jit.openni user map generator (silhouette) not showing all the image
I am trying to figure out why I can’t see the full processed image that is coming out of kinect via jit.openni (see attached image; click on it to see it full).
In the left and upper parts of the image you can see that the image is cut. I don’t get why.
For example for one, the imagemap generator output (RGB channels; 2nd output) output the full image. No problem there. And secondly, I don’t get the same problem in the right part and the lower part of the image coming from map generator (4th output).
Any ideas on why and how to fix this?
-- Pasted Max Patch, click to expand. --Copy all of the following text. Then, in Max, select New From Clipboard.----------begin_max5_patcher---------- 749.3oc0WtsbaBCDF9Z7SACW1wkpCfPoWk9bzoiGYiBQd.AUHm31L4cu5.lj zZa.6bvkYLdjPnc2O8i1UOLKHZY8VdaT3WC+dXPvCyBBbcY6HnqcPTEa6pRV qaXQU71VVAOZt+YZ9VsqeEmkGtVniqa3RoXwpZ4Mhh3sUk6FpbSU8FcIW6lH XWuML8paExhEJ9Js2QfjzXv7vzqr26Z.owfvez8NhbmIqWt9yHR1t4+lZoVx p3tm8Mkf0aYuY0+pg6MPTT+TYeoVwucO.hLVp2WExctJx12iylYuMejXRxu2 3d+Ck7.J7ZyuuDd8crxPDZ+7AcX9PQwoyCQ.fCLPGlPfCvmjS.OFubQESqDa ileY.qO0AqTvIPqDu.xKmPzihKz+A3xxkaZZK1HNNLppy4C.GHjhiIDRVB0. kLrkMTmtBmteBcJ5ooiE3Ihkl6Ex756mtHIwE5HrSqfRHVnPxPVcyA.ABLx3 sQwa4RMSKpkOyjoPmRDeEx8GB3s7yElenHIklMUj.omIR.tUAbV1kIRvXzTQ R14QjDpWjPtPEIYSWibl.wuGEN4MThnqKJJ4SurEXlMu7S4iORZFR+1nMLkY STMWsfKYKKcv.bHDIj5nWu37.oc+owcT0FDOYB3KLwW2lursCCfzSHOxRlr3 CtZDassgWmKZ0tEsP39gDYHHAQ9xZw3iiI5YWOx3acdoocuXToP92mmvw.a+ uj4s0aTq14x8kxG9DGx4sZgzsAvKFE8Ei5VQdNW97uZLKN1OjbTDr20+w6Qo Wbdj0V3g7Hz.NTkHuo1rYR6tCbAhwWYtRrm2hX0mTivLsq0SV6UNNPCDGP5D iCbVLAXtHFOOi3ODoKbbsdyhC3.wQ1DCCXRL0FF39isfSftvv15MKLFTnm79 oySFg+.e+bGzXvC3cbafjQswT5I5Q98xYMM2wUscSpyYLYJWWq5yvYTtReSW UAQJ9chciG65goL4+zljeaT9TWaojnYV673r+3ROFAC -----------end_max5_patcher-----------
By the way, the output is not zoomed out or anything like that. It is just cut out where it shouldn’t.
The depth camera doesn’t have the same field of view as the RGB camera. Therefore, you get a narrower field in the depth image. In the XML configuration file, it is possible to align this narrower depth data to the RGB camera using:
I suspect this is in your configuration
If you do not want this alignment, you can remove that line in the XML configuration file. This will not increase the field of view for the depth camera, but it will cause the output of the depthmap to be "more filled in".
argh! forum is filtering out the XML even in the code block.
Look for the AlternativeViewPoint line in the XML file. That is what I am referencing above that appears as strange red Image1 text.
No problem, Dale. I can work that out!
Thank you for your answer and for the object, of course! :)
May I ask how I can make the usermap generator to show detected human shapes even after a few meters?
In the actual configuration, if I get more than ~ 4 meters away from the kinect, my silhouette disappears. I wanted to be able to detect shapes while being even further away.
I am guessing it is something related to how it is set up right now, but I don’t know how/where to change those settings.
NITE and its use of the Kinect hardware is limited to 4m for identifying and tracking humans. There is no solution available today for distances longer than 4m.
Thank you, Dale.
I wanted to ask you something else about some weirdness I am experiencing with jit.openni. While trying to map on the body positioned in the center of the stage, if the person in the projection gets closer from the kinect, the shoulders from the projected mask get lower, showing up on the torso, and if further, they project on the wall behind higher than the shoulders of the person. I understand it is all about perspective and I pretty much have an idea on how to fix it.
What I don’t get is the following. If the person moves left and right (while not moving closer or further from the original position where the mapping has been done), the mapping used to stay on the body. Now it does not anymore.
I thought it was all in my head, but today, while trying to figure out the problem, I started an old patch and the mapping followed the body left and right perfectly. I closed the patch, put the new patch again which I knew it did not work, and it didn’t. And then I put the old patch again and, this time, it did not work either. And it doesn’t anymore since then. I didn’t change anything in the old patch, I didn’t change the setting (kinect or projector positions). The only explanation I have that this is related to the jit.openni object.
Do you maybe know anything about this?
I have no idea what you are describing/experiencing. A clear repro with a patch and series of steps will help us help you.
jit.openni codebase hasn’t changed in a very long time. I don’t think what you are experiencing would be related to it.
Forums > Jitter