Kinect HD
Hi,
I've worked with first generation Kinect quite a bit and was planning to transition one of my projects to the Kinect HD. I was wondering if anybody has worked on any projects with new Kinect and if there are any significant differences that I should be aware of. I've tried to research this online by myself and have found only Kinect 360 projects.
Any input is appreciated.
Thank you in advance.
Lethon
Hi. I have a lot of experience with the v1 and v2 Kinects. My externals dp.kinect and dp.kinect2 have a very high level of compatibility and sometimes you only have to add a "2" to the name to make your patches work. http://hidale.com/shop/dp-kinect2/
The spec differences between the sensors itself are best described by googling. Here's one post: https://social.msdn.microsoft.com/Forums/en-US/d158baa3-1779-439d-a9f0-c31053581dbc/what-is-the-difference-between-kinect-for-windows-and-v2-sensor?forum=kinectsdk
Finally, there are hardware and OS requirements that are different. Here are Microsoft's requirements for the v2 sensor https://developer.microsoft.com/en-us/windows/kinect/hardware-setup
Perhaps the most fundamental difference in the technology, is that any projects using multiple Kinects at once, may no longer be compatible with v2 technology (which is why "HD" may not be a good signifier for Kinect 2, they are using totally different underlying hardware implementations).
Kinect 2 will only support one device being used at a time, since otherwise they would interfere with one another.
It is true that the current drivers from Microsoft only support one Kinectv2 per computer. It is directly related to the tremendous USB3 bandwidth (and PCIe Gen2 slot) needed for the USB controller. Microsoft tried, but realized that at the time of release, the hardware on the marketplace couldn't handle two sensors. So they removed the codebase for multiple sensors on a single computer. The marketplace is better today, yet that legacy decision remains.
It is possible to use two computers. Microsoft did put in specific handling to allow for two Kinect v2's in the same workspace. The specific results you get are like always...specific to your scenario. Usually it is less an issue because the Kinects are not side-by-side and are instead pointed in different areas, regions, angles, etc.
I have specific features on dp.kinect2 to allows for repositioning all the data. With them, you can choose your über-world 0,0,0 point and then use the features to reorient all the data into a single coordinate space. One important feature that allows this is dp.kinect2's ability to work with jit.anim.node. You can create the hierarchies, offsets, rotations, etc. There is an example of this in the included helpfile.