Nvidia’s latest AI-driven painting tool produces breathtaking landscapes with remarkable realism, starting from the simplest brush strokes.
Transforming doodles into magnificent landscapes just got easier with the NVIDIA Canvas app.
Now in its free beta version, the app brings the real-time painting prowess of GauGAN to anyone equipped with an NVIDIA RTX GPU. Crafted by the talented NVIDIA Research team, GauGAN has captured the imagination of creative communities globally, using deep learning models to transform rough sketches into breathtaking scenes.
Taking it a step further, NVIDIA Canvas empowers creators to paint by material rather than color, leveraging AI to transform brushstrokes into lifelike images. The app provides an instantaneous display of the photographic result as users paint, eliminating the need to wait and allowing them to witness the realization of their vision in real-time.
NVIDIA Canvas is a proud addition to the NVIDIA Studio program, offering creators the hardware and software tools they need to bring their creative visions to life.
Begin with a few strokes and lines to sketch basic shapes, selecting from a palette of real-world materials such as grass or clouds. Instantaneously, the AI model transforms the screen, revealing captivating results. Sketch four swift shapes, and a breathtaking mountain range materializes. Add a few more lines, and a picturesque field unfolds.
Stunning Landscapes with NVIDIA Canvas
Introduce a pond, and adjacent elements like trees and rocks reflect in the water. Swap materials—changing snow to grass—and witness the entire image shift from a winter wonderland to a tropical paradise.
Begin by sketching uncomplicated shapes and lines using a palette featuring real-world materials such as grass or clouds. The AI model promptly populates the screen with dazzling outcomes—four swift shapes unveil a breathtaking mountain range, and a few additional lines craft a picturesque field.
Introduce a pond, and watch as nearby elements like trees and rocks manifest as reflections in the water. Experiment with materials, swapping snow for grass, and witness the complete transformation of the image, shifting from a winter wonderland to a tropical paradise.
The tool empowers artists to apply style filters, transforming a generated image to embrace the distinctive style of a particular painter. Unlike merely assembling fragments from other images or employing cut-and-paste techniques, NVIDIA Canvas crafts entirely new images, mirroring the process of a traditional artist.
For artists seeking to elevate their digital creations, NVIDIA Canvas allows them to save their work as a .psd file and seamlessly import it into Adobe Photoshop for further refinement or integration with other artwork.
The creative potential is boundless. NVIDIA Canvas serves as a catalyst for artists to swiftly generate ideas—an often daunting task when confronted with a blank canvas at the outset of a project.
Moreover, it introduces time efficiency across various creative domains. Imagine an art director facing a tight deadline, needing to enhance the hero product in an advertising campaign. A carefully crafted landscape is vital for completing the shot. With NVIDIA Canvas, the director is just a few clicks away from generating multiple exquisitely rendered images. In a matter of moments, the tool presents a variety of options for the client to evaluate, saving crucial time before finalizing the showcased product.
NVIDIA Canvas features nine styles that can alter the visual tone of a painting and offers 15 distinct materials, spanning from sky and mountains to river and stone. The ability to paint on different layers allows for the separation of elements. Whether starting from scratch or launching and modifying one of the app’s premade scenes, users can find a constant source of fresh inspiration.
The Creative Mechanism of NVIDIA Canvas
NVIDIA Canvas employs a variant of AI known as generative adversarial networks (GANs), comprising a generator and a discriminator.
The generator’s role is to transform material maps into visually coherent images. Meanwhile, the discriminator possesses knowledge about authentic ponds and lakes, recognizing the presence of reflections, and instructs the generator to produce a compelling imitation. Achieving lifelike portrayals for artists, the GAN has undergone training on an NVIDIA DGX system, utilizing an extensive dataset of over 5 million images.
Get a Taste of Canvas Beta, Share Your Artistry!
Grab the Canvas beta, explore its features, and share your thoughts on the NVIDIA forums.