This Fursona Does Not Exist

Welcome to This Fursona Does Not Exist! This unique website showcases AI-generated furry portraits created by arfa, using nVidia's StyleGAN2 architecture. The AI model was trained on a dataset of approximately 55,000 safe-for-work images from e621.net (excluding ponies and scalies), with each image carefully cropped and aligned using a custom YOLOv3 network. After 33 days of training on a TPUv3-32 pod, the result is highly detailed furry portraits. These final images are scaled up to twice their original resolution and denoised using Waifu2x, ensuring high-quality visuals. The best part? These AI-generated images are non-copyrightable, allowing for free and responsible use. Additionally, the model used to generate these portraits is available for download, enabling you to explore and experiment with this fascinating technology.

Key Features

• AI-Generated Furry Portraits: High-quality furry images created using advanced AI technology.

• Trained with StyleGAN2: Utilizes nVidia's cutting-edge StyleGAN2 architecture.

• Extensive Image Dataset: Built on a dataset of around 55,000 SFW images from e621.net.

• Enhanced Image Quality: Images are scaled up and denoised for superior resolution and clarity.

• Open-Source Model: The AI model is available for download, promoting transparency and further development.

FAQ

What technology was used to create the furry portraits?

The furry portraits were created using nVidia's StyleGAN2 architecture.

How long did the AI training process take?

The training process took 33 days, totaling 3.2 million iterations.

What is the source of the image dataset?

The dataset is comprised of approximately 55,000 SFW images from e621.net.

Can I use the images generated by this AI?

Yes, the images are non-copyrightable, and you can use them as long as it is done responsibly.

Is the AI model available for download?

Yes, the model is available for download to encourage further exploration and development.