By Jack Williams
This amazing AI app allows users to turn what were once the simplest of doodles into incredible landscapes.
From what were basic blobs, lines and block colours that could represent land, sea or mountain ranges, the technology then works to convert their layout into a realistic landscape ‘photograph’.
The project was overseen by researchers at NVIDIA, who recently released a video to showcase the technology.
Named GauGAN – a combination of famous painter Paul Gauguin and the technology used, generative adversarial networks (GANs) – the app uses different colours for each elements.
Bryan Catanzaro, VP of Applied Deep Learning Research at NVIDIA, said: “Wouldn’t it be great if everybody could be an artist?
“If we could take our ideas and turn them into compelling images.
“This technology allows us to create a smart paintbrush, so that if you wanted to create a new picture, you can just draw the shapes of the objects that you want and the neural network can then fill in all the details.”
A block of sky blue on the MS Paint-like app is converted into a photo-like sky, while placing greys on such a block would then generate clouds.
More impressive still, the deep learning network also picks up the likes of reflections and shadows – so if a tree is above a lake, a reflection will be generated on the water.
A research paper on the technology will be presented at the Conference on Computer Vision and Pattern Recognition conference, in June.
Bryan added: “I really think this technology is going to be great for architects, designers – people making virtual worlds to train robots and self-driving cars.
“It’s like a colouring book picture that describes ‘here’s where a tree is, where the sun is, where the sky is’ – and then the neural network is able to fill in all of the detail and texture, the reflections, shadows and colours, based on things that its learned form a large database of real-world images.”