We’ve seen some interesting uses of AI so far, and the latest prototype software launched by Nvidia could be of great help to many artists. It can turn your crude sketches into pretty realistic landscapes in a matter of a click.
Nvidia researchers explain in a blog post that their latest software uses generative adversarial networks, or GANs. It converts rough doodles into realistic images thanks to a “smart paintbrush.” You can use this paintbrush to draw a very basic outline of a scene and its elements (sea, rocks, sand, sky, clouds etc). Then, the tool lets you fill in the sketch with realistic textures for the elements you drew.
The researchers explain that the software is very intuitive. For example, if you draw a body of water in a scene, the software will add a reflection of the scene in it. In the video above, you can see a demonstration of the tool. It doesn’t create masterpieces, but the results are still very impressive. The software seems pretty simple to use and does the job very quickly.
It’s still not clear when and if the new software will find its application. But, the researchers note that it could help artists such as architects, urban planners, landscape designers, and game developers. It should help them do their jobs faster and make rapid changes to the scenes they’re designing.
I believe that a tool like this could also be helpful for photographers with poor drawing skills (such as myself). If you have a sketchbook for ideas, software like this could help you draw them and visualize them much better. The drawback is that this software is limited to landscapes, but hey – who knows what the future holds.
[via The Verge]