Realistic Batik Motif Generation Using the Series of StyleGAN2 and StyleGAN3
JOIV : International Journal on Informatics Visualization · 31.07.2026 · 10 (4) · 1866
DOI: 10.62527/joiv.10.4.4809 ↗Kurz gefasst
Fachartikel in der Zeitschrift JOIV : International Journal on Informatics Visualization, online erschienen am 31.07.2026. Verfasst von Muhammad Taufik Dwi Putra und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
Batik, as a UNESCO-recognized intangible cultural heritage of Indonesia, poses a significant challenge in motif innovation. Despite high market interest, the development of new motifs tends to stagnate, risking market saturation and hindering cultural evolution. This research aims to address this challenge by leveraging state-of-the-art Generative Adversarial Networks (GANs), specifically the StyleGAN architecture, to generate new, high-quality batik motifs automatically. The objective of this study is to implement and evaluate the performance of several GAN and StyleGAN models, namely StyleGAN2, StyleGAN3-r, StyleGAN3-t, Diffusion-StyleGAN2, Diffusion-Wassertein-StyleGAN2, DCGAN, and CGAN. These models were trained on a custom-built dataset of approximately 5,000 high-resolution batik images, collected via web crawling and manually curated into various traditional categories. For robust quantitative validation of the distinctiveness of the generated motifs, a classification model based on the EfficientNetV2 and EfficientNetB7 architectures was also developed and trained on the same dataset. The research findings show that all models qualitatively generated diverse and visually appealing motifs. However, a comparative analysis revealed that the StyleGAN2 architecture demonstrated a superior capability in replicating the intricate details, authentic textures, and complex color transitions characteristic of hand-drawn batik. Quantitatively, the validation process confirmed the quality of the generated images, with the EfficientNetB7 classification model distinguishing among motif categories with high precision, achieving a final validation accuracy of 98% and an average F1 Score of 0.98. These findings affirm that the StyleGAN architecture, particularly StyleGAN2, is a powerful and effective tool for designers, accelerating innovation in the batik industry and bridging the gap between preserving traditional aesthetics and exploring modern creative possibilities.
Zitieren
APA
Putra, M. T. D., Munawir, Khaerunnisa, Z., Pradana, M. H., Ismail, R. B., & Zamzami, M. F. A. (2026). Realistic Batik Motif Generation Using the Series of StyleGAN2 and StyleGAN3. JOIV : International Journal on Informatics Visualization, 10(4), 1866. https://doi.org/10.62527/joiv.10.4.4809
Vancouver
Putra MTD, Munawir, Khaerunnisa Z, Pradana MH, Ismail RB, Zamzami MFA. Realistic Batik Motif Generation Using the Series of StyleGAN2 and StyleGAN3. JOIV : International Journal on Informatics Visualization. 2026;10(4):1866. doi:10.62527/joiv.10.4.4809
BibTeX
@article{Putra2026,
author = {Putra, Muhammad Taufik Dwi and Munawir, - and Khaerunnisa, Zahra and Pradana, Muhammad Hilmil and Ismail, Rosli Bin and Zamzami, M. Fauzan Al},
title = {Realistic Batik Motif Generation Using the Series of StyleGAN2 and StyleGAN3},
journal = {JOIV : International Journal on Informatics Visualization},
year = {2026},
volume = {10},
number = {4},
pages = {1866},
doi = {10.62527/joiv.10.4.4809}
}
EUVIMED