Introduction · Ch. 1 · Figs 0.1, 1.4
Chair
In the mid-2010s, Paglen's Berlin studio built a computer-vision framework it called Chair. They pointed a webcam out the window and switched modes with a few keystrokes to watch the street the way the software did. Each mode here is a different machine Umwelt, a set of “lenses and blindfolds” (Introduction). Each one picks up a few things and misses everything else.
Loading the sample…
Lens
Blindfold
What does this image do?
Chapter 1 argues that most images are now made by machines for other machines, with no person looking at them. In every mode above, the frame stops being a picture: it becomes a table, a field of gradients, a set of points to match, boxes with scores, and detectors switching on. None of these outputs is meant for a person to look at. Each one is an input to a next step: a match, a count, a classification, a decision.
So the useful question is less what the frame shows than what it triggers. Point the camera at a coffee cup, a face or a window. Watch which units fire and which boxes appear, and which parts of the frame the machine doesn't register at all.