stylegan3 anime faces generator
akiyamashoIntroduction
The Anime Faces Generator is a StyleGAN3 model implemented in PyTorch, designed to generate anime-style faces. This model is based on NVIDIA's StyleGAN3 and trained with the Anime Face Dataset.
Architecture
The architecture employs the StyleGAN3 model from NVIDIA, which is a generative adversarial network (GAN) optimized for high-quality image generation.
Training
The model was trained using the Anime Face Dataset on a Paperspace Gradient platform with an RTX-5000 GPU. The configuration used was stylegan3-t
with a batch size of 8 and a gamma value of 0.5. The training completed at tick 102, achieving a fid50k_full value of 9.26043547642206.
Guide: Running Locally
- Clone the Repository: Clone the generator-script-only subset of the StyleGAN3 repo from here.
- Install Dependencies: Ensure you have the necessary Python environment and dependencies installed.
- Run Locally: Follow the instructions in the repository to run the model using the provided pickle file.
For leveraging cloud GPUs, consider using platforms like Paperspace or Google Colab for GPU acceleration.
License
The project is licensed under the MIT License, allowing for flexibility in modification and distribution.