by ryklin » 21 Oct 2014, 17:31
I wanted to update everyone on my experiences comparing Tobii's EyeX (TobiiEyeXSdk-Cpp-0.32.384) with the EyeTribe (9.41 Cpp SDK)
The EyeX Pros:
It seems that the EyeX may require less processing overhead than the eyetribe. I don't have data to confirm this yet.
If my pilot studies are correct, on slower computers, comparing the two trackers on the same exact benchmark code and hardware, the EyeX performs better. It has less latency (lag), and drops less samples. However, on faster machines, the two may perform equally, and in certain cases the EyeTribe may be preferred by some users and for certain applications. So this may truly be a neck to neck race. Keep in mind that other more expensive eye trackers ($30K USD + will totally outperform these).
The EyeX calibration system is much more developed. Among other features, it stores calibration profiles for different users, even with or without glasses for the same user. The EyeX also perpetuates the calibration even after the computer is restarted. This makes reconnecting your application to the eye tracker server seamless, at least in my experiences. The EyeX also has a magnetic bracket used to attach the controller to the surface of the display.
The Eye X cons:
The EyeX, so far in my opinion, is that it's bigger than the eyetribe and has annoying red lights (IR emitters). The eyeX is missing pupil diameter, an important and often critical measure.
The EyeTribe Pros:
The eyetribe has less noticeable IR emitters, and fits nicely on my 13 inch laptop. While the eyetribe *SDK* is less functional than the EyeX SDK, it is still more stable and reliable. Tobii is constantly updating the EyeX SDK, and this is annoying and terribly difficult to keep up with. Tobii keeps deprecating functions, and breaking old code. Eyetribe doesn't do this.
In my subjective opinion, the eyetribe gaze filter is better than Tobii's. It's certainly possible to disable all filtering for both devices, and then for programmers to implement their own filters, including replicating eyetribe's and using it for both controllers. However, the eyetribe filter is very good compared to anything else I've worked with.
The EyeTribe Cons:
The eyetribe, in my opinion, is slightly behind in overall development by comparison to the EyeX. Its calibration accuracy and precision is good and probably comparable to EyeX (no data to back that up, just my overall, personal, subjective impression), but is missing some important functions at the time of this posting. The eyetribe 60Hz mode doesn't seem to work for me. This effectively means that the eyetribe samples slower than the eyeX (I think the eyeX samples at 50Hz but I might be wrong). For some applications, and on slower computers, this may be a problem. The Eyetribe has issues with the Z coordinate (user distance).
I love the competition, in overall both companies are leading the eye tracking revolution that is at our door step.
.