
NVIDIA's AI that turns text and sketches into photorealistic landscape images
GauGAN2 is NVIDIA's AI-powered image generation tool that combines text-to-image generation, segmentation mapping, and inpainting into a single GAN framework. Built on NVIDIA's research and trained on 10 million landscape images, it enables artists and designers to create photorealistic scenes from simple text descriptions or rough sketches in real time. The technology powers NVIDIA Canvas, a free desktop application for RTX GPU users.
Type a phrase like 'sunset over mountains' and instantly generate a photorealistic landscape image
Combine text prompts, segmentation maps, sketches, and style references in a single workflow
Generate and preview images instantly as you type or draw, powered by NVIDIA RTX GPU Tensor Cores
Transform simple brushstrokes into detailed, realistic landscape artwork using AI-powered material recognition
Apply nine different artistic styles to modify the look and feel of generated images
Choose from 20+ materials including sky, mountains, river, stone, and vegetation for granular scene control
Separate elements into different layers for precise control over each part of the composition
Rapidly generate landscape concepts and environment art for films, games, and creative projects without starting from a blank canvas
Create photorealistic background scenes for film and TV production, quickly iterating on compositions and lighting
Prototype game world environments and terrain layouts with AI-generated landscapes that can be refined in professional tools
Experiment with different scenes, moods, and compositions by combining text prompts with simple sketches for artistic inspiration
Export artwork as PSD files for seamless integration with professional editing workflows

AI character and video generation with unmatched consistency