I’m trying to use the Raspberry pi AI camera for some machine vision stuff for our pioreactor, but I’m getting a “No cameras found” error from libcamera. With the same hardware but a different MicroSD with stock raspbian I can get the camera working.
I’m can try to track down what might be happening, but figured I’d ask here first in case there was some reason this behaviour is there/an easy fix.
sudo libcamera-hello
[0:05:52.512190594] [861] INFO Camera camera_manager.cpp:327 libcamera v0.4.0+53-29156679
[0:05:52.567777276] [865] WARN CameraSensorProperties camera_sensor_properties.cpp:473 No static properties available for 'imx708_wide'
[0:05:52.567960755] [865] WARN CameraSensorProperties camera_sensor_properties.cpp:475 Please consider updating the camera sensor properties database
[0:05:52.724463452] [865] WARN RPiSdn sdn.cpp:40 Using legacy SDN tuning - please consider moving SDN inside rpi.denoise
[0:05:52.729070985] [865] WARN CameraSensor camera_sensor_legacy.cpp:501 'imx708_wide': No sensor delays found in static properties. Assuming unverified defaults.
[0:05:52.732199648] [865] ERROR RPI vc4.cpp:421 Unicam driver does not use the MediaController, please update your kernel!
[0:05:52.732397460] [865] ERROR RPI vc4.cpp:216 Failed to register camera imx708_wide: -22
Preview window unavailable
ERROR: *** no cameras available ***
So the kernel module is out of date perhaps, or deliberately not loaded?
Hm, and I’m guessing you followed the instructions here already?
Check your linux kernel version with pio version -v, look for the Operating system line. It’s possible you are using an older kernel version (we don’t update kernels in our Pioreactor updates) vs the latest image from RPi Foundation.
There is possibly a solution here (translate to english unless your a Deutschsprachiger) - they report the same error as you, too.
This is really cool, I am planning on adding some cams into our incubator. Is there anyway this could be implemented as a pluggin for those (like me😅) that have no coding experience?
Hey! Happy to have a more in depth conversation, and will try to set something as a plugin at some point, but two quick lessons for now:
We’ve found in our case snapshots every 5 minutes to be more useful than video. We mainly just want to check for foam and have a log of what was happening over the course of the experiment and that does the job. It takes up much less room than continuous video. Streaming was sort of helpful but actually we’re more often interested in whether we missed something or what’s changed than exactly what’s going on right now.
Webcams tend to be better than the rpi camera because I think things might have changed since the last time I posted here getting it working. The rpi cameras have really short leads. Webcams take up a lot of CPU if you’re recording continuously but if you’re ok with snapshots as discussed above they’re just more convenient and cheaper
We have a job that is triggered by cron every 5 minutes and then rsyncs to a Google cloud storage bucket
Ok, it makes total sense. In my case I want to track the Pio when I am not in the lab, I work with pressure fermentation and being able to check that there haven’t been any unfortunate explosions (it has happened before😅) let us turnoff the system from anywhere to avoid damage. So I think that for this maybe having a webcam makes more sense, in witch I can set some alarms if an object moves drastically (like a vial lid).
Thanks a lot!!! I’ll keep you posted if we make more advances.