Nvidia’s latest AI technology can convert text into realistic images

Nvidia's GauGAN technology has demonstrated its ability to turn simple sketches into realistic images. Since then, we have seen it applied to NVIDIA Canvas, but it seems that this GPU giant is targeting higher goals with its artificial intelligence (AI), launching a new version that can convert text into images.

Nvidia first showcased its GauGAN technology in 2019, but it was not until recently that we saw it applied to products aimed at the general public. This software called Canvas is very interesting to use, allowing users to create stunning images similar to photos using basic sketches.

Several months have passed since Canvas announced, but GauGAN's work has continued and has now reached version 2.0. This technology has become even more impressive as it can now turn text into realistic images, providing results similar to those obtained using drawing functionality.

As seen in the video above, writing something on a text box will immediately generate an image based on your text. Adding an adjective or replacing a noun image in a phrase will change accordingly.

To increase personalization, users can combine the functions of text and drawing into images. By using written text to generate foundations and painting to refine images, users can change the shape, size, and texture of any object in the image.

To achieve these results, Nvidia's GauGAN 2 text image feature utilizes an AI model based on generative adversarial networks, which combines segmentation mapping, inlining, and text image generation. This model was trained with 10 million landscape images, so it should be well prepared for anything the user provides.

Prev: How Robot Technology Helps Environmentally Friendly Buildings