Why it’s interesting
PatternPortrait uses a custom graph neural network trained on one pattern scribble made by the artist. Visitors do not train the style themselves. Instead, the installation takes their photograph, removes the background, detects edges, applies the learned pattern language, and sends the vector drawing to a plotting robot. The result combines a visitor’s likeness, the artist’s marks, and the machine’s transformation in a physical sheet of paper.
How to try it
On Bleeptrack’s official page, begin with the portrait collection and then compare the plotting-process images with the video documentation. The complete loop requires an exhibition visit, a photograph, and time for the robot to draw. The listed events from 2022 through 2025 are historical records, not a current permanent installation or a public upload tool.
Requirements and sources
The official page verifies the GNN, vector-path input, one-sample training, photography, edge detection, and robotic plotting workflow. The earliest documented exhibition year is 2022, so the stored date uses year-level precision without inventing a day. Facial images are captured on site; retention and consent depend on the venue’s current notice. The 3840×2160 cover is served by the artist’s project page and credited to Bleeptrack.