3D Printed Paintings started in USC’s chapter of 3D4E and continued as a personal project. It’s primary goal is to transform 2D images into 3D models that can then printed. The source code can be found here.

---

Process

To create a 3D model from a 2D image I employed numpy-stl. The image is first converted into grayscale, then each pixel’s grayscale value is mapped to the height of the point in the 3D model.

Iteration

The initial program created 3D models that were spiky and difficult to print due to the possible high difference in grayscale values in a picture, along with each pixel only being assigned one point. The first solution to solve this issue was to apply a blur over the image before converting it into a model which made the images easier to print but lost detail and recognizability. The solution that produced the best reults was giving each pixel of the source image a face instead of a point. This makes the art much more recognizable and introduces a pixel art style that works really well with the Pokemon that were printed.

---

Results

Various Pokemon with pixel variation.

Van Gogh portrait using a blur and other post processing mounted in a frame.

Initial test print with an early version of the program of the Mona Lisa.