Create digital worlds from sketches
Creating a virtual digital environment is something that typically takes time, as those involved in games on computers know. In order to create the latest games and meet cutting-edge standards, many 3D artists work hard on creating graphics that are at the same level of quality as a human actor. However, this may change, due to a powerful new AI algorithm that can produce photorealistic details of a scene on the spot.
This software will make life a bit easier for software developers with automated virtual environments. Alongside that, it can be used to automatically create virtual reality or autonomous vehicles!
“We can create new sketches that have never been seen and render those,” said Brian Catanzaro, vice president of applied deep learning at Nvidia. “We basically teach the model how to draw based on real video,” he said.
The researchers used a standard machine learning approach to identify different objects in an image scene: Cars, trees, buildings, etc. Then they used something known as a GAN to learn from real images. They used it to “train” a computer to fill in realistic 3D images and it worked really well!. From there, the system can take a scene’s outline and locate different objects, and it will then fill in the details.
If you enjoyed this article, then you may also like this one.
Author: PC-GR
The World of Technology
