
Limitations
- The paper states that most experiments used only one prompt, “SUBJECT in the STYLE of,” with VQGAN+CLIP; future work could test other prompts, models, image conditioning, intermediate iteration controls, and modifiers such as “4k” or “2048px.”
- The paper states that qualitative analysis of styles and subjects could be extended to understand cultural nuance and whether generations move beyond flat reproductions.
- Evident from the text: annotation was highly subjective, with low inter-rater reliability in Experiment 1 (Cohen’s kappa 0.0013) and slight agreement in Experiment 2 (0.13); only two annotators rated most tasks, one combination was removed for inappropriate content, and the subject-only pilot was abandoned because generations were too poor to evaluate.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 6 Not valid · 2 Not checked · 149 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“since many of the computer science datasets relevant to art focus on Western schools of art (i.e. WikiArt and MetFaces) [25, 47].”
What the sources say
- Partly supportedcites [25] [25] Tero Karras, Miika Aittala, Janne Hellsten, Samuli Laine, Jaakko Lehtinen, and Timo Aila. 2020. Training Generative Adversarial Networks with Limited Data. CoRR abs/2006.06676 (2020). arXiv:2006.06676 https://arxiv.org/abs/2006.06676 - sourceExperimental — accuracy not yet measured
Section 4.3 (p. 8) identifies MetFaces as the authors’ dataset of faces drawn from the Metropolitan Museum of Art collection, supporting its use as an example of an art-related computer science dataset. The paper does not establish that MetFaces focuses on Western schools of art, discuss WikiArt, or assess whether many such datasets have that focus. (LLM, conf=0.84)
- Not matchedcites [47] [47] wikiart.org. 2022. Visual Art Encyclopedia. https://www.wikiart.org/ -
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“While momentum has started to build in prompt engineering for text generation purposes, less work has been done to rigorously examine how users can prompt generative frameworks with natural language for visual generation purposes, which is the focus in this paper.”
What the sources say
- Partly supportedcites [17] [17] Songwei Ge and Devi Parikh. 2021. Visual Conceptual Blending with Large-scale Language and Vision Models. arXiv:2106.14127 [cs.CL] - sourceExperimental — accuracy not yet measured
The Introduction (p. 1) and Related Work (p. 2) describe using prompt engineering and textual descriptions to generate images, so the paper addresses natural-language-guided visual generation. It does not establish that less research has examined this than text-generation prompting, or examine how users formulate prompts. (LLM, conf=0.78)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
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
“For example, for the style unreal engine, most of the text prompts returned a scene with textured rocks and grass akin to what might be rendered by a game engine.”
hedgedstates an association- 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.
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 first recurring theme across successful styles was the presence of salient color schemes.”
plainstates an association- 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.
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
“We found that with a p-value of <0.01, the number of generations judged as outliers was significant when compared to the number of generations deemed not outliers.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
100 more claims
Create a free account to see all 160 claims and every source verdict.
- The quote only partly states this claim. 1
- It may state two separate claims as one. 30
- Probably not a substantial claim. 32
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Although recent work has begun to probe at design considerations for generative AI, this work has been limited to specific application domains or technologies. For example, guidelines of various maturity levels exist for GAN-based interfaces [57, 195], image creation [102, 104, 173], prompt engineering [102, 104], virtual reality [169], collaborative storytelling [146], and workflows with co-crea”
From Design Principles for Generative AI Applications (2024) · cites this paper as [104]
The cited source says what the claim attributes to it.
In the citing paper
“Teachers also emphasized the importance of prompt engineering education, which involves crafting words into sentences so that AI results are appropriate [17].”
From Design implications of generative AI systems for visual storytelling for young learners (2023) · cites this paper as [17]
In the citing paper
“This process is typically informal and relies on trial-and-error [104, 131, 165, 190].”
From Design Principles for Generative AI Applications (2024) · cites this paper as [104]
In the citing paper
“The recommendation to, 'Make clear what the system can do' may be difficult to implement in light of the emergent and unanticipated behaviors of generative foundation models [19], as well as the trial-and-error methods by which users iterate toward a desired outcome [104, 131, 165, 190].”
From Design Principles for Generative AI Applications (2024) · cites this paper as [104]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.1 Generation Methodology · p. 5
“We generated images using 10 randomly generated seeds 796, 324, 697, 11, 184, 982, 724, 962, and 805”
§ 4.1 Generation Methodology · p. 5
“To study the effect of seeds, we generated 1296 images from 12 subjects, 12 styles, and 9 seeds.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Experiment 4. Mean Ratings (Aggregate and Individual) for Partitions of Styles · p. 13
“Figure 9: For Experiment 6, in the left subgraphs, averages are reported for each category of the three partitions studied”
§ Experiment 4. Mean Ratings (Aggregate and Individual) for Partitions of Styles · p. 13
“## Experiment 4. Mean Ratings (Aggregate and Individual) for Partitions of Styles”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ B EXPERIMENT 3. RANDOM SEEDS · p. 22
“## B EXPERIMENT 3. RANDOM SEEDS”
§ 4 EXPERIMENT 2. RANDOM SEEDS · p. 5
“## 4 EXPERIMENT 2. RANDOM SEEDS”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“Table 1: In Experiment 4, we generated generations from 51 styles listed in the table above. Eight styles were general mediums of visual art. Twelve styles were Internet aesthetics. The remaining 33 were styles that were balanced for representation”
Reported 51 · recomputed 53
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
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
Analyzing and Improving the Image Quality of StyleGAN
Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., & Aila, T. (2020) · 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr42600.2020.00813 · Cited by 6,332 · cited 1×
“[6, 16, 21] Likewise, HCI researchers have sought to understand how generative AI can become a creativity support tool for artists.”
