Download | Reference | Changelog | License (MIT)
Random Art Generator is a command-line application to produce generative art.
When passed a target image, it will start painting elements on a blank canvas, essentially creating new images at random. Resulting images that are mathematically closer to the target image are kept, and used as a new canvas; results that aren't get discarded. This repeats until the desired number of attempts or generations is achieved.
For example:
rag profile.jpg --generations 100Technically known as hill climbing, this process allows the creation of new images that progressively converge into the target image, with a style all its own depending on the chosen painting algorithm and its arguments.
This application is written in Rust, and downloadable binaries are currently available for macOS, Linux, and Windows. A webasm target - so it can be used as a JavaScript library - is planned for later.
Notice: most of the development happens in the dev branch, as the move to GPU-accelerated filters (replacing the manual pixel buffer filtering) is finalized.