Batik, a unique blend of art and craftsmanship, is a distinct artistic and tech nological creation for Indonesian society. Research on batik motifs is primarily focused on classification. However, further studies may extend to the synthe- sis of batik patterns. Generative Adversarial Networks (GANs) have been an important deep learning model for generating synthetic data, but often face chal lenges in the stability and consistency of results. This research focuses on the use of StyleGAN2-Ada and Diffusion techniques to produce realistic and high- quality synthetic batik patterns. StyleGAN2-Ada is a variation of the GAN model that separates the style and content aspects in an image, whereas diffu- sion techniques introduce random noise into the data. In the context of batik, StyleGAN2-Ada and Diffusion are used to produce realistic synthetic batik pat- terns. This study also made adjustments to the model architecture and used a well-curated batik dataset. The main goal is to assist batik designers or crafts men in producing unique and quality batik motifs with efficient production time and costs. Based on qualitative and quantitative evaluations, the results show that the model tested is capable of producing authentic and quality batik pat terns, with finer details and rich artistic variations. The use of the Wasserstein loss function tends to produce batik motifs that are relatively new but less neat than the use of the StyleGAN2-Ada loss. The quality of the dataset also has a positive impact on the quality of the resulting batik patterns. Overall, this research contributes to the integration of Diffusion-GAN technology with tra ditional arts and culture, especially in the synthesis of batik motifs. However, there is still room for further development in increasing skill and accuracy in pro ducing more detailed batik motifs.
- 20,000 batik images from 20 types of batik at: Batik datasets
- Chrystian, C.: Itb-mbatik dataset (2023)
To prepare the datasets use commands such as example, or please refer python dataset_tool.py --help for more details on dataset preparation.
python dataset_tool.py --source=/source_folder/ --dest=/destination_folder/ --width=256 --height=256Please refer diffusion-stylegan2-ada-pytorch/config/ for configuration details of 8 experiments conducted in this study.
| Model | Dataset | Resolution | FID | Checkpoint |
|---|---|---|---|---|
| StyleGAN2 | 20.000 Batik Images | 256x256 | 46.799 | download |
| StyleGAN2 | Combined Datasets with ITBmBatik | 256x256 | 57.4302 | download |
| Wassertein-StyleGAN2 | 20.000 Batik Images | 256x256 | 48.781 | download |
| Wassertein-StyleGAN2 | Combined Datasets with ITBmBatik | 256x256 | 42.9192 | download |
| Diffusion-StyleGAN2 | 20.000 Batik Images | 256x256 | 29.045 | download |
| Diffusion-StyleGAN2 | Combined Datasets with ITBmBatik | 256x256 | 33.0104 | download |
| Diffusion-Wassertein-StyleGAN2 | 20.000 Batik Images | 256x256 | 36.756 | download |
| Diffusion-Wassertein-StyleGAN2 | Combined Datasets with ITBmBatik | 256x256 | 43.4331 | download |
If you use the batik dataset, or find this helpful, please cite this paper:
@misc{octadion2023diffbatik,
title={Synthesis of Batik Motifs using a Diffusion -- Generative Adversarial Network},
author={One Octadion and Novanto Yudistira and Diva Kurnianingtyas},
year={2023},
eprint={2307.12122},
archivePrefix={arXiv},
primaryClass={cs.CV}
} Please visit the original Diffusion-GAN and StyleGAN2-Ada-Pytorch on https://github.com/Zhendong-Wang/Diffusion-GAN and https://github.com/NVlabs/stylegan2-ada-pytorch