F

OverviewConceptExhibitingBuild

F

mrrior

Type
Academic
Domain
Interaction Design • p5js
machine learning • posenet
Year
2021
Mrrior is contemplative look at ourselves through the reflective lens of a mirror made possible through machine learning.

Concept

When coming across a centuries old mirror at The Met, I couldn’t help but wonder the countless faces that have stared into it, and the countless faces it had stared back into. If mirrors had a memory, what could it possibly remember?

Using the PoseNet deep learning model to track bodily movements, the mirror recounts its sightings in textual descriptions. Its recollections are on display through a mirrored surface.

We gaze upon it gazing upon us, and we take in its observations as we recount ours.

Exhibiting

build

Mrrior is a union of secondhand objects. An old Acer monitor was removed from its casing, leaving the display and the its logic board. The display was affixed to a picture frame, with its glass treated with privacy film. Areas of exposed glass between the display and the frame was covered with matte paperboard.

Laterally, the display is held in place by “ribs” made of folded paperboard as a lightweight solution. To secure the display in the vertical direction, MDF strips were mounted across the frame just above and below the logic board casing.

The display connects to a computer through a DVI to HDMI cable, and is rotated to accommodate the vertical dimension.
Special thanks — Karolina Luckiewicz.