[BUG] {FFC-6P is not detecting the camera with correct name and not able to stream video}

[gajena]

Check if issue already exists

• Google it (e.g. error xy github luxonis depthai)
• Check troubleshooting in documentation.

Describe the bug

1. Recently bought FFC-6P flashed with LuxOS 1.2 here
2. Attached Camera OAK-FFC 48mp and IMX477 on CAMA and CAMSB
3. When following the setup document the cmd gst-inspect-1.0 kmbcamsrc output was showing two cameras with wrong names:- camera-id : (0): kmb-imx412-sensorkeembay_cam_a2 (1): kmb-imx412-sensorkeembay_camex_
4. Also cmd ls /usr/bin/*.bin show only some of the camera bin files

   /usr/bin/calib_data_0.bin           /usr/bin/imx335_database.bin            /usr/bin/imx412_database_imesh.bin
   /usr/bin/imx219_database.bin        /usr/bin/imx412_database.bin            /usr/bin/kmb_dtp.bin
   /usr/bin/imx327_database.bin        /usr/bin/imx412_database_1920x1080.bin  /usr/bin/ov9282_database.bin
   /usr/bin/imx334_database.bin        /usr/bin/imx412_database_3200x1800.bin
   /usr/bin/imx334_database_imesh.bin  /usr/bin/imx412_database_3280x2464.bin

5. I tried accessing it from host system using depthai-python with rvc3_support branch with only OAK-FFC 48mp on CAMA port

   $ python3 utilities/cam_test.py -cams rgb,c -cres 48mp
   Enabled cameras:
   rgb : color
   DepthAI version: 2.19.1.0.dev+e23b04d694ce1fcd52955b3bde265602eb1ce87a
   DepthAI path: /home/gn/src/luxonis/depthai/luxonis/lib/python3.8/site-packages/depthai.cpython-38-x86_64-linux-gnu.so
   Connected cameras:
   -socket RGB   : imx412   -1 x   -1 focus:fixed - 
   USB speed: UNKNOWN
   IR drivers: []
   Cam:      rgb      [host | capture timestamp]
   FPS:   0.00|  0.00(base) 

   tried with low res same result.

6. tried color camera example code, got stuck at the frame get method.

Minimal Reproducible Example Append the MRE to the bug report, instructions here

Expected behavior A clear and concise description of what you expected to happen.

Screenshots If applicable, add screenshots to help explain your problem.

Pipeline Graph

Please also provide a screenshot of your pipeline using the DepthAI Pipeline Graph.

Attach system log

• Provide output of log_system_information.py

Additional context Add any other context about the problem here.

[themarpe]

Hi @gajena

RVC3 is still in development - WRT camera naming, you should look into rvc3_support or rvc3_develop branches and run the device.getConnectedCameraFeatures() function. Also, for 6P you should likely update the FIP as well, to get the new naming for the camera ports.

Otherwise, there are some WIP tmp_s3 examples & rvc3_develop can run quite a number of regular examples already as is.

cam_test.py utility is not yet tested to be RVC3 compatible. Use at your own discretion.

48MP sensor has a known issue with streaming from DepthAI at the moment, something which we are working forward to address.

[gajena]

Hi @themarpe, thank you for the reply. I am currently using the rvc3_develop branch and output of device.getConnectedCameraFeatures() is (IMX477 with motorized focus on CAME and 48mp OAK on CAMD):

[{socket: VERTICAL/CAM_D, sensorName: imx412, width: -1, height: -1, orientation: AUTO, supportedTypes: [], hasAutofocus: 0, name: keembay_cam_c1}, {socket: CAM_F, sensorName: imx412, width: -1, height: -1, orientation: AUTO, supportedTypes: [], hasAutofocus: 0, name: keembay_cam_b2}]

I did update FIP from here with the LuxOS1.2 if there is any updated version of FIP please let me know or can I build the same myself using the latest commit?

I am able to receive images from the IMX477 camera with rgb_video example but it has a weird stutter(or uneven frame pacing), I am using this arducam camera and with the FFC cable. Is there any modification required if I use the arducam directly with this FFC cable?

[gajena]

IMX477 works fine when I used other camera ports. Few queries: does the current rvc3_develop branch has support for

• IMX477 motorized focus control,
• OAK 48mp camera,
• ISP support for the supported sensors? And what is the best way to know what nodes are currently supported on s3 hardware since commit-msg says IMage manip is there but when I run it RuntimeError: Pipeline node with name: 'ImageManip' doesn't exist? (probably develop branch overwrote the Image Manip firmware)

[themarpe]

I am able to receive images from the IMX477 camera with rgb_video example but it has a weird stutter(or uneven frame pacing)

The high resolution is likely the culprit. We are addressing this.

• IMX477 doesn't yet have focus control available
• 48mp is partially supported but not yet from DepthAI perspective
• Yes, ISP is implicitly supported alongside the sensor if sensor is claimed "supported'.
• As of right now there isn't any API to discover this. We might add this in the future though.

[gajena]

On some ports, the same camera performs worse and on some ports, it looks fine. Tested the following config:

• CAM-SA Works okay with IMX477
• CAM-A very low fps with IMX477
• CAM-E has very low fps with stutter issues
• CAM-F works okay
• CAM-SB fps seems better than CAM-E but same issues with stutter
• CAM-D seems fine

So seems like three ports works okay and other do not. Probably related to this:

will contact support soon. Thank you for your time.

[nonathecoda]

@gajena

Are you running python3 utilities/cam_test.py -cams rgb,c -cres 48mp on the oak device or on your host computer? I tried running it on the oak device, but keep getting an error:

cv2.error: OpenCV(4.5.5-kmb) ../opencv/modules/highgui/src/window.cpp:1334: error: (-2:Unspecified error) The function is not implemented. Rebuild the library with Windows, GTK+ 2.x or Cocoa support. If you are on Ubuntu or Debian, install libgtk2.0-dev and pkg-config, then re-run cmake or configure script in function 'cvWaitKey'.

However, libgtk2.0-dev isn't suitable for the device's distribution. Did you run into the same problem?

[gajena]

Hi @nonathecoda, No never seen this issue. Are you trying to run this command on the device itself or running it on the host machine?

[nonathecoda]

@gajena i'm running it on the device itself. It might have something to do with some x11 forwarding issues though, so i'll look into that.