Page 1 of 1

TrackIR 5 Performance compared to v4 and FaceTrackNoir

Posted: Thu Aug 23, 2012 6:24 pm
by drliebs
I realize I am in a forum with TrackIR 5 users and I would expect that you all approve of the product. What I am really asking is how does v5 compare to v4, how does TrackIR compare to FaceTrackNoir, so that I can estimate how well it works compared to something I have tried. If you have used the earlier version, and more importantly, if you have tried FaceTrackNoIR, I want to know how you rate the products.

I wish I could try it before I buy it. I don't know how to do that so before I make a big investment I would like to know if v5 and the trackclip pro make a big difference too. I have been playing around with FaceTrackNoIR (that is how I found out about TrackIR) and it is really cool, but leaves me wanting more precision. This has been my only experience with head tracking.

FaceTrackNoIR works, but is just a little sluggish and not precise enough to really satisfy (it's free and that is awesome). If that was all I had, I would use it in some circumstances. In basic flight it is pretty cool, but in combat things jump around a little too much. For the money, does TrackIR give you high precision compared to the freebie?

For example, I am playing Apache Air Assault, and in this game you can control the cannon with the mouse. Response and accuracy with a mouse are what you would expect. Controlling it with FaceTrackNoIR works, but it is not accurate enough; i ended up drawing circles around the target.

I suppose I am looking for mouse like precision in the head tracking. I feel like I can control my head well enough to do this if the hardware is capable. How about some feedback? I would rate FaceTrackNoIR a solid 5, and given nothing else would be happy to use it. Is TrackIR v4 a 9? Is v5 is a 10? I'll buy tomorrow >:) Thanks !

Re: TrackIR 5 Performance compared to v4 and FaceTrackNoir

Posted: Fri Aug 24, 2012 9:50 am
by Seth Steiling
It sounds like you're primarily looking for feedback from other customers, but I'll chime in with some information as well. As you might expect, I've also tried FaceTrackNoIR. :)

TrackIR is especially superior to FaceTrackNoIR (or, really, face tracking in general) in two main aspects: precision and latency.

Precision:
TrackIR 5's raw resolution is higher than many webcams, at 640 x 480. Additionally, we analyze each image on the camera, frame by frame, to granularity down to 1/150th of a pixel. Compare this to TrackIR 4 at 1/20th of a pixel and a Wii Remote at 1/8th of a pixel. Between raw resolution and image processing, TIR5 is topping the charts in what we call reporting resolution--the net precision we're delivering.

Additionally, the method we're using to extract data from that resolution is more precise. FaceTrackNoIR uses edge detection, which is a relatively imprecise method of tracking nuanced motions. We track reflective markers or LEDs, which provides a very high contrast object to track--perfect for nuanced control.

Latency/Responsiveness:
With a sample rate of 120 FPS, TrackIR provides data 2x as fast as the typical targeted render rate for competitive gaming (60 FPS). The lower your sample rate on a tracking system, the more likely that the head tracking data will be "old" and thus feel laggy, or out of sync with what your actual motion is.

The sense of latency is magnified by a need for smoothing. Basically, when a system has a lower level of precision in tracking, the result is noise or jitter. To make up for this, you can smooth the data, but this results in delayed delivery (because the smoothing typically requires that a few frames of data be stored up so they can be averaged, creating a more "smooth" or stable experience). The lack of sufficient resolution in web cameras, combined with the difficulty of face tracking via edge detection, creates a low-precision output of data from anything like similar to FaceTrackNoIR.

But with TrackIR (especially 5), the precision is so high that the need for smoothing is super minimized. I frequently use almost none at all. This keeps the data very responsive, reducing lag to essentially unnoticeable levels. I think the difference between in feel compared to FaceTrackNoIR would really please you. :)

Re: TrackIR 5 Performance compared to v4 and FaceTrackNoir

Posted: Fri Aug 24, 2012 9:53 am
by Seth Steiling
Also, my personal recommendation if you do buy a TrackIR--if you're budget constrained, and can't afford both the TrackIR 5 and the TrackClip PRO upgrade, I would order performance as:

TrackIR 5 + TCP > TrackIR 5 >>> TrackIR 4 + TCP > TrackIR 4

In other words, it makes more sense to me to purchase the standard TrackIR 5 than it does the TrackIR 4 + TCP. The TCP won't nearly make up for the improvements from TrackIR 4 to 5.

Hope this helps. :)

Re: TrackIR 5 Performance compared to v4 and FaceTrackNoir

Posted: Fri Aug 24, 2012 1:47 pm
by drliebs
Thanks for sharing. I'm going to jump in, it sounds like TrackIR has the precision I want. I don't know if the ClipPro is going to work for me, I don't use headphones, do you guys have an alternative method for using it? IR is not picky about lighting conditions is it (seems silly asking that but who knows what else is at play)?

My hats off to the TrackNoIR guys, if that was my only choice I would gladly use it. I think they have done a great job, especially considering the technical hurdles with the platform. And it's free...just thought I would tip my hat to them too.

Re: TrackIR 5 Performance compared to v4 and FaceTrackNoir

Posted: Fri Aug 24, 2012 4:38 pm
by Seth Steiling
The standard clip for TrackIR attaches to the brim of a hat or a visor. So you'll be fine without a headset, as long as you wear a hat.

Regarding lighting conditions--the only problem would be sunlight. If you have a strong source of sunlight in the camera's view, the IR from the sun will disrupt the tracking.

- Seth