BackgroundAnalyzing and Improving the Image Quality of StyleGAN
Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., & Aila, T. (2019) · arXiv · doi:10.48550/arxiv.1912.04958 · Cited by 6,332
Generative Adversarial Networks
Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2014) · arXiv · doi:10.48550/arxiv.1406.2661 · Cited by 4,477
Liu, P., Yuan, W., Fu, J., Jiang, Z., Hayashi, H., & Neubig, G. (2021) · arXiv · doi:10.48550/arxiv.2107.13586 · Cited by 4,352 · cited 1×
“[28], who referred to this emerging paradigm as "pretrain, predict, and prompt".”
BackgroundLarge Scale GAN Training for High Fidelity Natural Image Synthesis
Brock, A., Donahue, J., & Simonyan, K. (2018) · arXiv · doi:10.48550/arxiv.1809.11096 · Cited by 3,717 · cited 1×
“[6, 16, 21] Likewise, HCI researchers have sought to understand how generative AI can become a creativity support tool for artists.”
BackgroundSemantics derived automatically from language corpora contain human-like biases
Caliskan, A., Bryson, J. J., & Narayanan, A. (2017) · Science · doi:10.1126/science.aal4230 · Cited by 2,887 · cited 1×
“[8] used sets of words to quantifably demonstrate bias within word embeddings.”
MethodA Neural Algorithm of Artistic Style
Gatys, L., Ecker, A., & Bethge, M. (2016) · Journal of Vision · doi:10.1167/16.12.326 · Cited by 2,612 · cited 1×
“Communities conversing about artistic AI have developed for a long time around techniques, algorithms, and networks such as DeepDream [26], neural style transfer [11], and generative adversarial networks.”
BackgroundZero-shot text-to-image generation
Ramesh, A., Pavlov, M., Goh, G., Gray, S., Voss, C., Radford, A., Chen, M., & Sutskever, I. (2021) · arXiv · doi:10.48550/arxiv.2102.12092 · Cited by 2,417 · cited 6×
“The authors of DALL-E demonstrated how the model could handle image operations, perform style transfer, and produce novel combinations of elements.”
BackgroundMethodInfluential
Research works printed but not linked: 1
- wikiart.org. 2022. Visual Art Encyclopedia. https://www.wikiart.org/ wikiart.org. 2022. Visual Art Encyclopedia. https://www.wikiart.org/
Cited by
Generative AI and ChatGPT: applications, challenges, and AI-human collaboration
Fui-Hoon Nah, F., Zheng, R., Cai, J., Siau, K., & Chen, L. (2023) · Journal of Information Technology Case and Application Research · doi:10.1080/15228053.2023.2233814 · Cited by 1,363
Why Johnny Can’t Prompt: How Non-AI Experts Try (and Fail) to Design LLM Prompts
Zamfirescu-Pereira, J. D., Wong, R. Y., Hartmann, B., & Yang, Q. (2023) · Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581388 · Cited by 1,274
Mou, C., Wang, X., Xie, L., Wu, Y., Zhang, J., Qi, Z., & Shan, Y. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i5.28226 · Cited by 923
Generative artificial intelligence
Banh, L., & Strobel, G. (2023) · Electronic Markets · doi:10.1007/s12525-023-00680-1 · Cited by 579
“The field of prompt engineering deals with systematically constructing prompts to improve the generated outputs (Liu & Chilton, 2022).”
BackgroundGenerative AI Literacy: Twelve Defining Competencies
Annapureddy, R., Fornaroli, A., & Gatica-Perez, D. (2025) · Digital Government: Research and Practice · doi:10.1145/3685680 · Cited by 333
“It is important to note that prompt engineering is not limited to LLMs, but to all types of generative AI that use text as input, including text-to-image models, as evidenced by the existing literature and guidelines for text-to-image prompt engineering [57, 108].”
Metadata
- Authors
- Liu, V., & Chilton, L. B.
- Affiliations
- Columbia University (US)
- Venue
- CHI Conference on Human Factors in Computing Systems
- Publisher
- ACM
- Published
- 2022-04-29
- Pages
- 1-23
- OpenAlex
- W3199400376
- Semantic Scholar
- 0968f1592f9401d72bf0d97e740496818c1a3135
- Cited by
- 639
- References
- 28
- Citations by year
- 2026: 112 · 2025: 212 · 2024: 220 · 2023: 91 · 2022: 4
- FWCI
- 47.78
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- CHI Conference on Human Factors in Computing Systems (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- arXiv (Cornell University) (cc-by)
- Funders
- NSF (National Science Foundation) (DGE - 1644869)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Generative Adversarial Networks and Image Synthesis
- Physical Sciences › Computer Science › Computer Graphics and Computer-Aided Design › Digital Media and Visual Art
- Social Sciences › Arts and Humanities › Visual Arts and Performing Arts › Art, Technology, and Culture
- Keywords
- text-to-image generation, prompt engineering, design guidelines
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-13, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-13 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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: Design Guidelines for Prompt Engineering Text-to-Image Generative Models. https://app.cochat.ai/papers/10.1145/3491102.3501825
- The paper
- Liu, V., & Chilton, L. B. (2022). Design Guidelines for Prompt Engineering Text-to-Image Generative Models. CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3491102.3501825
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.