The existing style transfer apps in Prisma and Facebook are stylistic but not photorealistic but now a Cornell University research group in collaboration with Adobe have come up with a process to augment the style transfer in such a way that it transfers the look of one photo onto another while still looking like a photo. The paper is titled as, “Deep Photo Style Transfer” and the process involves adding another layer of neural networks on the process to the existing methods of style transfer in such a way that the details of the original images are preserved. Cornell professor Kavita Bala says, ” People are very forgiving when they see [style transfer images] in these painterly styles. But with real photos there’s a stronger expectation of what we want it to look like, and that’s why it becomes an interesting challenge.”
Why not share this article?
