Microsoft Azure Kinect DK

Will it be supported in versions of TD following the new Kinects release this summer?
Thank you, Graham

Details are still thin on the device I think, but yes we plan to support it if it makes sense.


any news on the possibility of support for Azure DK ?


well the hardware, drivers and SDK’s are not even released yet as you can see here:
Microsofts official statement is only: “coming soon”, so I can imagine any third parties can’t guarantee anything about any future integration until that time.

Just got mine in the mail today. Curious if anyone has anything working yet.

I received mine today. Would be awesome to get this new sensor integrated !

Getting 4 devices soon,
hopefully will able to test the multi camera case with TD.

Does anyone know if Body Tracking SDK is going to support input from multiple Azure Kinect devices? I guess it should, but I have not found it officially stated on their website so far.

it’s starting to heat up! nice… >D

I just received mine as well! Very much hoping to get this working in TD for some upcoming installations this fall - would love any feedback on integration progress!

We expect to be getting ours at the start of Aug and will try to get it integrated ASAP

If you guys will able to integrate the full multi sensor features,
it will probably place TD as the best environment for volumetric content generation.

I have few of the K4A around, cant wait to build full volumetric capture system.
:smiley: :smiley: :smiley:

Excelent news!

Besides the nesting / syncing of n sensors, what do you think it will be better with this Kinect?

I have a couple in the office rn. On first blush they are faster, more accurate and overall push a much cleaner point cloud. Build quality is much better and they have a cool microphone matrix on the top.

They have a lot of work to do on the data \Ai\ skeleton stream part.
skeleton ( which run differently from from v2 skeleton)
suffer from high latency issues and as far as i checked not all features imported
from the old v2 kinect. ( not sure about that )

depth wise
The" TOF noise" and kinect v2 issues got better over all.

one device default stream + record on ssd
runs on rtx 2080ti, 8700k machine
cost about 5% of cpu and 13% of gpu , what ever that means.

seem like there is lot of room for additional streams ( cameras ) which is great.

overall it seem they started from scratch, kind of one step backward…
hopefully to gain many steps forward later. … SDK/issues

my modest impression so far.

There is also an IMU in the sensor - gyroscope and accelerometer, this will be helpful if the sensor is attached to a moving platform.

I’m working on a CPlusPlus TOP with a friend that will stream the depth image in from this new Kinect. It’s currently in a working state - we’ll post it once we have a chance to make it a bit more user friendly!

@b34npour that is awesome. Good work ! Is this on GitHub ? I just finished implementing a wrapper in openFrameworks and would love to fork your c++ project

@wooliii it’s not up on GitHub yet - but will post soon. Just cleaning it up so that it’s easier to set up (currently you have to manually copy Kinect SDK DLLs into your touchdesigner bin folder, and building the project in VS requires some manual include directory adjustments)

Any more updates on the Kinect DK? More just curious to get other first impressions with it. I am working out a rig that is for live facial projection mapping on Kinect 2 and am curious to see if this provides significantly higher framerates /smoother tracking.