Generative Adversarial Networks (GAN’s)
The end of the line for Stock Photography?
Chris Marquardt and Ade Stock recently discussed the advances in AI technology on their Future of Photography Podcast Episode 63
Chris described the application of Generative Adversarial Networks to create virtual, photo-realistic faces. As the technology matures it seems that companies requiring a specific look will be able to ‘dial it in’ to a computer program and generate a virtual model to order. It is certainly worth a listen and to spend an hour or so googling GANs.
AI Generated Cats
I came across a few interesting sites that are worth a look including an example of GAN to generate a cat from a sketch of a cat… pix2pix
Nvidia have a number of interesting demos of this technology including using AI to remove noise from photographs which will be an amazingly useful feature to take advantage of in those low-light situations