Hi all, hoping someone can help. I have two IMX477 HQ cameras on a Raspberry Pi CM4, one on each CSI port, using the picamera2 library. Camera 0 captures images fine, but camera 1 comes out almost entirely pink - the colors are completely wrong. I noticed one difference between them: camera 0 is a Rev D board, camera 1 is Rev E. Could the hardware revision really cause this? I’ve attached image_cam_0.jpeg (correct) and image_cam_1.jpeg (pink), plus the terminal output for each camera (output_camera_num_0 and output_camera_num_1).
Thanks for the detailed report. The pink output with the Rev E board is suspicious. Could you share the capture script you’re using (capture_example.py) so I can see the exact still configuration? Also, when both cameras are enumerated by picamera2/libcamera, do they report the same sensor modes and resolutions, or is there any difference in the logs?
Sure, capture_example.py is attached - it just opens both cameras, applies the still configuration, captures, and saves. The terminal outputs are the two files I mentioned: output_camera_num_0 is camera 0 (Rev D) and output_camera_num_1 is camera 1 (Rev E). Both cameras enumerate and stream without errors, but camera 1’s image is pink every single time. I’ve tried different still configurations with the same result - I never got camera 1 to work properly.
Thanks, that helps narrow it down. From the outputs both cameras are detected and list the same sensor modes, so the difference really does look like the Rev E board behaving differently - we’ve seen a hardware revision difference cause exactly this kind of wrong-color/pink image after a mode switch, while Rev D stays correct under the same picamera2 setup. One thing to double check: B0449 is actually the IMX519 module, not the IMX477. The IMX477 HQ SKU you’d want is B0466R - can you confirm which SKU is printed on the Rev E board? I’ve also sent you a small firmware troubleshooting package through private message in case we need to compare firmware behavior between the revisions.
Good to know about the SKU - I’ll double check the board marking. I’d really like to get the Rev E working if possible, because the vendor told us the D version is discontinued and only the latest E version is currently available. So if there’s a way to fix the E, that would be ideal. What should I try next?
Let’s rule out the picamera2 path first. Can you test camera 1 (Rev E) directly with libcamera-still and check whether the colors come out correct:
libcamera-still -t 5000 -o test.jpg
If this produces a normal image, the problem is in the picamera2 usage path. If it’s still pink, the issue is in the Rev E camera hardware itself.
To summarize where we are: the issue is isolated to the Rev E camera (camera 1), while the Rev D camera (camera 0) works correctly under the same picamera2 setup. Running the Rev E board through libcamera-still directly will tell us whether this is a hardware revision issue or a picamera2 issue. Please run the command and let me know the result.