very cheap hand tracking with color glove

Or just use the free OpenNI 3D Hand Tracking Library that does not need any glove:
http://www.openni.org/files/3d-hand-tracking-library




Although the video above looks like he is wearing a glove, that is a VIRTUAL glove. His hand is bare.

And here is the "zigfu development kit for Unity3d", which works with OpenNI:
http://www.openni.org/files/zdk-for-unity3d

Why does the forum editor have a "YouTube" tag, and then just ignore it (as can be seen in post such as this one).

I should try openni, but i think that hand tracking accuracy will be better with color glove.

"ulula" wrote:
I should try openni, but i think that hand tracking accuracy will be better with color glove.

I agree ulula. The colors give useful hints to help tracking. Another thing to keep in mind with any camera-based tracking, is that with the Rift on your face, you can have bright lighting -- all the better for a camera to get a better exposure with less noise. Not sunlight-bright, mind you -- unless the camera you use is fine with such intensity.

That is cool, ulula. The good thing that this colored glove method has over bare-hand tracking methods appears to be a lack of confusion for the camera when it comes to occlusion and fingers overlapping each other. Even the leap has been shown to suffer from this issue, so for camera-based, tracking methods, it seems like a jump in the right direction. I'm not so sure about the camera being used to track this being on the rift itself though, as I don't expect many games to have the user's hands in their sight at all times.

"ulula" wrote:


Oculus team, you should hire this guy or buy is tech.
ok, these gloves are ugly but with 2 mounted mini camera on oculus rift this can be super effective.

Data gloves, just like in the OASIS.

Throw some tiny vibration motors in there for basic haptics, and we are good to go. The Hydra is too kludgy to last long.

"rmcclelland" wrote:

Throw some tiny vibration motors in there for basic haptics, and we are good to go. The Hydra is too kludgy to last long.

No need for motors. Various gloves of the '90s used piezoelectrics (http://en.wikipedia.org/wiki/Piezoelectricity) at the fingertips for touch.

Bad news everyone, looks like it is not possible to obtain the source code of this project

ulula: So if you are no longer developing the color-glove-based system may be you can give us the code source of this project ?

Robert Wang: I'm sorry, but we can't do that at this time. The code for that project is still proprietary (and technically the property of MIT).
Our current system however does provide an easy to use and open API with many code examples.
http://threegear.com/latest/doc/api.html

Best,
Rob

Wouldn't the latency of standard camera based system outweigh the benefits when used with something like the oculus?

If you're looking at this sort of low latency tracking, something like the Leap would be more fitting, although range and drift are issues to consider.

From what I've been hearing, latency concerns are primarily tied to head movements.

The tracking of head motion has to have as little latency as possible, because your brain is very sensitive to shifts in what you can see not matching your actual movements.

But as I've heard so far, this is far less critical for tracking other parts of the body. (assuming of course, that these body parts that you're tracking don't somehow figure into the view transform.)

Basically, people are very sensitive to the camera movement not matching what they expect, but not quite so sensitive to their hands and feet and so on not quite being where they think they are.

(When you think about it, that kind of makes sense, since your head movement can influence the balance sensors in your inner ear, while body position is actually largely deduced through the equivalent of correlating what you see with the angular measurements of your joints.)

This looks very cool.

Also, regarding the YouTube tags. You have to use just the video code (the part after the v=), not the whole URL.