I am able to get the depthai_demo.py program to work when I choose model source of left or right to work but color does not seem to work. The face detection worked for the left and right cameras but not color. Also this tool said the same thing about the calibration data and it said…loading default oak d calibration data.
Video Window Does not render
Hi BrianHouse ,
Could you share the full terminal output? It might be you have a very old camera, which has old calibration. If you want to use all the features, you'd unfortunately need to recalibrate the device: https://docs.luxonis.com/projects/hardware/en/latest/pages/guides/calibration.html
Hi Erik,
What command should I run to get you the diagnostics that wold be helpful?
Thanks,
Brian
Hi BrianHouse
The terminal output you get when running the demo (the one that prints the 'Device Contains calibration data of version 1437204487 which is not supported in this version of depthai' error) should be fine atm.
Thanks,
Jaka
Here is the requested output:
pi@raspberrypi:~/depthai $ python3 depthai_demo.py
Using depthai module from: /home/pi/.local/lib/python3.9/site-packages/depthai.cpython-39-arm-linux-gnueabihf.so
Depthai version installed: 2.13.3.0
Setting up demo...
Available devices:
[0] 184430100149620E00 [X_LINK_UNBOOTED]
[184430100149620E00] [1182.272] [system] [warning] Device contains calibration data of version 1437204487 which is not supported in this version of depthai
USB Connection speed: UsbSpeed.SUPER
Warning: calibration data missing, using OAK-D defaults
Hi Erik,
I just bought this camera this week, is it possible i still got an old model? Yes, I want to use all the features, that is why I bought the camera. I will try the recalibration. I am new to this device, hopefully I want make it worse.
Brian
Hi @BrianHouse , Where did you buy the camera - via our official store, or via distributor?
The Luxonis store, Order #13634.
Shouldn't the RGB camera work out of the box?
Hi BrianHouse ,
In theory yes, but I am not sure if there are some processes (eg dewarping) which rely on the calibration part. An alternative would be to just use an old depthai version, which doesn't require the new calibration. Thoughts?
Whatever makes sense. If i need to recalibrate the device to make all the cameras work with the newer depthai I can do that.
This link will show me how to do that?
https://docs.luxonis.com/projects/hardware/en/latest/pages/guides/calibration.html
Which version should I install of depthai? what is the best way to install that specific version making sure the newer version is 'removed'?
Hi BrianHouse ,
Yes, I think you should just use latest depthai (pip install depthai -U) and calibrate the device, the link you posted would be the best resource for that.
Thanks, Erik
Thanks Erik.
Hi Erik,
One last thing I decided to try before the calibration was to set this up on my MacBook Pro running Ventura 13.3.1 (a). So when I ran the depthai_demo.py script, the RGB camera still showed no video, just like the Raspberry PI, but I feel like I got more output that may help determine what is going on with the RGB camera. I am trying to rule out of this is a physical or configuration issue with the OAK-D Lite. See below:
(base) Brians-MacBook-Pro:depthai brianhouse$ python3 depthai_demo.py
Using depthai module from: /Users/brianhouse/.local/lib/python3.9/site-packages/depthai.cpython-39-darwin.so
Depthai version installed: 2.20.2.0
Setting up demo...
Available devices:
[0] 184430100149620E00 [X_LINK_UNBOOTED]
USB Connection speed: UsbSpeed.SUPER
/Users/brianhouse/.local/lib/python3.9/site-packages/depthai_sdk/previews.py:186: RuntimeWarning: invalid value encountered in cast
dispFrame = (dispFrame * 255. / dispIntegerLevels).astype(np.uint8)
Stopping demo...
=== TOTAL FPS ===
File "/Users/brianhouse/Luxonis/depthai/depthai_demo.py", line 656, in run
self.instance.run_all(self.conf)
File "/Users/brianhouse/Luxonis/depthai/depthai_demo.py", line 133, in run_all
self.run()
File "/Users/brianhouse/Luxonis/depthai/depthai_demo.py", line 317, in run
self.loop()
File "/Users/brianhouse/Luxonis/depthai/depthai_demo.py", line 373, in loop
self._pv.prepareFrames(callback=self.onNewFrame)
File "/Users/brianhouse/.local/lib/python3.9/site-packages/depthai_sdk/managers/preview_manager.py", line 148, in prepareFrames
packet = queue.tryGet()
RuntimeError: Communication exception - possible device error/misconfiguration. Original message 'Couldn't read data from stream: 'color' (X_LINK_ERROR)'
Hi BrianHouse
Do scripts that don't utilize the color camera work? e.g. Mono preview
Thanks,
Jaka
Yes all scripts referencing the right and left mono cameras work perfect. Just the RGb does nothing.
Specifically I have been trying to compare and contrast the rgbpreview.py (that does nothing, generates no errors but does not even render a black window) and mono_preview.py that does work. I am trying to debug these variables in pycharm:
xoutLeft = pipeline.create(dai.node.XLinkOut)
According to the hellow world example provided in the documentation, this is where the camera gets 'added' whether it be RGB or Monjo. I am trying to figure out what is different with the camera that works and the one that doesnt.
Anything specific I could look at to determine where the issue may lie.
I was looking at the connection attribute but is says 'None' for the value for both cameras so that is not a good indicator why one would work and one would not. Any variable specifically I could set a debug 'watch' to shed more light on this?
Hi BrianHouse
I feel like the XLink for color just isn't sending frames to a queue correctly. When reading from the rgb queue with .get() the script probably blocks, since there is no frame to be read and .get() is a blocking function.
I believe since you are running all the latest versions of software, it might be a hardware fault.
Could you:
- try to run the pipeline graph tool to see where the stream stops
- try the same script with the same camera and usb cable on a PC to confirm
Thanks,
Jaka
Hi BrianHouse
Sorry for not linking to the repo. Here.
Run it using pipeline_graph run "python3 <name of script>.
Thanks,
Jaka
