Parallel Multiscale Autoregressive Density Estimation — mentioned by gwern

Thu, 06 Jul 2017 23:27:38 GMT

It's a combination of GPU RAM, slowdowns (remember it's squared in dimensions), and stability (larger is more unstable end-to-end). Arguably, state of the art in image synthesis is DeepMind's PixelCNN: "Parallel Multiscale Autoregressive Density Estimation" https://arxiv.org/abs/1703.03664 , Reed et al 2017: generating 512px photorealistic images & video with PixelCNNs rather than GANs. Also good is StackGAN which does ~200x200ish but there's no reason it couldn't go up to 500x500 (just pop in a third upscaling stage). There's far more work on GANs than PixelCNNs (see the…

Paper and discussion history. A mention is not an endorsement.

Parallel Multiscale Autoregressive Density Estimation — HN submission

Sat, 27 May 2017 17:40:22 GMT

Generating 512px photorealistic images and video with PixelCNNs

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Parallel Multiscale Autoregressive Density Estimation · HackerPapers discussions

Parallel Multiscale Autoregressive Density Estimation — mentioned by gwern

Parallel Multiscale Autoregressive Density Estimation — HN submission