A Generative Model-based Emulator for mocking Galaxies with spiral arms

In the study of spiral galaxy morphology, spiral arm structures are valuable for intuitively reflecting active physical and chemical processes within galaxies. However, long-term scarcity of high-quality one-, three-, and four-armed galaxy samples has limited deep learning model performance.To address this, this study developed a spiral galaxy data simulation program with a three-stage workflow: first, screening highly reliable training samples; second, selecting the best-performing Imagen3 architecture as the generative model after comparing 9 mainstream ones; finally, training Imagen3 to generate an open dataset of 9402 one-/three-armed galaxies, expanding the original sample size by 6 times.Multi-dimensional evaluations verified reliability and usability: FID scores for N=1 and N=3 tasks were 6.05 and 9.13; t-SNE showed generated data covered and expanded real data distribution; SSIM confirmed no sample duplication. In downstream validation, data augmentation improved 7 classification models’ average accuracy by 8.7% (DenseNet peaked at 97%), and SHAP analysis showed model decisions focused on spiral arm topology.In conclusion, the program and dataset support spiral galaxy morphology deep learning research and are publicly available at https://github.com/TuAstroAILab/AstroGS.