Deleting objects from our photos has been commonplace for a while. There isn’t much you can’t ‘paint’ away in photoshop but what about adding things to our photos?
I am not talking about composites where you import other images and layer things to make a final image. I am talking about generating things from scratch to add to our photos.
Words like ‘artificial intelligence’ and ‘computational photography’ are buzzwords – but what do they mean in practice?
GANPaint Studio has made a demo to show you how to add objects into your photos by using neural networks.
The project is based on research by IBM and MIT and utilizes a GAN algorithm to do its magic. Through machine learning, the software is able to ‘paint’ objects as opposed to colors. You can watch the video below or go to http://ganpaint.io/ for a demo. The demo is currently only able to work with church images.