Nvidia Brings AI Technology To Draw With Just Phrases

Nvidia has demonstrated its ability to turn simple sketches into photorealistic images. (Photo: Nvidia)

JAKARTA - Nvidia is showing off again with its technology, GauGAN. Armed with AI, Nvidia has demonstrated its ability to turn simple sketches into photorealistic images.

It was first introduced in 2019, but has recently been featured in products available to the general public.

Named Canvas from the product GauGAN version 2.0 (GauGAN2), this software can be a lot of fun to use, allowing users to create stunning photo-like images with basic sketches. This app is free in beta for all RTX GPU users.

The first version of GauGAN was trained on public images from the Flickr platform of more than one million images. However, GauGAN2 was trained 10 times and could understand natural language descriptions relating to landscapes.

GauGAN is a Microsoft Paint-style platform that allows users to create landscape images, with models that can then turn them into photorealistic images. However, GauGAN2 can generate images using only text.

Citing TechSpot, Friday, November 26, users can type in phrases such as winter, foggy or rainbows and the AI model can generate images that match the desired descriptor.

“With the push of a button, users can create segmentation maps, high-level outlines that show the location of objects at the scene. From there, they can move on to drawing, altering scenes with rough sketches using drawings by scribbling on the sky, trees, rocks, and rivers, enabling smart brushes to combine these doodles into stunning images."

Unfortunately, Nvidia did not mention any commercialization plans, or whether it would be integrated with Canvas and made available to the general public. But the demo does illustrate future possibilities for powerful image-making tools for artists.