Introducing the CPPN Playground!
The main interface of the CPPN playground; reproducing a cool rendering of a black hole.
As we saw in the business card blog post we love “CPPNs” here.
We wanted to share the fun, so I made a little JavaScript app where you can play around with them. You can access it here: CPPN Playground.
There’s a few main things you can do:
- Look at the cool pictures! And adjust the network to get more.
- Train the network to reproduce a specific picture.
- Download a big (1000x1000) version of the image!
Here’s a big version of the black hole image:
The way it works is almost exactly as laid out in the business card post. The reason we’re able to reproduce images is that we use the machinery of deep learning to ask the network to match a given image.
Over time I’ll add a better explanation to the page, but for now I just wanted to get it out there so people can have a go and give some feedback!
It’s also open-source, built using TensorFlow.js and Elm, so feel free to clone it and hack around. There’s plenty more features to add! (I’ve put a list at the bottom of the readme).
Hope you enjoy it!
Reach out if you have any questions/comments/suggestions!
