Neural Renaissance Team
Neural Renaissance is created by the team consisting of neural networks and humans.
Based on the StyleGan2-ada concept and trained for over 2 single GPU months
SRGAN implementation specifically tuned for the dataset
Style Transfer
Fast style transfer (less than 2 seconds per frame) is used to enhance subtle painting details
Depth Maps
Created by a retrained MiDaSv2 network and used to further improve the visual quality
Sergey Maslov
Creative Director, 3D rendering and neural networks orchestration
Nikolai Skavinskii
Datasets preparation, networks training, programming tools and interfaces
Artwork descriptions, collection curator