ANIME CHARACTER GENERATION USING STYLEGAN AND LATENT SPACE MANIPULATION

  This paper presented an anime character generation framework using StyleGAN and latent space manipulation. The model successfully generates high-quality stylized anime characters and enables semantic editing through vector operations in latent space.   Future work includes improving feature disentanglement using supervised attribute learning, integrating conditional inputs for precise control, exploring StyleGAN3 for enhanced alias-free generation, and developing user-friendly interfaces for interactive character customization. Additional research may also investigate multimodal generation combining text prompts with style-based generative models.

[1] I. Goodfellow et al., “Generative Adversarial Nets,” NIPS, 2014.   [2] A. Radford et al., “Unsupervised Representation Learning with Deep Convolutional GANs,” ICLR, 2016.   [3] T. Karras et al., “Progressive Growing of GANs for Improved Quality,” ICLR, 2018.   [4] T. Karras et al., “A Style-Based Generator Architecture for Generative Adversarial Networks,” CVPR, 2019.   [5] T. Karras et al., “Analyzing and Improving the Image Quality of StyleGAN,” CVPR, 2020.   [6] T. Karras et al., “Alias-Free Generative Adversarial Networks,” NeurIPS, 2021.   [7] J. Shen et al., “Interpreting the Latent Space of GANs for Semantic Face Editing,” CVPR, 2020.   [8] E. Härkönen et al., “GANSpace: Discovering Interpretable GAN Controls,” NeurIPS, 2020.   [9] Y. Choi et al., “StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation,” CVPR, 2018.   [10] A. Brock et al., “Large Scale GAN Training for High Fidelity Natural Image Synthesis,” ICLR, 2019.