Skip to main content

Projects

An AR demo you can try right now, and open-source Unity toolkits that other labs actually run.

All three are years past the work that produced them, and I still maintain them. That is the part worth stating plainly: they are public APIs that other people's work depends on. Keeping one alive means holding an interface stable for strangers and documenting it well enough that they can adopt it without asking me. It also means a new version still has to ingest the files an old one wrote, and where a format had to break, it shipped with a migration script and a deprecation notice.

AR Business Card​

Interactive AR business card that runs in the browser. Personal project, sole developer.

Point a camera at my business card and an augmented reality scene anchors to it, with links you can tap. Nothing gets installed. Image tracking runs on OpenCV compiled for the browser. It hands the card's position to a Unity WebGL build, which draws the scene over the live camera feed.

Trying it takes a camera and the card itself. Open the card image on a second screen or print it, then load the demo on a phone and point the camera at the card.

Try the demo

bmlTUX​

Unity experiment framework. Open source, sole developer, still maintained.

A framework for building perception experiments in Unity. It handles the parts every study needs and nobody wants to write again: trial tables, block randomization, counterbalancing, and per-participant output. A VR-ready experiment goes together in minutes with very little code, which matters when the person building it is a vision scientist rather than a programmer.

It is used by vision-science labs internationally and the design is peer-reviewed, in i-Perception. See the publications page for the citation.

biomotionlab.github.io/TUX

bmlSUP​

Statistical body avatars in Unity. Open source, sole developer, still maintained.

Displays realistic animated body models in Unity and VR with minimal coding, built on SMPL-X statistical body models and drivable from motion capture. It came out of needing believable, measurable human bodies for perception research, where a stylized game avatar will not do because the body's proportions are the thing under study. It was presented at IEEE VR in 2021, and the publications page has the citation.

biomotionlab.github.io/SUP

UnityH5Loader​

Unity package for loading HDF5 data. Open source, sole developer, still maintained.

A small library for reading H5 files directly in Unity, wrapping PInvoke so a project can pull in float, int, and string datasets, including 2D arrays, without a manual native-interop layer. It exists for the same reason bmlTUX and bmlSUP do: research pipelines produce data in Python and need it inside Unity, cleanly and without friction.

Distributed as a versioned package through Unity's Package Manager, with semantic-release handling releases.

github.com/AdamBebko/UnityH5Loader