Skip to content

RUN_THESE_TESTS.md

Latest commit

 

History

History
132 lines (100 loc) · 3.96 KB

RUN_THESE_TESTS.md

File metadata and controls

132 lines (100 loc) · 3.96 KB
Raw

How to Test Nvidia CSI Camera Exposure/Gain Settings

IMPORTANT: Run Tests on HOST, Not in Docker!

The nvarguscamerasrc GStreamer element ONLY works on the Jetson host, not inside Docker containers. You'll get the error "no element nvarguscamerasrc" if you try to run it in Docker.

Step-by-Step Testing

1. Exit Docker (if you're in it)

exit

2. Navigate to the project directory on the host

cd /path/to/DefectDetectionApplication

3. Run the environment check

chmod +x check_where_running.sh
./check_where_running.sh

This will tell you:

  • If you're on the host or in Docker
  • If nvarguscamerasrc is installed
  • If the capture service is installed

4. Install missing packages (if needed)

If nvarguscamerasrc is not available:

sudo apt-get update
sudo apt-get install -y gstreamer1.0-plugins-nvargus nvidia-l4t-gstreamer gstreamer1.0-tools

5. Run the simple camera test

chmod +x simple_csi_test.sh
./simple_csi_test.sh

This will capture 3 images:

  • /tmp/csi_test_default.jpg - Default settings
  • /tmp/csi_test_dark.jpg - Low exposure/gain (should be dark)
  • /tmp/csi_test_bright.jpg - High exposure/gain (should be bright)

Compare the images: If the dark and bright images look the same, the camera is ignoring manual settings (auto-exposure is overriding).

6. Check the capture service

chmod +x check_csi_service.sh
./check_csi_service.sh

This will:

  • Show service status
  • Display current config
  • Test config updates
  • Show recent logs

7. Watch service logs in real-time

sudo journalctl -u nvidia-csi-capture.service -f

Then in the DDA UI:

  1. Go to Edit Image Settings for your Nvidia CSI camera
  2. Change exposure or gain values
  3. Click Save
  4. Watch the logs - you should see "Settings updated - Gain: X, Exposure: Y"

8. Run full diagnostics (if issues persist)

chmod +x diagnose_gstreamer.sh
./diagnose_gstreamer.sh

Expected Results

If Everything Works:

  • simple_csi_test.sh creates 3 images with different brightness levels
  • Service logs show "Settings updated" when you change values in UI
  • Image preview in UI changes brightness when you adjust exposure/gain

If Exposure/Gain Don't Work:

  • All 3 test images have similar brightness (auto-exposure is active)
  • Service logs show settings updated, but image doesn't change
  • This means nvarguscamerasrc is ignoring the manual parameters

Next Steps if Manual Control Doesn't Work

If the camera ignores manual exposure/gain settings, we may need to:

  1. Try different nvarguscamerasrc properties:

    • Check available properties: gst-inspect-1.0 nvarguscamerasrc
    • Look for aelock, awblock, or other AE control properties

  2. Use v4l2-ctl to control camera directly:

    v4l2-ctl -d /dev/video0 --list-ctrls
v4l2-ctl -d /dev/video0 --set-ctrl=exposure_absolute=200

  3. Check Jetson forums for IMX219-specific manual exposure control

  4. Consider alternative approach: Use post-processing to adjust brightness instead of camera settings

Files Created for Testing

  • check_where_running.sh - Verify you're on host, not Docker
  • diagnose_gstreamer.sh - Full GStreamer diagnostics
  • simple_csi_test.sh - Quick camera test with 3 exposure levels
  • check_csi_service.sh - Service status and config test
  • test_nvargus_params.sh - Comprehensive parameter testing (advanced)

Common Issues

"no element nvarguscamerasrc"

  • You're running inside Docker - exit and run on host
  • OR nvarguscamerasrc is not installed - install packages above

"Could not get EGL display connection"

  • You're running inside Docker - this confirms nvarguscamerasrc needs host access

Service not found

  • Run installation: sudo bash src/host_scripts/install_nvidia_csi_service.sh

Config updates but image doesn't change

  • Camera has auto-exposure enabled that overrides manual settings
  • Need to find correct parameters to disable AE
  • May need alternative approach