ofxOpenNi with automatic skeleton recognition/ no pose required

I’ve just added to roxlu’s ofxOpenNi the capacity for automatically loading a standard user calibration file whenever a new user is recogniced.
I included an example app.
So far compiled OpenNI an NITE libs just for mac.
In win or linux it should work with no problem using the latest unstable releases of both libraries.
comments are welcome.

Check at:
https://github.com/roymacdonald/ofxOpenNI

cheers!!

Hi Roy,

Thanks fro this. I’m keen to try this out but seems some pieces may be missing? I notice the auto skeleton example still references ofxHardwareDriver but this class no longer exists in your version of the addon? I removed this dependency from the example but then I came up against some other problems.

Any advice?

Hello Roy!! Congratulations!! and thank for sharing!

With Guido Corallo want to contact you to start making together spanish tutorials for oF… see you!!

This sounds awesome - I have been hoping this was possible without the MS SDK.

In addition to nay’s comments I see that it looks like you modified ofxTracker user as well to take in a context

  
ofxTrackedUser* tracked_user = new ofxTrackedUser(context);  

but your version on github doesn’t reflect this.

https://github.com/roymacdonald/ofxOpenNI/blob/experimental/src/ofxTrackedUser.h

Let me know if I can help!

HI there.
This is my first github commit so there might be some faults.
I just downloaded it in a different computer and realized there are several things to correct.
Mi first guess is that the reference to the openframeworks lib is broken.

I didn’t modified the ofxTrackedUser class. just ofxUserGenerator.

I’ll correct right now all the mistakes and upload it asap.
I’ll let you know when its done.

OK. so now it works.
My mistake was to use roxlu’s latest commit, in which he modified and reimplemented almost all classes, but I was actually using a previous version by gameoverhack. So now i reverted all to gameoverhack’s version and uploaded everyting to github.
Now it should run with no problems.
Keep in mind that you need to update to the latest versions of OpenNI and NITE.

cheers!

hello roy!

can you upload your usercalibration.bin file?:slight_smile:

cheers
andré

Duhh!
Sure.
My fault.
It wasn’t uploaded because I left the whole bin folder ignored.
I’m uploading it to GitHub right now.

BTW, I gues that my Usercalibration.bin file might not be the best one. You can easily create your own ones. The classes for doing so are implemented in ofxUserGenerator. I think I might have a working example that generates UserCalibration files. I’ll upload it asap to github.

cheers!

anybody else getting a crash and this in the console?

  
dyld: Library not loaded: ../../../Bin/Release/libnimCodecs.dylib  

still testing but it seems to works pretty well…nice work roy :slight_smile:

i have it working in visual studio in case you want it…

cheers
andré

I patched together a version using a recent project based off gameoverhack’s and roy’s ofxUserGenerator

It seems the updated libraries were throwing me. Still testing too but once I created my own usercalibration file it picked me up right away - pretty amazing!

@jvcleave: the error you are getting is because you have not installed the latest OpenNI and NITE libs.
Even though the updated libs are in data/openni/libs/ folder you need to install the libs to your computer.
I don’t know why this is this way, I was getting the same error and got solved after installing the lastest libs.
I left my comuuter uploading the usercalibration file, I didn’t checked if it was uploaded well; it didn’t.
I’ll uploadit again to github. Also attached here.

Cheers.

UserCalibration.bin.zip

@andremaia:
If you can send me the Visualstudio project files I’d be glad to add it to the repository.

cheers!

thanks so much roy - no need for the M$ API now!
after updating openni and nite the latest git works out of the box
cheers!

@roymacdonald

here it is the VS project.

cheers

VS-project.zip

@andremaia
Great!
many thanks.
I’m adding the VS project to the github repo.

Best!