CoChat

Sketchar: Supporting Character Design and Illustration Prototyping Using Generative AI

Open access
Ling, Long, Chen, Xinyi, Wen, Ruoyu, Li, Toby Jia-Jun, LC, RAY · 2024 · Proceedings of the ACM on Human-Computer Interaction

The DOI matches its Crossref record; 67 of the 68 references it cites were found in a registry or an index; 15 of the 48 sources its claims cite were checked for support.

Abstract

Character design in games involves interdisciplinary collaborations, typically between designers who create the narrative content, and illustrators who realize the design vision. However, traditional workflows face challenges in communication due to the differing backgrounds of illustrators and designers, the latter with limited artistic abilities. To overcome these challenges, we created Sketchar, a Generative AI (GenAI) tool that allows designers to prototype game characters and generate images based on conceptual input, providing visual outcomes that can give immediate feedback and enhance communication with illustrators' next step in the design cycle. We conducted a mixed-method study to evaluate the interaction between game designers and Sketchar. We showed that the reference images generated in co-creating with Sketchar fostered refinement of design details and can be incorporated into real-world workflows. Moreover, designers without artistic backgrounds found the Sketchar workflow to be more expressive and worthwhile. This research demonstrates the potential of GenAI in enhancing interdisciplinary collaboration in the game industry, enabling designers to interact beyond their own limited expertise.

Overview

CoChat Summary

TL;DR

Sketchar is a web-based GenAI tool that uses ChatGPT and DALL-E to help game designers prototype character profiles and reference images for communication with illustrators. In mixed-method studies, it improved perceived collaboration and was especially valued by designers without artistic backgrounds, though image quality and simulated collaboration limited conclusions.

Key findings

  • Sketchar’s CSI score was significantly higher than the sketching-only baseline (p = .028); no significant interaction was found between workflow and art skill (p = .146, Scheirer-Ray-Hare test).
  • Participants without professional artistic skills scored significantly higher on CSI than those with artistic skills (p = .011), with differences above 2 points in Expressiveness and Results Worth Effort.
  • Expert illustrators identified an average of 6.84 out of 10 generated images per set as relevant, valuable references for game character artists.
  • Qualitative findings: artistic backgrounds led to higher demand for detail; GenAI classification/expansion helped designers express ideas and refine details; generated reference images assisted visual expression of design requirements.

Methods

  • Formative interviews with 10 game design professionals, analyzed via thematic analysis and keyword coding, informed Sketchar’s design. The tool uses hierarchical prompt engineering with ChatGPT, DALL-E, and CLIP in a React/Flask web application.
  • Qualitative study: 13 game designers (5 female, 8 male) completed a baseline character design proposal and a Sketchar-supported task, followed by semi-structured interviews. Quantitative study: 17 participants aged 21–27 (Mean = 23.47, SE = 1.87) completed counterbalanced baseline and Sketchar tasks with CSI surveys after each.
  • Expert evaluation: 5 illustrators assessed 10 generated images for each of 5 prompts. Analyses included Wilcoxon signed-rank, Wilcoxon rank-sum, Scheirer-Ray-Hare two-way nonparametric ANOVA, and Cronbach’s alpha (>0.7 for all CSI factors).

Limitations

  • Stated: Sketchar sacrifices image intricacy for faster generation; illustrator P9 needed 5 iterations to achieve distinct facial tattoos. The paper suggests switching to more controllable models like Stable Diffusion or DALL-E 3 could improve quality but may increase generation time.
  • Stated: testing was limited to mainland China, where ChatGPT and DALL-E were inaccessible and most participants lacked GenAI experience; collaboration was simulated with single individuals rather than real designer–illustrator teams; no direct comparison was made to ChatGPT plus DALL-E; and the CSI Collaboration subscale normally measures synchronous collaboration, so its use for asynchronous collaboration may be speculative.
  • Evident: samples were small (13 qualitative, 17 quantitative, 5 expert evaluators), and the paper notes it did not statistically test each CSI factor score. The paper also reports DALLE-2 stereotype effects, such as a Maori brave man prompt producing a white male with indigenous-style tattoos and Chinese businessperson prompts always producing a young Chinese male in a suit.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1145/3677102

Cites

67 of 68 references verified0 citation errors · 1 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

8 Valid · 10 Not valid · 20 Not checked · 96 Own results and interpretations

49 more claims

Create a free account to see all 134 claims and to verify the 5 references still open.

