Corrupted RGB values when using two kinects

classic Classic list List threaded Threaded
30 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Corrupted RGB values when using two kinects

Michael Vitus
Hello,

    I am running into a little weirdness when using two kinects.  It appears that the RGB values are being corrupted by the other kinect stream, but the xyz values don't seem to be affected.  This only occurs intermittently as well.

    For example, here are the correct images for the two kinects: kinect 1 kinect 2 and here is the corrupted image for the kinect 1.  As shown, the lower half has the RGB values from the kinect 2.

    A minimal code example is here: http://pastebin.com/gdRGXrk5

    I am using pcl-1.5 on Ubuntu x86_64 Natty. 

cheers,
Mike


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Radu B. Rusu
Administrator
Michael,

Thanks - I am able to replicate this. We'll try to find a fix tomorrow.

Cheers,
Radu.
--
http://pointclouds.org

On 02/25/2012 11:32 PM, Michael Vitus wrote:

> Hello,
>
>     I am running into a little weirdness when using two kinects.  It appears that the RGB values are being corrupted by
> the other kinect stream, but the xyz values don't seem to be affected.  This only occurs intermittently as well.
>
>     For example, here are the correct images for the two kinects: kinect 1 <http://postimage.org/image/57isfz1kh/>
> kinect 2 <http://postimage.org/image/kwecsxayd/> and here is the corrupted image <http://postimage.org/image/jh6t4fstt/>
> for the kinect 1.  As shown, the lower half has the RGB values from the kinect 2.
>
>     A minimal code example is here: http://pastebin.com/gdRGXrk5
>
>     I am using pcl-1.5 on Ubuntu x86_64 Natty.
>
> cheers,
> Mike
>
>
>
> _______________________________________________
> [hidden email] / http://pointclouds.org
> http://pointclouds.org/mailman/listinfo/pcl-users
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Radu B. Rusu
Administrator
In reply to this post by Michael Vitus
FYI, this was fixed by moving to a newer version of OpenNI - the bug was on their side. See
http://dev.pointclouds.org/issues/627 for more information.

Cheers,
Radu.
--
http://pointclouds.org

On 02/25/2012 11:32 PM, Michael Vitus wrote:

> Hello,
>
>     I am running into a little weirdness when using two kinects.  It appears that the RGB values are being corrupted by
> the other kinect stream, but the xyz values don't seem to be affected.  This only occurs intermittently as well.
>
>     For example, here are the correct images for the two kinects: kinect 1 <http://postimage.org/image/57isfz1kh/>
> kinect 2 <http://postimage.org/image/kwecsxayd/> and here is the corrupted image <http://postimage.org/image/jh6t4fstt/>
> for the kinect 1.  As shown, the lower half has the RGB values from the kinect 2.
>
>     A minimal code example is here: http://pastebin.com/gdRGXrk5
>
>     I am using pcl-1.5 on Ubuntu x86_64 Natty.
>
> cheers,
> Mike
>
>
>
> _______________________________________________
> [hidden email] / http://pointclouds.org
> http://pointclouds.org/mailman/listinfo/pcl-users
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Michael Vitus
Hello Radu,

   Great!  Thank you for looking into this.  

cheers,
Mike

On Mon, Mar 12, 2012 at 2:52 PM, Radu B. Rusu <[hidden email]> wrote:
FYI, this was fixed by moving to a newer version of OpenNI - the bug was on their side. See
http://dev.pointclouds.org/issues/627 for more information.

Cheers,
Radu.
--
http://pointclouds.org

On 02/25/2012 11:32 PM, Michael Vitus wrote:
> Hello,
>
>     I am running into a little weirdness when using two kinects.  It appears that the RGB values are being corrupted by
> the other kinect stream, but the xyz values don't seem to be affected.  This only occurs intermittently as well.
>
>     For example, here are the correct images for the two kinects: kinect 1 <http://postimage.org/image/57isfz1kh/>
> kinect 2 <http://postimage.org/image/kwecsxayd/> and here is the corrupted image <http://postimage.org/image/jh6t4fstt/>
> for the kinect 1.  As shown, the lower half has the RGB values from the kinect 2.
>
>     A minimal code example is here: http://pastebin.com/gdRGXrk5
>
>     I am using pcl-1.5 on Ubuntu x86_64 Natty.
>
> cheers,
> Mike
>
>
>
> _______________________________________________


