Hi everyone. I’m working on a steel coil defect detection project and I’ve got an Arducam 64MP camera with the OV64A40 sensor paired with a Raspberry Pi 5. I need some help on two fronts: first, what are the optimal camera settings to get the sharpest possible images? And second, I need some guidance on getting libcamera installed and configured properly for this particular camera. This is my first time using Arducam hardware so I want to make sure I don’t mess up the software setup. Any advice would be appreciated.
On the optimal settings question — there isn’t really a universal answer we can give you. Image quality depends a lot on your environment: lighting conditions, how far the camera is from the steel coils, what kind of defects you’re trying to catch (scratches, discoloration, etc.), and so on. You’ll need to dial in parameters like resolution, exposure, focus, shutter speed, and lighting yourself based on your actual setup. The docs at 64MP OwlSight - Arducam Wiki walk through how to adjust those parameters, so I’d start there and experiment.
One thing worth noting upfront: the Arducam 64MP (OV64A40) is what we call a “Native camera” — it’s designed to work with the Raspberry Pi’s built-in camera software stack. A support engineer will jump in shortly to help with the libcamera side of things.
Thanks for the doc link. We’ve been working through that guide and we’re in the testing phase now. The exact model I have is B0483 — the Arducam 1/1.32" 64MP Autofocus Camera Module. Running Raspberry Pi OS on the Pi 5.
I’m still a bit confused about the libcamera situation though. Do I need to install Arducam’s own version of libcamera, or some special build? Or does the stock libcamera that comes with Raspberry Pi OS just work? The image quality is critical for our defect detection algorithms so I want to make sure I’m not missing a step on the software side. Any pointers to get me unblocked would be great.
You don’t need any custom or Arducam-specific libcamera installation — that’s the whole point of it being a native camera. The B0483 (64MP OV64A40) works out of the box with the Raspberry Pi’s built-in rpicam-apps stack (which replaced the older libcamera-apps). If you’ve already been through the documentation, your Pi should detect the camera automatically.
Let’s verify everything is working. Run this in the terminal first:
ls /dev/video0
If the camera is recognized, you’ll see /dev/video0 listed. Assuming it shows up, start a live preview with:
rpicam-still -t 0
The -t 0 flag means no timeout, so you’ll get a continuous preview until you close the window.
Does the preview window appear? Let me know what you get and we’ll troubleshoot from there if needed.