Yesterday I attended the Dassault Systèmes Developer Conference (Devcon 2008), and had a good time playing with Softkinetic technology of markerless body tracking. They use a special camera that gives you a depth map and from that they can get blobs corresponding to voxels of objects in the field of view of the camera (correct me if I’m wrong Gilles ;). They can also analyze those information and get you human skeletons to work with.

And they have really funny games!

[youtube]http://youtube.com/watch?v=6cnjoVZuCvo[/youtube]

– Stéphanie from our Marketing team taking a sportive break –

– Eric and yours truly playing coop –

The advantage of this technology is that people don’t have to “make the first step” like putting trackers, markers or anything. As soon as they are in front of the screen, they’re in the game! And the interaction is plain natural, no learning curve.

Here’s the human tetris, virtual style :

[youtube]http://youtube.com/watch?v=wUGau2HnKVw[/youtube]

Following the same idea, CamSpace by Cam-Trax technologies is using really basic markers that can simply be everyday objects, like a bottle of water. It seems the marker simply has to be of a uniform color to be initialized. I don’t have much info about the techno, what kind of data they can get from that, but I’d gess you can have the 2d position of the object, an estimation of its depth, and an orientation along the camera optical axis.

[youtube]http://www.youtube.com/watch?v=v0srY37kkMw[/youtube]

Ah I really hope we’ll soon get markerless, wireless, full body, precise, cheap, small, <insert your wishlist here>, lightweight, undestructible, 6dof tracking with haptics…