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Radu B. Rusu
Administrator
It's really all Suat :) I was just the guinea pig testing it as his laptop had some issues :)

Cheers,
Radu.
--
http://pointclouds.org

On 03/12/2012 03:29 PM, Michael Vitus wrote:

> Hello Radu,
>
>    Great!  Thank you for looking into this.  
>
> cheers,
> Mike
>
> On Mon, Mar 12, 2012 at 2:52 PM, Radu B. Rusu <[hidden email] <mailto:[hidden email]>> wrote:
>
>     FYI, this was fixed by moving to a newer version of OpenNI - the bug was on their side. See
>     http://dev.pointclouds.org/issues/627 for more information.
>
>     Cheers,
>     Radu.
>     --
>     http://pointclouds.org
>
>     On 02/25/2012 11:32 PM, Michael Vitus wrote:
>     > Hello,
>     >
>     >     I am running into a little weirdness when using two kinects.  It appears that the RGB values are being
>     corrupted by
>     > the other kinect stream, but the xyz values don't seem to be affected.  This only occurs intermittently as well.
>     >
>     >     For example, here are the correct images for the two kinects: kinect 1 <http://postimage.org/image/57isfz1kh/>
>     > kinect 2 <http://postimage.org/image/kwecsxayd/> and here is the corrupted image
>     <http://postimage.org/image/jh6t4fstt/>
>     > for the kinect 1.  As shown, the lower half has the RGB values from the kinect 2.
>     >
>     >     A minimal code example is here: http://pastebin.com/gdRGXrk5
>     >
>     >     I am using pcl-1.5 on Ubuntu x86_64 Natty.
>     >
>     > cheers,
>     > Mike
>     >
>     >
>     >
>     > _______________________________________________
>     > [hidden email] <mailto:[hidden email]> / http://pointclouds.org
>     > http://pointclouds.org/mailman/listinfo/pcl-users
>
>
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Michael Vitus-2
Thank you Suat :-).

--Mike

On Mon, Mar 12, 2012 at 3:30 PM, Radu B. Rusu <[hidden email]> wrote:
It's really all Suat :) I was just the guinea pig testing it as his laptop had some issues :)

Cheers,
Radu.
--
http://pointclouds.org

On 03/12/2012 03:29 PM, Michael Vitus wrote:
> Hello Radu,
>
>    Great!  Thank you for looking into this.
>
> cheers,
> Mike
>
> On Mon, Mar 12, 2012 at 2:52 PM, Radu B. Rusu <[hidden email] <mailto:[hidden email]>> wrote:
>
>     FYI, this was fixed by moving to a newer version of OpenNI - the bug was on their side. See
>     http://dev.pointclouds.org/issues/627 for more information.
>
>     Cheers,
>     Radu.
>     --
>     http://pointclouds.org
>
>     On 02/25/2012 11:32 PM, Michael Vitus wrote:
>     > Hello,
>     >
>     >     I am running into a little weirdness when using two kinects.  It appears that the RGB values are being
>     corrupted by
>     > the other kinect stream, but the xyz values don't seem to be affected.  This only occurs intermittently as well.
>     >
>     >     For example, here are the correct images for the two kinects: kinect 1 <http://postimage.org/image/57isfz1kh/>
>     > kinect 2 <http://postimage.org/image/kwecsxayd/> and here is the corrupted image
>     <http://postimage.org/image/jh6t4fstt/>
>     > for the kinect 1.  As shown, the lower half has the RGB values from the kinect 2.
>     >
>     >     A minimal code example is here: http://pastebin.com/gdRGXrk5
>     >
>     >     I am using pcl-1.5 on Ubuntu x86_64 Natty.
>     >
>     > cheers,
>     > Mike
>     >
>     >
>     >
>     > _______________________________________________
>     > [hidden email] <mailto:[hidden email]> / http://pointclouds.org
>     > http://pointclouds.org/mailman/listinfo/pcl-users
>
>


_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

felix
Hi,

I'm running into the same issue, because I'm using the precompiled PCL version (1.5.1 on Win7, 64bit).

