Eye Tracking



Virtual Reality



Augmented Reality

We have created eye tracking add-ons for the current range of leading VR and AR platforms. Our add-ons enable eye tracking without compromising your VR or AR experience.

You can use the VR or AR hardware you already own, and augment it with our tool-free, drop-in eye tracking hardware.

You can use our software platform, or roll your own. With our open source software and direct access to the raw eye video feeds you are never locked into a platform.

Hardware Add-Ons

Our add-ons give your AR/VR Headset eye tracking superpowers. Eye videos are transferred to your host via USB 2.0.

The hardware is easily attached and removed. Eye cameras can run up to 120hz and have a latency of 5.7ms. Fast enough for interaction applications and detection of saccades and quick fixations.


There is no one size fits all approach - that is why we made Pupil Service.

Pupil Service is based on Pupil Capture and gives you access to all data: eye videos, to pupil data, gaze data, and more. Get the lastest release below.

We also maintain and contribute to a community driven repository of tools for eye tracking in head mounted displays called HMD-eyes

Technical Specifications & Performance

VR and AR add-ons share many of the same specifications due to our modular design approach. There are subtle differences in specs and performance where noted in the table below.

Basic Capabilities
Mono | Stereo Both, depending on desired setup
Tracking Frequency 120Hz
Field of View HTC Vive, Oculus DK2: up to HMD limits. Epson BT-200: greater than 100 deg
Gaze Accuracy ~1.0deg
Gaze Precision ~0.08deg
Camera Latency 5.7ms
Processing latency 3-4ms on i5 CPU
Resolution 640x480
Connection USB 2.0
Saturation Interconnected bandwidth USB 2.0 ~60% saturation
Data Output
Combined Eye Gaze Yes
Independent Eye Gaze Yes - 1 gaze point per eye
Pupil Position Yes
Pupil Diameter Yes - units in pixels using 2d appearance, units in mm using 3d model
Velocity | Acceleration Can be implemented if desired
Eye Accomodation Yes, using binocular vergence
Method 5 point
Power | Performance | Memory
Host CPU % Utilization 50% of one core per eye
SDK Through network message based API
Plugin for Game Engines Reference plugin implementation for Unity3D


Our goal is to work with and support a community of researchers and developers in creating reusable open source software for eye tracking in HMDs.

Many researchers and developers have already thought of, or implemented eye tracking software HMDs. Instead of continuing to reinvent the wheel, we have formed a community with the goal of sharing knowledge and tools.

Fork the hmd-eyes code on Github. Chat with the hmd-eyes community on Gitter.