Fihan

The Machine Looks At Art

A sample photograph as seen through Fihan's lens

Give it a photo, video, or sound. Watch six different algorithms — some decades old, one from 1908 — decide how to see it.

Coming soon

A few things inside the machine

1958 Rosenblatt Perceptron Learns a single line that splits the image in two — one of the first trainable neural networks ever built.
1908 Voronoi Tessellation Splits the photo into cells, each claimed by its nearest point — originally a tool for pure mathematics.
1968 Sobel Edge Detection Pulls the structure and edges hidden inside an ordinary photograph.
1980s Demoscene Plasma Field Turns the photo's own colors into flowing motion — a trick born on the Amiga.