Using a newer *unpatched* OpenNI version (1.5.2.23-dev) did not work. And since http://dev.pointclouds.org/issues/627 states that executable recreation is pending, I guess I need to compile the patched OpenNI (and Sensor?)?
I see that "http://svn.pointclouds.org/pcl/trunk/3rdparty/openni/" might contain some related source, but I'm not sure if this is the place to look for the patch or the patched OpenNI version.
So the question is how can I get the new and patched OpenNI version? Where to look for the code?

Thanks for your help. Felix
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Radu B Rusu
Administrator
Suat found the bug to be in OpenNI not PCL. However while updating to the newest OpenNI fixed the corruption problem, it
also brought a lot of XnSensorServer instabilities. We e-mailed our colleagues at PrimeSense about this, but didn't hear
about a fix yet.

Cheers,
Radu.

On 03/30/2012 06:03 AM, felix wrote:

> Hi,
>
> I'm running into the same issue, because I'm using the precompiled PCL
> version (1.5.1 on Win7, 64bit).
>
> Using a newer *unpatched* OpenNI version (1.5.2.23-dev) did not work. And
> since http://dev.pointclouds.org/issues/627 states that executable
> recreation is pending, I guess I need to compile the patched OpenNI (and
> Sensor?)?
> I see that "http://svn.pointclouds.org/pcl/trunk/3rdparty/openni/" might
> contain some related source, but I'm not sure if this is the place to look
> for the patch or the patched OpenNI version.
> So the question is how can I get the new and patched OpenNI version? Where
> to look for the code?
>
> Thanks for your help. Felix
>
> --
> View this message in context: http://www.pcl-users.org/Corrupted-RGB-values-when-using-two-kinects-tp3776530p3870741.html
> Sent from the Point Cloud Library (PCL) Users mailing list archive at Nabble.com.
> _______________________________________________
> [hidden email] / http://pointclouds.org
> http://pointclouds.org/mailman/listinfo/pcl-users
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

felix
Ok, thanks for the infos.

Felix
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

teixeira0
This post was updated on .
Hello,

Some friends and I also tried to use this code : http://pastebin.com/gdRGXrk5

However, we got different results. Every time we run this program, we are able to see only one image, from the first kinect. The data from the second one seems to disappear in the process, because the window that was supposed to show kinect 2 sowed only a black screen.
We tried to use openni_viewer in two different terminals using the parameter -dev with different IDs. With this, both windows showed the first kinect's data.

Please, if you have any information about this issue, tell me.

We used in 2 different computers (Ubuntu 10.10 and 11.04) both with PCL 1.5.1.

Thanks for your help,

teixeira0
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

gneilhaven
In reply to this post by Radu B Rusu
Sorry to be a pest, but since my application is still exhibiting the rgb image corruption I'm trying to understand the current state of things. Irrespective of its effect on pcl:

1. Precisely which version of OpenNI fixes the ring buffer corruption? As felix mentioned, using version 1.5.2.23-dev from the OpenNI project in github does not seem to fix the problem. (At least using Windows 7, 64-bit.)

2. Using *patched* version 1.3.2 from the pcl.org website also does not seem to fix the problem. Could you supply a pointer to the code patch you are using to fix the ring buffer corruption? Is it implemented within the patched version 1.3.2 from pcl.org?

3. When you mention "OpenNI" I am assuming you are referring to the OpenNI layer of the drivers, and that it is not necessary to update the PrimeSense layer, but perhaps I am confused. Could you please confirm or correct my understanding?

Radu B. Rusu-2 wrote
Suat found the bug to be in OpenNI not PCL. However ... Cheers, Radu.
felix wrote
On 03/30/2012 06:03 AM, felix wrote: > Using a newer *unpatched* OpenNI version (1.5.2.23-dev) did not work.
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

filipecaixeta
Someone who was able to make two Kinects work could say how you did it and what version of OpenNI are using?
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Radu B Rusu
Administrator
I don't think this got fixed yet, as the errors are in OpenNI, not in PCL as far as I know.

Cheers,
Radu.

