The new GauGAN2 text-to-image feature can now be experienced on NVIDIA AI Demos, where visitors to the website can experience AI through the most recent demonstrations from NVIDIA Research. With the flexibility of text prompts and sketches, GauGAN2 lets users develop and personalize scenes faster and with finer control.
An image worth a thousand words now takes just 3 or four words to produce, thanks to GauGAN2, the most recent version of NVIDIA Research studys wildly popular AI painting demo.
With the press of a button, users can produce a division map, a top-level outline that shows the place of items in the scene. From there, they can change to drawing, tweaking the scene with drafts utilizing labels like sky, tree, rock and river, enabling the clever paintbrush to integrate these doodles into sensational images.
The deep learning model behind GauGAN enables anybody to funnel their creativity into photorealistic masterpieces– and its easier than ever. Simply type a phrase like “sundown at a beach” and AI generates the scene in genuine time. Include an additional adjective like “sunset at a rocky beach,” or swap “sunset” to “afternoon” or “rainy day” and the design, based on generative adversarial networks, instantly customizes the photo.
An AI of Few Words
The AI model behind GauGAN2 was trained on 10 million top quality landscape images utilizing the NVIDIA Selene supercomputer, an NVIDIA DGX SuperPOD system thats among the worlds 10 most powerful supercomputers. The researchers used a neural network that finds out the connection in between words and the visuals they represent like “winter,” “foggy” or “rainbow.”.
The demonstration is one of the first to combine several modalities– text, semantic segmentation, sketch and design– within a single GAN structure. This makes it faster and simpler to turn an artists vision into a high-quality AI-generated image.
Compared to state-of-the-art models specifically for text-to-image or segmentation map-to-image applications, the neural network behind GauGAN2 produces a greater variety and higher quality of images.
GauGAN2 combines segmentation inpainting, mapping and text-to-image generation in a single model, making it a powerful tool to create photorealistic art with a mix of illustrations and words.
The GauGAN2 research demo shows the future possibilities for effective image-generation tools for artists. One example is the NVIDIA Canvas app, which is based upon GauGAN innovation and readily available to download for anybody with an NVIDIA RTX GPU.
It does not just develop reasonable images– artists can also utilize the demo to portray transcendent landscapes.
NVIDIA Research has more than 200 scientists around the world, focused on locations including AI, computer vision, self-driving cars, robotics and graphics. Find out more about their work.
Envision for example, recreating a landscape from the iconic planet of Tatooine in the Star Wars franchise, which has two suns. All thats required is the text “desert hills sun” to develop a beginning point, after which users can quickly sketch in a 2nd sun.
Its an iterative procedure, where every word the user types into the text box adds more to the AI-created image.
Instead of requiring to extract every component of an envisioned scene, users can get in a quick expression to quickly produce the essential features and style of an image, such as a snow-capped range of mountains. This beginning point can then be tailored with sketches to make a specific mountain taller or add a couple trees in the foreground, or clouds in the sky.