Posters
:
42. Realistic Postprocessing of Rendered 3D Scenes
SessionDisplay & Rendering
TimeNone
LocationWest Building, Ballroom Foyer, Vancouver Convention Centre
DescriptionPoster Sessions: Monday, 13 August and Wednesday, 15 August, 12:15-1:15 pm
We show realistic post-processing rendering based on generative adversarial network CycleWGAN. We propose using unpaired image-to-image translation with CycleGAN architecture and Wasserstein loss function with additional identity component in order to transfer modern graphics from GTA V video-game
to the older version GTA, San Andreas, using a deep learning approach.
Contributors
Anastasia Feygina
National Research University Higher School of Economics
Dmitry I. Ignatov
National Research University Higher School of Economics
Ilya Makarov
National Research University Higher School of Economics