On 04/22/2012 10:26 AM, filipecaixeta wrote:
> Someone who was able to make two Kinects work could say how you did it and
> what version of OpenNI are using?
>
> --
> View this message in context: http://www.pcl-users.org/Corrupted-RGB-values-when-using-two-kinects-tp3776530p3930327.html
> Sent from the Point Cloud Library (PCL) Users mailing list archive at Nabble.com.
> _______________________________________________
> [hidden email] / http://pointclouds.org
> http://pointclouds.org/mailman/listinfo/pcl-users
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

goutham_mr
Hi,

I am also facing with the same RGB corruption problem using multiple kinects. I am confused reading through the mailing list whether upgrading to openNI 1.5.2 fixes the corruption problem or not. Either way, I would like to try a patched version of openNI 1.5.2 in windows. However, I am unable to find either precompiled binaries or source files for compilation in windows. Can someone point me to that?

Thanks,

Goutham
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

xnoybis
Is there any further information on how to resolve this? I know this topic is 6 months old, but i'm still having the same issues using OpenNI 1.5.4 and latest unstable PCL. Can someone recommend a combination of OpenNI/PCL revisions that will work correctly with 2 kinects?
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Jochen Sprickerhof
Administrator
Hi xnoybis,

* xnoybis <[hidden email]> [2012-11-16 07:59]:
> Is there any further information on how to resolve this? I know this topic is
> 6 months old, but i'm still having the same issues using OpenNI 1.5.4 and
> latest unstable PCL. Can someone recommend a combination of OpenNI/PCL
> revisions that will work correctly with 2 kinects?

Please try the new versions from our website (1.5.4.0/5.1.0.41), they
are tested to work with more than one Kinect.

Cheers Jochen
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

xnoybis
Hi Jochen,

Thanks for the reply. I was pretty sure that I was using all the most up-to-date dependencies, but I
re-downloaded and rebuilt everything just to be sure. Unfortunately the problem is persisting - i've uploaded
a brief video of it in action here: http://youtu.be/b5dBTUX5zSs . Any suggestions on how else I might be
able to diagnose the cause of this?

Failing that, does anyone have an example code snippet that is verified to work with 2 kinects simultaneously?
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Radu B. Rusu
Administrator
Just to confirm, can you show us the output of:

$ dpkg -l | grep openni

and

$ dpkg -l | grep ps-engine

on your machine?

Cheers,
Radu.
--
http://openperception.org

On 11/19/2012 07:43 AM, xnoybis wrote:

> Hi Jochen,
>
> Thanks for the reply. I was pretty sure that I was using all the most
> up-to-date dependencies, but I
> re-downloaded and rebuilt everything just to be sure. Unfortunately the
> problem is persisting - i've uploaded
> a brief video of it in action here: http://youtu.be/b5dBTUX5zSs . Any
> suggestions on how else I might be
> able to diagnose the cause of this?
>
> Failing that, does anyone have an example code snippet that is verified to
> work with 2 kinects simultaneously?
>
>
>
> --
> View this message in context: http://www.pcl-users.org/Corrupted-RGB-values-when-using-two-kinects-tp3776530p4023932.html
> Sent from the Point Cloud Library (PCL) Users mailing list mailing list archive at Nabble.com.
> _______________________________________________
> [hidden email] / http://pointclouds.org
> http://pointclouds.org/mailman/listinfo/pcl-users
>
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

xnoybis
Hi Radu,

Unfortunately this is being developed on Win7/MSVC2010. Is there some equivalent data on the
windows installation I can give you that would be of help?

Thanks,
Reply | Threaded
Open this post in threaded view
|

Re: Corrupted RGB values when using two kinects

Radu B. Rusu
Administrator
We will try to roll out new OpenNI binaries for Windows soon, maybe even today.

Cheers,
Radu.

On Nov 19, 2012, at 8:44, xnoybis <[hidden email]> wrote:

> Hi Radu,
>
> Unfortunately this is being developed on Win7/MSVC2010. Is there some
> equivalent data on the
> windows installation I can give you that would be of help?
>
> Thanks,
>
>
>
> --
> View this message in context: http://www.pcl-users.org/Corrupted-RGB-values-when-using-two-kinects-tp3776530p4023940.html
> Sent from the Point Cloud Library (PCL) Users mailing list mailing list archive at Nabble.com.
> _______________________________________________
> [hidden email] / http://pointclouds.org
> http://pointclouds.org/mailman/listinfo/pcl-users
_______________________________________________
[hidden email] / http://pointclouds.org
http://pointclouds.org/mailman/listinfo/pcl-users
12