As part of the experience prototyping course at UID, I built a small controller that manipulates two video feeds using physical controls. It also controls an optional audio feed.

I used a person's face as the subject matter. Two slightly offset cameras (one being a laptop's built-in webcam; the other a mounted webcam) focus on and display the face looking at the screen. The person turns knobs and fader to create distorted versions of their face.

A still image can be saved by pressing the spacebar at any time.

The program was built in Processing and the physical controls were scrapped together using the Phidgets toolkit.

The core image is intentionally pixelated from the beginning, but colour, filter, and other artifacts were user selectable. The LCD display gives feedback on current settings.

The range of possible outcomes is quite wide. I enlisted volunteers to help me explore the limits of the tool, letting them save stills of images they found interesting.

Asked to spend just three minutes with the tool, many stayed much longer playing.

The audio response was a bit trickier. If turned ON, the beat visualizer draws a grid of dots over the video feed according to the audio input. A monome device controls which dots in the grid should be on, and which should be off.

This part of the tool was less successful, as generating believable real-time response to the audio was a difficult task. Also, the visual component of the project demanded enough attention from the users on its own.