Kinect2 - Projector Calibration

Hey,

here is a little example showing how you would use the Kinect Calibration Component as a Camera.

Best
Markus
kinectRightHandExample.toe (21.9 KB)

thnx for the new version. do I need to use the latest td version for that always or does it work for older versions also?
knut

Hi Knut,

you will have to use 2018.27550 (latest official).

cheers
Markus

Hi Snaut,

Thank you for this tool.

this was my first post but i found what i did wrong:

My data points where to close together, now i went all over the place to get points and it worked.

Kind regards
Groeten
Gertjan

Thanks for this! It works fine with calibration and everything! Really good job!

But I experience a bit of delay. I guess that is hard to avoid though … But is there a way to at least to minimize the delay? At the moment if I am drawing just lines between all the joints creating a stick man and projecting it back onto my body the delay is pretty noticable as long as you dont move really slow.

thanks

It worked perfectly fine at home but when Im at the theatre where I shall run a show Im constantly getting error message. “error to high, try again” when I am trying to calibrate.

Anyone who knows what that means?

Hey Acroscene,

I would first check the lighting conditions and if the camera sees the checkboard ok. I also find it useful to project the checkboard onto a black surface rather than a white one. You can play around with the Grid Level Parameter to see if you get better results.
Also projecting onto reflective material can be difficult.

Essentially the message tells you that the reprojection error is too high meaning in essence that the algorithm figures the combined errors are so high that it can’t return a satisfactory solution for the camera position.

Regarding the delay, this is how TouchDesigner receives the data from the Kinect. The delay can be reduced by turning off Joint Smoothing in the Kinect CHOP though…

Best
Markus

Hi Markus
I would like to project the mask, that I derive from the kincet player index on the moving player.
Is there a way, that I can use the camera data out of this calibration for that purpose?
greetings knut

Hi Knut,

that should be possible. Your Kinect is assumed to be positioned at root (0,0,0) looking straight down the z axis. Trick would be to position a rectangle now in front of the kinect (straight down the z axis, no other translation or rotation) that holds the video texture from the kinect. A good question is how far out it would have to be - Needs to be far enough to be seen by the TouchDesigner Camera… Maybe you would have to play around with distance and size a bit to get it right…

Best
Markus

Hi Markus
OK i understand. I will try that.
Thnx for your help
Knut

Does this work for Player Index? If so , any pointers pls

Thanks

Justin

I tried to follow the advice from markus with no success.
Need a little Bit of Time to understand where the problem is and to come back with questions if necessary.
Knut

I was trying to get this to work with the Intel Realsense for awhile.

I couldn’t get any reliable results. I consider two things may be an issue.
It may be that the camera FOV is different for the Kinect vs Realsense so I tried the different OpenCV toggles to guess focal length and aspect, didn’t help.

The next guess was that the realsense rgb camera doesn’t have the same FOV as the 3d depth image. I would guess if the point pairs are compared, it would be sampling an incorrect position.
I tried the “depth aligned to color” settings, but it seems like the pointcloud top doesn’t have this option. The “color aglined to depth” doesn’t find any checkerboards.

Any more suggestions about what to try to get the realsense to calibrate?

I was also guessing that the depth data would have different values considering the realsense can be used closer proximity. But that shouldn’t matter the calibration shouldn’t need actual distances, should it?

Sometimes It appeared that the depth data was not discovered on the white board I was holding. Would that record 3d position of -1 or discard the point pair?

I dont have no realsense so I cant comment on that.

But some learnings from using the kinect a lot

  1. 7 to 10 point pairs a normally sufficient with a kinect 2.
  2. Lots of problems if I try to get data when using a short throw projector. If I include data
    from checkerboards that are closer to the edges of the projection area, the data becomes unreliable - the projected image does not hit the person any more.
    Never saw that on projectors with a “normal” lens. Seems to be a problem of the projectors lens
    although I would expect that the calibration would cover that…
  3. still problems to use the kinect player index to project a mask on the person. I cant get it
    to work in my enviroment

good luck

I have got problems using the component as a camera. Anyone successfully calibrated the camera object to be used and calibrated with the rgb-camera from the kinect or maybe that is not possible?

A workaround would be to use the rgb camera as a colormap and instance it on the pointcloud? Maybe that would be the only solution or are there an more straightforward approach?

Sorry, not sure I quite understand. Does it correctly calibrate the pointcloud for you?

Cheers
Markus

Yes. The pointcloud calibrates fine. I wanted to calibrate my camera, only using the rgb (ordinary webcam), not the pointcloud. But guess that is not what this method does. :slight_smile:

Hi @snaut thanks for that! Finally played with it for a second, here’s a junky proof of concept of interactive projection mapping :wink:

1 Like

Snaut (Markus)

Thanks so much for this tox. Super effective and clever. I love using the projector for the sizing the grid on a big white board that I move around the room. So much cooler than using multiple sizes of cardboard with checkerboards printed on it.

What I would like to do is add a high def camera input that will also get calibrated to the Kinect + projector combo. By using the projector as the grid source, I don’t have to use the IR output of the kinect and clever real-world lighting to show the checkboard like older methods of aligning a Kinect with a different RGB camera.

I think it would be a super clever and useful object that perhaps others would use.

It would allow calibrating a Kinect with a Hi-Def camera (and projector) without the nasty process of using the IR stream to try and read a checkerboard.

Win Win Win

I’ve spent some time going through the KinectCalibration.tox and have a basic understanding of what’s going on in there. My python and OpenCV skills are where I’m coming up short. I think I understand most of the OpenCV script, but don’t know how to add the other camera and associated code to get it’s transform matrix and extrinics and intrinsics and tie the depth data to that camera as well without breaking what 's already there.

Is this a feasible? Overly complicated ? Tips ?

Cheers

Stuart

Hi Snaut,

thank you for sharing this Kinect Calibration Tool.
I’ve been trying to get it to successfully calibrate on and off, but i am struggling.

The kinectCalibration version that is shipped with TouchDesigner (2020.20020 & 2019.33840) seem to have some bugs in the python code?

the lines 40-41 & 87-88 seem to have to be updated to

gridResX = int(parent().par.Gridresw)
gridResY = int(parent().par.Gridresh)

for it to capture points.

It calculates intrinsic & extrinsic matrices but the intrinsic matrix seems to have weird negative values ?
the Kinect2 is just placed behind the projector.

image

Do you have suggestions what would be the best way to debug this ?

Thanks for your help!
Seb