NE 1

My inaugural upload, a VGGAN-CLIP image, early stopping, of a colorful nuclear explosion a la Takashi Murakami.

I was initially wary of deep networks for generative art, and I found a lot of the images appearing on Twitter less than appealing. But after a lot of experimentation, this image - and several others like it - just blew me away (no pun intended). I'm now rather bullish on the whole enterprise, and can't wait to see where this will go.

Special thanks to the authors of VQGAN (https://compvis.github.io/taming-transformers) and CLIP (https://openai.com/blog/clip), and Katherine Crowson for putting together the colab combining them.