|
Robo Realm Problem
Matt Beaver from United States [8 posts] |
12 year
|
I work with my local robotics team and I did everything right for installing the drivers and my kinect sensor worked the first time on windows 7 64 bit but now it just gives me an unidentifiable image. Also notice all the drivers are installed and I also have the kinect sdk removed and I have rebooted my computer at least 7 times now.
 
|
|
|
|
Matt Beaver from United States [8 posts] |
12 year
|
Picture 1 is it working, picture 2 is the drivers, and picture 3 atttached to this post, is the glitch.
|
|
|
|
Anonymous |
12 year
|
Matt,
Couple questions to help narrow this down:
1. Did you update RR inbetween the working and not?
2. Did you recall if anything else was unstalled or installed during that period?
3. Do you see these effects when you switch off the depth image, or switch off the RGB image?
4. Try switching off the depth and increase the RGB size to 1280x1024 and see if this continues.
5. Also check the IR image to see if you get the same distortion.
The error appears to be a timing issue that is out of sync. What is strange is that it is happening on both the depth and RGB image which have different encodings.
Thanks,
STeven.
|
|
|
|
Matt Beaver from United States [8 posts] |
12 year
|
1) No
2) Nothing was installed or uninstalled
3) Yes
4) Tried it same thing
5) Same
Pic 1 is IR and pic 2 is 1280x1024.
1280x1024 gives fine RGB but then it will randomly blink green and the depth view is still distorted.
 
|
|
|
|
Anonymous |
12 year
|
Matt,
Actually the last two images that you posted are actually what we would expect. The IR is bad but that can be solved by exiting RR and then starting it again ... it should pull in the same module and initialize the Kinect. Sometimes switching from RGB to IR causes that pattern that you see. Restarting appears to clear that up (or chaning the IR size will also).
The second image is correct for a 320x240 depth and 640x480 RGB. Are you expecting to see something different? Note that the screen is a bad reflector of IR so the Kinect is not going to pick that up.
In the previous images there is a clear disconnect between certain parts of the image ... which doesn't appear in the second set of images. Or does this error phase in and out of the current display?
STeven.
|
|
|
|
Matt Beaver from United States [8 posts] |
12 year
|
|
|
|
|
Matt Beaver from United States [8 posts] |
12 year
|
I have tried all combinations and this is what I am still getting.
|
|
|
|
Anonymous |
12 year
|
Matt,
We've managed to replicate your issue (phew!). This is going to sound a little strange but try your setup on AC power and then just on battery. We are theorizing that the Kinect requires quite a bit of power in order to function correctly ... which when on battery mode (on some machines) may not be getting enough juice to function through the USB port.
Also, watch your CPU utilization and see if that changes when on AC and on battery when running the Kinect. We notice quite a jump when moving from AC to battery ... so this may also be a CPU power (in terms of number of calculations) which typically is also reduced when on battery mode.
If your experience is similar to what we are seeing then at least we know where the problem exists and have a better idea on how to fix that.
STeven.
|
|
|
|
Anonymous |
12 year
|
Matt,
Ok, I think we have another solution. Try downloading 2.44.5 which has more checks in for image formatting errors. It appears that on slow machines they are not quite able to keep up with the data throughput from the Kinect and will cause sections of the image to go missing. This new version will ignore those incomplete frames ... so you will notice a fps reduction if more errors are being ignored.
For example, a netbook at we saw the issue on was getting 25 fps on AC power with 640x480 on both RGB and Depth. Removing the AC cord reduced the fps to 15!!
While we are not 100% if the issue is CPU speed or actual power distribution to the Kinect I'd recommend wiring an external power supply to the netbook to keep the CPU happen ... or see if you can find battery settings that do not reduce the CPU speed. Either way the newest version should help to clean up your issue quite a bit.
Cheers,
STeven.
|
|
|
|
Matt Beaver from United States [8 posts] |
12 year
|
|
I am running it on a quad 3.6 OC'd to 4.2 GHz processor, with 16 GB's of 1600 MHz RAM, with a 512 GB SSD, and Dual SLI Nvidia 560 Ti's, and watercooling. It's not my machine.
|
|
|
|
Matt Beaver from United States [8 posts] |
12 year
|
|
Either way though, I updated and it worked the first time and now it decides when it wants to work.
|
|
|
|
Anonymous |
12 year
|
Wow, nice machine! I guess this issue doesn't have anything to do with the CPU speed! I assume this is a desktop or laptop?
By your last comment you are now seeing the problem less frequently but it still happens once in a while? Or how would you characterize the error?
While we had duplicated a similar effect and have that fixed on our side we will continue to test on other machines till we can once again see an issue.
Thanks,
STeven.
|
|
|
|
Matt Beaver from United States [8 posts] |
12 year
|
Well it's works waaaay better on my laptop for some reason -_-
|
|
This Thread
