Layered 3D

Layered 3D is a small, static glasses-free 3D display. We introduced it in a technical paper at SIGGRAPH 2011. The key idea here is that you can build a simple and inexpensive 3D display by stacking a few printed transparencies. What’s scientifically interesting about Layered 3D is that the optimal patterns to depict a given 3D scene can be automatically¬†computed using computed tomography algorithms that are usually employed for medical imaging in hospitals. This project was featured as a SIGGRAPH 2011 Paper Highlight, as a SIGGRAPH 2011 Proceedings Cover Feature, and in The Boston Globe.



Building a Layered 3D display is relatively straightforward – we usually have students in our MIT Media Lab course on Computational Photography build one as an assignment. First, you download the software that allows you to generate the patterns for a desired 3D scene from this website. Then you either capture or render your own light field or download one from my synthetic light field archive. You run the code with the desired light field and print the resulting patterns by following these instructions. You will get something like these patterns


I prepared transparencies for a couple of difference scenes, you can just cheat and skip the code & processing part and download the archive: Layered3D_transparencies. These layers are computed to roughly fit on an iPhone (which you can use as a uniform backlight) – you need to print them such that each of the patterns (as indicated by the cutting marks) has is precisely 59.88 x 44.91¬† mm large when printed. They are computed to work with¬†5/64″ clear acrylic spacers in between them. You can order the acrylic at McMaster (get the 5/64″ strengthened UV-resistant acrylic sheets) but you’ll have to cut it somehow. We are fortunate to have a laser cutter, but you can probably use a saw as well.

Once you’ve cut the acrylic, printed the patterns, and stacked them (aligning them carefully using the marks) you just need to rear-illuminate them with some sort of a light box. The easiest thing is to use your phone screen and make it white (e.g., with the flashlight app). Zach Honig from engadget took a few good shots of some of the Layered 3D displays we presented at SIGGRAPH 2012 Emerging Technologies.

Here is a shot of a a prototype that shows a couple of different scene. I built it for the original SIGGRAPH 2011 paper:





2 Responses to Layered 3D

  1. psartor says:

    Hi all,

    in the “synthetic light field” section, you’ve a collection of image set created with POVRay, I understood.
    Would it be possible to archive not only the LF itself, but also the POVRay script used to generate them?
    The problem I’m having is that some crucial parameters, like focal length of the camera, are missing.




  2. psartor says:

    Hi all,

    I’m interested in evaluating the performances of the 3D layered displays. Specifically I would like to compare one original picture, for example one from the source views, with the generated one.
    The Matlab code should provide such images (matrices), unfortunately I was not able to find any documentation explaining which variables are what.
    I assume the “reconstructLayers.m” script will generated everything that is needed, but it would be nice to know which variable (matrix) generated contains the proper inverse view.
    I assume it is “lightFieldRec”, but I would like to have a confirmation from you.

    Thanks a lot in advance,