Face-Off: My response to the 2018 RISD Assignment “Threshold”

As part of my 2018 application to RISD’s BFA program, I took part in the RISD Assignment to create two responses to the word “threshold.” I was admitted but ultimately chose to accept CCA’s Interaction Design BFA instead. Although RISD recently removed the Assignment, I’d like to share my project and process to inspire others who might be considering art school!

Face-Off: My response to the 2018 RISD Assignment “threshold”

Threshold measures a line of absolute that divides two aspects onto either side of the line. I began by exploring this duality that a threshold creates through a common entity — the human face. The human face can easily recognize other faces, and is itself easily recognizable. Thus, we can imagine the face as a platform that gives and receives information, and I decided to measure the threshold of this information exchange — how easily a face can be recognized.

In the first of my two explorations, I applied the threshold of facial recognition to our interactions with technology. I chose Snapchat’s facial recognition system as a representative for digital facial detection, as it was one of the most accessible systems with an extremely large database of facial samples during 2017. I used the detection system to determine the minimum threshold of information required by the program to determine a face.

Early tests to determine which alterations that can obscure the face from digital detectors. You can test for yourself which faces can be picked up by smartphone filters!

I experimented with images of fellow artists around me using found materials to find the minimum distortion required to obscure the face from Snapchat’s detection system. This resulted in a collection of portrait collages that appeared both random and intentional. I used a found assortment of materials combined with careful staging and iterative testing. For example, in one variation I kept the overall position of the facial features and created uneven and sporadic ripples away from the central traits. Ultimately, I created a collection of images that are easily recognizable by the human eye as portraits of faces, but unrecognizable to a digital facial detection program. (This project was created in 2017 and was tested on Snapchat’s facial recognition program at the same time. Results with newer technology might differ.)

My first response to the RISD Assignment.

As the first exploration concluded in a set of images, I decided to bring my second round of explorations into a three-dimensional space. Furthermore, I wanted to push the threshold of recognition further to challenge the ability of a human to recognize a fellow face. I began by formulating a list of base features that allow a face to be detected by Snapchat, then sculpted the features into a bust-like model, which resulted in an organic structure that holds an uncanny human-like appearance.

Notes on the minimum threshold of information required for a facial detector to determine a face, as well as a paper mache bust with the features sculpted into it.

I then projected images from my first response onto the model, turning them from flat surfaces to three-dimensional landscapes with seemingly human faces emerging from the extrusions. The overlay of images from the threshold of recognition for a digital recognition program on top of a structure on the threshold of human recognition creates an object that challenges the facial detection system of the human eye.

My second response to the RISD Assignment. Which images do you recognize as faces, and which facial features do you find to be most easily recognizable?

The surfaces distort themselves at different angles as the observer circulates the object, weaving the threshold into every perspective, while changing its visual appearance. In future iterations, I would like to explore a responsive model or environment that can provide the user with a mirror-like reflection that distorts the user’s image and perception in real-time.