"No v4l2 loopback device found" after using pyvirtualcam once with exclusive_caps=1
#61
Comments
|
I'm unable to reproduce this on Ubuntu 18.04 (kernel 4.15.0-143) using the same v4l2loopback version with/without running |
|
Also, if you choose a specific device you will see a better error message, for example: with pyvirtualcam.Camera(width=1280, height=720, fps=20, device="/dev/video0") as cam:This must match the device created by v4l2loopback obviously. |
|
Thanks for your quick response. This issue occurs even without any consumer connected at any point. I just started a script and quit it (Ctrl+c) and then tried to restart it, resulting in above error. I also tried specifying the device to use which resulted in this error: Until now I didn't try connecting another producer after having run a pyvirtualcam script. I just tried with OBS, and OBS can connect to the loopback device. I even added another v4l2 source in OBS and selected my loopback device, and it worked. Yesterday, I tried the virtualvideo module which uses ffmpeg for accessing the loopback device. I just tried running their showFish.py sample after having used pyvirtualcam and it worked, too. Neither Basically, it looks like pyvirtualcam leaves the loopback device in a state that makes it unaccessible for pyvirtualcam again but not for any other application. I attached while in the failing case it shows: Interestingly, pyvirtualcam doesn't detect (after): I am not especially apt in reading |
|
Not exactly sure what's going on. How do you create the device? Are you using |
If I remove There's still the question why every other application but pyvirtualcam isn't bothered by having the loopback device been used before. But that's more out of curiosity than urgency. |
letmaik
changed the title
exclusive_caps=1
|
On my end I still can't reproduce it with the same modprobe line you've given, with For me, before running the pyvirtualcam script, |
|
I do face the same issue on gentoo linux (pyvirtualcam==0.8.0). |
|
After modprobe, befor using: After terminating: |
|
I am using Since in my case If there is any other additional information that you need to recreate the issue, please ask me : ) |
|
I'm experiencing a very similar (perhaps, identical?) problem. If I load v4l2 by running:
I can only start pyvirtualcamera once. The second time I do it, I get the following error:
One workaround is to load the module with Details of my system: I'm using Python 3.10.4. |
|
Same issue for me under Ubuntu 22 |
|
Similar issue for me running Ubuntu 22.04.3 LTS:
This is the code I'm running, it's the basic greyscale animation with a SIGINT handler for a graceful exit (which I thought was the issue...): main.pyimport signal
import pyvirtualcam
import numpy as np
GLOBAL_SIGINT_RECEIVED = False
def onSigint(sig, frame):
global GLOBAL_SIGINT_RECEIVED
GLOBAL_SIGINT_RECEIVED = True
signal.signal(signal.SIGINT, onSigint)
with pyvirtualcam.Camera(width=1280, height=720, fps=20, device="/dev/video0") as cam:
print(f'Using virtual camera: {cam.device}')
frame = np.zeros((cam.height, cam.width, 3), np.uint8) # RGB
while True:
frame[:] = cam.frames_sent * 4 % 255 # grayscale animation
cam.send(frame)
cam.sleep_until_next_frame()
if (GLOBAL_SIGINT_RECEIVED) :
breakThis is the output of Output of v4l2-ctl BEFORE running the code, WITH exclusive_caps=1Output of v4l2-ctl AFTER running the code, WITH exclusive_caps=1This is the exception raised by my program:
And this is the output i have creating the device without Output of v4l2-ctl BEFORE running the code, WITHOUT exclusive_caps=1Output of v4l2-ctl AFTER running the code, WITHOUT exclusive_caps=1 |
|
The first execution of the same camera is normal, but the execution after ctrl+c is abnormal. |
|
Fix: Stop the video stream before closing the device, e.g: ioctl(vcam->device, VIDIOC_STREAMOFF, &parm) A fix might be to insert this before 'close(_camera_fd);': Maybe as an extra function, as the close() is used 3 times. |
|
@GermanWarez Are you able to test this and create a PR if it works? |
|
@letmaik TODO: Add ioctl to set both VIDIOC_STREAMON and VIDIOC_STREAMOFF, using the fix for obs as reference. |
|
I hit the same issue. Wait, nobody tested the PR since March? |
Describe the bug
After running a script using pyvirtualcam the next run causes this error:
Only after unloading and reloading the kernel module I can run another script using pyvirtualcam.
To Reproduce
Every single sample code causes this error.
I even rewrote a script to not use the
withstatement and callcam.close()(and evencam._backend.close()) directly.The text was updated successfully, but these errors were encountered: