3/22/2019 0 Comments
A few days ago, Nvidia introduced an impressive artificial intelligence that created incredible images from almost ordinary shorts. The user will definitely need a few seconds for this picture, but what makes the artificial intelligence of Nvidia with the image is impressive.
It's an interesting drawing tool in which the user chooses what he draws and gives him a form. On behalf of the networks (generative adversarial networks), the application is called GauGAN. Brushes are a kind of material selection, and the user only gives some form, and then the artificial intelligence decides what and how it enters into this form, certainly considering what has been selected. It's amazing that artificial intelligence has learned that if someone draws "water," it should add reflection.
What must be pointed out is that the tool does not use parts of existing images that fill the space, but creates a unique part.
The tool is on the Tensor platform, the hardware that is presented is the RDX Titan. According to the speed of drawing, the events appear to occur in real time, from pulling the line to the creation of superficial. Nvidia used millions of Flickr images to train artificial intelligence. For training, images published under the Creative Commons license are used.
Probably most artists and designers will not use this tool, but it might be useful when a picture is needed quickly.
So far, the software is a test prototype and no commercial version. Nvidia hopes GauGAN will be part of the AI Playground. There is a test version announced soon enough for anyone who wants to test the software.