GauGAN2 is a tool for creating AI-generated photorealistic landscapes. Developed by NVIDIA and available as a web demo, GauGAN2 allows you to modify the image as much as you like, using a combination of sketching and text input to guide the AI ​​system.
With a little time, you can refine the image by adding or subtracting details until it looks just right. The images it can produce often look realistic, but one wrong stroke and the images can start to look like fantasy art, rather than real-life landscapes.
Here’s what you need to know about NVIDIA’s GauGAN2.
What is NVIDIA GauGAN2?
tonightGAN2 is an AI art generator with a unique editing window. Developed by the research arm of NVIDIA, the same company that makes GPUs, it’s available for anyone to use via a web demo.
It’s similar to using Dall-E 2, an AI text-to-image generator, where you can type phrases like “sunset over the desert” and the AI ​​system will create an image based on that description. However, that’s not all you can do with GauGAN2; There are several ways to continue editing the image.
One option is to use the segmentation map to see an outline of the objects in the image. Using this information, you can change the shape of things like a mountain range by redrawing its boundaries or erasing elements entirely.
You can also use the pencil tool to draw a circle, for example, which could be represented as a sun. The same goes for drawing some peaks in the background, which could turn into a mountain range, especially if you combine it with text input describing the scene.
After trying NVIDIA’s GauGAN2, you may want to learn how to use NVIDIA’s Canvas application, which is based on the same technology. It is a free app aimed at artists and creatives, however it only works with one of NVIDIA’s RTX GPUs.
How does GauGAN2 work?
Behind the scenes, GauGAN2 was built using Generative Adversarial Networks (GANs). In fact, its name is a play on the French Post-Impressionist artist Paul Gauguin and the GANs.
The reason it can generate images that convincingly resemble real landscapes is that during its training period it received 10 million high-quality landscape images. The computing power required to train the model came from the NVIDIA Selene supercomputer, one of the most powerful supercomputers in the world.
GANs are finding uses in many applications, including many AI text-to-art generators. The key difference to GauGAN2 is the ability to modify the AI-generated image in a dedicated editor, using tools like pencil, fill, and brush to make changes.
The interface is minimal, but not exactly easy to use, and its retro look may remind you of the early days of Microsoft’s paint, unlike last-gen software. If the controls don’t seem obvious at first, you can run a quick tutorial or watch a video tutorial; just click the buttons in the top right corner of the page.
What art can GauGAN2 produce?
The goal of GauGAN2 is to create photorealistic images of landscapes such as mountains, lakes, and beaches.
There is a library of tags you can choose from when drawing that help further tag the sketch input, so the AI ​​generator can render an appropriate image. These include things like fog, hill, stone, snow, and water, to name a few.
However, it is not difficult for things to go wrong, and you are likely to produce absurd and surreal images. Just grab the pencil tool and draw some weird shapes, and GauGAN will spit out something that looks more like experimental artwork than anything recognizable.
Entering nonsense phrases into the text box can also lead to strange results. Sometimes you don’t even need to go that far; simply asking for a mountain landscape produced what looks like an odd combination of snow and fur.
In any case, some artists may find a tool like this useful. It can be used to speed up the storyboarding process or as a quick way to create concept art for a game or video.
While this demo is far from a polished out-of-the-box product, it does show some interesting potential.
GauGAN2: An AI Hybrid Art Generator
GauGAN2 joins the collection of AI image generators that convert text to images. Unlike other apps, however, you can edit the image in stages, using sketches and text as input, adjusting the AI ​​image as you like.
If used correctly, GauGAN2 can produce quite convincing photorealistic images of nature, which could be useful for artists who need to complete a storyboard or produce concept art quickly. However, it’s all too easy for things to get weird, and a few pencil strokes can lead to some interesting results.