5 ms·
Although there is one higher res image available, which is better than usual. Of course it's high enough res to show that there's visible artifacting. But this
by p1necone 8y ago
Although there is one higher res image available, which is better than usual. Of course it's high enough res to show that there's visible artifacting. But this is pretty impressive.
I wonder if an easy hack would be to train a second separate neural network to clean up the kind of artifacting that you get out of the first neural network...
- squarefoot 8y ago"I wonder if an easy hack would be to train a second separate neural network to clean up the kind of artifacting that you get out of the first neural network..." Shhhh, don't give the evil web overlords ideas for new nasty captchas :)
- pontifier 8y agoMy understanding is that this is exactly how these images are generated. One network makes the image, and another tries to find flaws... They both get better together. Here's the first 2 Google results for "GAN" meaning "Generative Adversarial Network" https://skymind.ai/wiki/generative-adversarial-network-gan https://skymind.ai/wiki/generative-adversarial-network-gan https://en.m.wikipedia.org/wiki/Generative_adversarial_network https://en.m.wikipedia.org/wiki/Generative_adversarial_netwo...