Print by Numbers is my MIT SuperUROP project. SuperUROP allows students to do advanced undergraduate research in their field, and for EECS majors, it is also accompanied by a 2 semester seminar-style class in which a video and poster of the research is produced after the first semester and a research article is produced at the end of the second semester.

I am currently enrolled in the SuperUROP program for the Fall 2022 and Spring 2023 terms.

Print by Numbers is an investigation that aims to digitalize the art of printmaking and giving it a space in the digital world. I am working in the Human Computer Interaction Lab in MIT CSAIL with Stefanie Mueller as my faculty supervisor and Ticha Sethepakdi as my direct supervisor. The investigation is divided into three main components:

1. Image processing: Create a program that can take an input image and a number k, and quantize the colors of the image to k colors and produce layers in Black and White of these colors ready to be laser cut.

2. Printmaking technique: We are using a laser cutter to create the layers for stamping. But experimentation with what material to use to create them is necessary since we are using high-quality images in our image processing algorithm that can have different results when put in the laser cutter and generating the stamps. For example, rubber stamps allow for more detail but wood stamps are more rigid and stable.

3. User interface and Outputs: We aim for Print by Numbers to be an useful tools with people with no previous technical or artistic background. So it is important to have an investigation on the best way for users to use our program, what things should be available for them to define and what things are better kept in the black-box of our program.

So far, the investigation has been mostly targeted to the image processing section. We have successfully created an image processing algorithm using K-means clustering that quantizes an input image into a specific number of colors and then creates the layers that are ready to be laser cut and a color palette of what the colors of these layers should be. We are still fixing some details like problems due to the pixelation of input images.

We also did some initial trial in of the laser cutting of layers using a draft of our quantization algorithm testing for both wood and rubber as layer materials. However there is still a lot of room for improvement in our technique.