NVIDIA creates technology that can transform procedurally generated worlds

NVIDIA has shared a video showing the progress in developing artificial intelligence capable of creating images. The company has gone even further and allows you to create realistic graphics based only on textual descriptions.

The fruit of recent months of work by NVIDIA specialists is GauGAN2 (I really like this name). Check out what machine learning software is capable of in the video below. Interestingly, you can even help him by painting with a special brush.

I can imagine that with the appropriate development of this technology, AI could also create 3D models that depict different places in the same way. If the results were satisfactory, I see no obstacle to developing a game development engine that would allow you to create scenery in this way. The topic of procedurally generated levels, shaders and textures has been known to game developers for many years.