AI-Powered Sprite Generation for Visual Novel Games: A Latent Diffusion Model Approach
DOI:
https://doi.org/10.11594/Keywords:
Game development, Generative AI, ISO 25010, Latent diffusion models, Sprite generation, Visual novelsAbstract
Creating character sprites for visual novel games has long been a resource-intensive process, erecting significant barriers for independent developers and small studios. Accessible, high-quality asset generation remains an unmet need—one this study addresses by developing a specialized AI-powered sprite generator built on a Latent Diffusion Model (LDM) fine-tuned with Low-Rank Adaptation (LoRA). Unlike general-purpose image generators, the system is optimized specifically for 2D anime-style character sprites with multiple emotional variants, making it suitable for narrative-driven gameplay. Development followed a seven-phase Agile methodology, and evaluation involved 50 student game developers working against ISO/IEC 25010 software quality standards. Results confirm that the system reliably generates up to nine cohesive emotional variants from a single text prompt, substantially streamlining the asset-creation workflow. User scores indicate strong satisfaction with Functional Suitability (M=3.73), Security (M=3.74), and Usability (M=3.66). Performance Efficiency received a lower rating (M=3.20, Agree), attributable to a generation latency of approximately 2–3 minutes per sprite set—acceptable for final production, but less so for rapid prototyping. Taken together, these findings support the viability of specialized LDM applications as practical tools for democratizing visual game development and empowering collaborative storytelling.
Downloads
References
Duan, H., & Zhang, J. (2022). AI-driven envi-ronmental concept design: A framework for adaptive visual stimuli in interactive storytelling. Journal of Digital Media Arts, 14(2), 112-128.
Ergen, T. (2023). Procedural generation of video game characters using StyleGAN2: techniques and applications. Internation-al Journal of Computer Games Technolo-gy, 2023, Article 451239.
Gonzalez, M. (2020). The artificial art director: AI as a creative partner in game devel-opment. Game Studies, 20(1). http://gamestudies.org/2001/articles/gonzalez
Ma, X., Li, Y., & Chen, S. (2022). AI Illustrator: A cross-modal framework for narrative visualization. IEEE Transactions on Mul-timedia, 24, 2890-2902.
Park, J., Kim, S., & Yoo, D. (2021). Enhancing digital literacy in education through AI-assisted storytelling tools. Computers & Education, 168, 104201.
Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., & Chen, M. (2022). Hierarchical text-conditional image generation with CLIP latents. arXiv preprint arXiv:2204.06125.
Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022). High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat-tern Recognition (pp. 10684-10695).
Shi, Y., Wang, X., & Liu, Z. (2020). PokerFace-GAN: Generating neutral face templates for game character customization. ACM Transactions on Graphics, 39(4), 89:1-89:12.
Tanaka, H., & Kobayashi, S. (2021). Automat-ing game asset creation: A survey of AI-based approaches. Entertainment Com-puting, 38, 100412.
Wu, L., Zhang, Y., & Wang, K. (2023). The im-pact of generative AI on video game visu-alization workflows. Journal of Graphics Tools, 27(1), 15-29.
Xie, Y. (2024). Consistency challenges in AI-generated character art for serial narra-tives. Digital Creativity, 35(1), 45-60.
Zhao, L., & Zhang, W. (2024). From GANs to Diffusion: A comparative analysis of text-to-image models in digital art. AI & Socie-ty, 39, 215-230.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Eliza B. Ayo, Pio Honesto Rico Belleza, Aaron Dwayne F. Esmaquel, Edmund N. Lazaro
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See the Effect of Open Access).
