I stole this writer's thumbnails

we’ll use Generative AI, specifically a CycleGAN model. It can do cool things like this: <img alt="" src="https://miro.medium.com/v2/resize:fit:593/1*13ngOMY7T1qFtX1IL5FsVw.jpeg" style="height:276px; width:593px" /> <a href="https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix/blob/master/imgs/edges2cats.jpg" rel="noopener ugc nofollow" target="_blank">Image Source</a> To be able to reproduce Martin’s image, we’ll need to train the network. Essentially, we input a bunch of images into the network, and we have two agents. One agent is the painter, generating fake images. The other agent is the guesser, trying to guess if the image is fake or real. Usually, you have thousands of images for this process, but I was only given around 30 to work with. We have to start somewhere else. In preparation for this story, I trained a network that is able to generate landscapes from edges: <a href="https://medium.com/write-a-catalyst/i-stole-this-writers-thumbnails-fbe9f64f43dd">Read More</a>