This beach does not exist

StyleGAN2-ADA network trained in HD resolution 1280×768 on dataset with ~20.000 images of the beach. All of these beaches are AI-generated.

Website monitoring by Testomato.com

Training progress

The video shows the learning process of the network. The learning progress is measured in kimg (kilo images). The network was trained for 25,000 kimg (until it had seen 25,000,000 images).

Random images

Generate random images

K-Means Clustering

Using the K-Means Clustering technique to group similar images. K-Means clustering is the unsupervised machine learning algorithm for image segregation.

Cluster #0/376

Download

Download the network for your experiments
or retrain the network with your own dataset (transfer learning).

  Download (362 MB)

Technical details

Architecture StyleGAN2-Ada
Dataset 20.000 images
Network size 362 MB
Resolution 1280×768 px
Network layout 5×3 squares (with 28 = 256 px sides)
Training options --min_w=5 --min_h=3 --res_log2=8
Trained using RoyWheels/stylegan2-ada (github)
Training steps 25.000 kimg
Metric FID50k 3.14

Credits

This beach image generator was made using StyleGAN2-Ada, the GAN architecture published by researchers from NVIDIA Labs.

Paper: Training Generative Adversarial Networks with Limited Data
Original implementation: NVlabs/stylegan2-ada
Non-square support and other improvements: RoyWheels/stylegan2-ada