See all claims — free account
  • It may state two separate claims as one. 44
  • Probably not a substantial claim. 56
  • The quote only partly states this claim. 1

Create a free account to see these claims.

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.

3 of 3

Citations

Cites

Show all 68 references

Research works printed but not linked: 11

  1. AlgorithmWatch. 2024. Image Generators Reinforce Stereotypes, Lack Diversity: A Study by AlgorithmWatch. AlgorithmWatch (2024). https://algorithmwatch.org/en/image-generators-stereotypes-diversity/ Accessed: 2024-05- 26.
  2. Janna Anderson and Lee Rainie. 2023. As AI spreads, experts predict the best and worst changes in digital life by 2035. (2023).
  3. Marianne Krawczyk and Jeannie Novak. 2006. Game development essentials: Game story & character development. (No Title) (2006).
  4. Petri Lankoski. 2004. Character design fundamentals for role-playing games. Beyond Role and Play (2004), 139--148.
  5. Jonathan Lazar, Jinjuan Heidi Feng, and Harry Hochheiser. 2017. Research methods in human-computer interaction. Morgan Kaufmann.
  6. Sola Lee, Suin Jun, Mihye Park, Sui Park, and Jongin Choi. 2021. Character Generation and Storytelling using Modeling Tool and Game Engine. In Proceedings of the Korean Institute of Information and Commucation Sciences Conference. The Korea Institute of Information and Commucation Engineering, 620--622.
  7. Eliott Lilly. 2017. The Big BadWorld of Concept Art for Video Games: How to Start Your Career as a Concept Artist. Design Studio Press.
  8. Margaret Mark and Carol S Pearson. 2001. O herói e o fora-da-lei: como construir marcas extraordinárias usando o poder dos arquétipos. São Paulo: Pensamento (2001).
  9. John Naughton. 2023. The ChatGPT bot is causing panic now--but it'll soon be as mundane a tool as Excel. The Guardian 7 (2023).
  10. P Tavares and PM Teixeira. 2011. Drawing and project: tradition and innovation. In Proceedings of Conferência Designa.
  11. Jesse Voimala. 2023. Creating a Video Game Character Design Workflow: a literature review on the theories and methods used by professionals in video game character design. (2023).

Cited by

Metadata

Authors
Ling, L., Chen, X., Wen, R., Li, T. J.-J., & LC, R.
Affiliations
  • Tongji University (CN)
  • Zhejiang University (CN)
  • University of Canterbury (NZ)
  • University of Notre Dame (US)
  • City University of Hong Kong (HK)
Venue
Proceedings of the ACM on Human-Computer Interaction
Published
2024-10-14
Volume
8
Issue
CHI PLAY
Pages
1-28
ISSN
2573-0142 (ISSN-L)
OpenAlex
W4403424095
Cited by
52
References
44
Citations by year
2026: 18 · 2025: 33 · 2024: 1
FWCI
45.11
Type
journal article
Language
en
Open access
Locations
  • Proceedings of the ACM on Human-Computer Interaction (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • CityU Scholars (submittedVersion, other-oa)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Computer Graphics and Computer-Aided Design › Digital Media and Visual Art
  • Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
  • Social Sciences › Arts and Humanities › Visual Arts and Performing Arts › Art, Technology, and Culture
Keywords
character design, game design, generative AI, interdisciplinary collaboration, co-creation, reference images, design communication

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 2026-10-12, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-12 (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

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.