Lines of light ranged in the nonspace of the mind, clusters and constellations of data. Like city lights, receding.
- William Gibson, Neuromancer
It’s no secret that – in some eternal quest to palliate my nostalgia – I’ve developed a mild obsession with text-mode art. Years ago, I learned all the relavent algorithms, techniques, and code pages in order to best simulate text-mode in modern browsers – an effort that resulted in (long since abandoned) projects like ASCIImeo and Flashterm. Recent explorations in bridging the digital-physical divide (plus a sense of withdrawal) have caused me to revisit image-to-text conversion with a quick experiment.
For many of us that have grown up with computers, text-mode art represents something deeper than nostalgia. It is an artform manifested from technological constraints, inspired by the same hacker ethos that build the early machines used to produce and view it. Fundamentally, it is both an expression and prisoner of the system it inhabits. This latest experiment attempts to free ASCII art from the confines of the screen and enable it to exist in physical space – with light and paint.
I wanted to approach image-to-text conversion a bit differently this time around, focusing primarily on speed. In order to achieve this, all the real-time conversion work was moved to the GPU. The resulting program (available as an openFrameworks add-on) can scale to huge dimensions with little strain.
A nice side-effect of writing the new library in OpenGL was easy portability to WebGL, enabling a unique twist on exploring the physical world:
