I am currently running debian jessie with kernel 3.14.17 . I assume there is a kernel bug with the camera so I will just make a new kernel. guvcview acts as if it works, just showing a screen screen, until I shine a light in front of it, and then I see a pattern across the image where there should have been a single light source.
Now I am using kernel 3.17.2 What is weird is that my system refuses to boot with the camera plugged in. The camera doesn't want to be configured until I replug it. I think the camera got in a weird state and needs a real power cycle. Anyway: with the new kernel, the resolution seems always 320x240, but it still crashes, and I found these in the dmesg: [ 227.453126] CameraThread[4772]: segfault at 7ff8a03f4ec0 ip 000000000049e2b9 sp 00007ff8a03d1820 error 4 in oculusd[400000+11e000] [ 337.807704] CameraThread[5394]: segfault at 7facdcef9ec0 ip 000000000049e2b9 sp 00007facdced6820 error 4 in oculusd[400000+11e000] [ 435.141965] CameraThread[5820]: segfault at 7f5ff0eeb000 ip 00007f5fef68f767 sp 00007f5ff0f0aaa8 error 4 in libc-2.19.so[7f5fef559000+19f000]
oh, forget it, after the first few crashes the resolution changes too.
I had exactly this problem too, and finally found a solution in my case:
The camera needs to be attached by a very short USB cable.
The long story: oculusd kept crashing for me all time. So I checked the camera with vlc player, and it seem to work. However, plugging the camera into my laptop reveals far higher resolution and framerate. So I tried to get the same experience on my desktop pc. After getting some weired USB related error messages on dmesg, I tried a short cable, and voila, I got the same stability. My laptop is a lenovo thinkpad, which provides superior USB quality (stability and USB power supply), so some other devices working well there had glitches on my desktop pc.
The camera may have timing issues or a too large power demand to run stable on mediocre USB connections.
"neuefarben" wrote: The long story: oculusd kept crashing for me all time. So I checked the camera with vlc player, and it seem to work. However, plugging the camera into my laptop reveals far higher resolution and framerate. So I tried to get the same experience on my desktop pc. After getting some weired USB related error messages on dmesg, I tried a short cable, and voila, I got the same stability. My laptop is a lenovo thinkpad, which provides superior USB quality (stability and USB power supply), so some other devices working well there had glitches on my desktop pc.
The camera may have timing issues or a too large power demand to run stable on mediocre USB connections.
Yes, I have used a different cable, and it works now. Guvcview now shows me my environment again... I can see my glasses, and I can see where the sun hits the walls.
Oculusteam: this problem needs to be adressed on consumer hardware. I don't give a crap about how I got it to work. But a consumer doesn't know. For developers this needs to be a FAQ.
Anyway: oculusd works now, using 752x480 . On to the next thing: RiftConfigUtil segfaults on show-demo-scene ;-).
I agree that it's a bad experience to get a defective cable, but it's difficult to guarantee every cable is perfect. Glad to hear you figured it out, though.
In terms of the segfault, can you provide more information about the error and your OS / system config?
