NVIDIA has developed a generative art system that employs artificial intelligence to transform words into aesthetically stunning art pieces. This isn’t the first time a concept like this has been proposed, let alone realized. However, this is the first time we’ve seen such a system operate at such a high rate and with such precision.
OpenAI has a project named DALLE that you can look at. That’s a GPT-3-based image-generation project that you can learn more about at Cornell University. With Deep Dream Generator, you may start making wild interpretations of styles or learn about some of the sources for the NVIDIA Research project we’re looking at today — see the paper Generative Adversarial Networks to learn more about GAN!
NVIDIA Project GauGAN2 is a follow-up to NVIDIA Canvas, which the company’s researchers developed. That program, which is now in Beta, works with the initial GauGAN model. With artificial intelligence at their disposal, anyone can create a somewhat realistic-looking work of art using only the information needed to do a finger painting.
NVIDIA researchers enhanced the capabilities of simple input and artificial intelligence interpretation using GauGAN2. This program draws on an extensive library of visual knowledge (about 10 million high-quality landscape photos) to determine what your words may represent in a work of art.
GauGAN2 incorporates various modalities in a single GAN architecture. Text, semantic segmentation, drawing, and style are all mentioned by NVIDIA. This new text input element is demonstrated in an interface simply an extension of NVIDIA Canvas in the video below.
The demonstration isn’t nearly as important as what it stands for. A smartphone can now magically remove parts of a photograph. Artificial intelligence is already in your life if you use a system like Google Photos, and it’s getting brighter as you feed it more photographs captured by your phone.
The next wave has arrived, with NVIDIA’s demonstration demonstrating how the machine can do more than recognize components in photos; it can also develop imagery based on the images it has been fed. NVIDIA has created a model that convincingly demonstrates how graphics processing power combined with the correct combination of codes can provide astonishingly accurate representations of what we humans understand as reality.
Source: nvidia
, ...){kind=link}