Hi, hoping someone can help me out here. I’m running the Arducam 64MP camera (the pre-order one) on a Raspberry Pi 64. The setup is basically ARDUCAM 64 + Raspberry Pi. For reference, I’ve been using a 12 megapixel camera with a C mount for a project I run and it’s been fine, and I was planning to order 4-5 of the 64MP units.
My problem: I’m capturing an area of about 59 by 42 cm and I’m getting areas that are out of focus in the image. I’m mostly photographing written characters, so the out-of-focus issue really matters for me. I’m using the maximum resolution.
What I see is unevenness — some areas are blurry, some are very sharp, in the same image. I’m trying to figure out if this is a limitation of the technology / the lens / the CMOS, or something specific to my unit. Honestly I’m hoping the camera has a defect (from transport or production), because then I’d just order another one today. If it’s a limitation of the lens or the CMOS, that’s a much bigger issue for me.
Hi, thanks for the detailed description. A few things to look at here before we can point at the hardware.
First, can you share a couple of the actual images so we can see the uneven focus you’re describing? Full-resolution originals are best, not re-compressed ones, since JPEG compression can hide or fake detail. If you can also tell us how the camera is mounted and the subject distance, that would help.
A couple of quick questions in the meantime:
Are you focusing manually and locking focus before capture, or letting the autofocus run?
Are the blurry and sharp regions consistent across shots (same part of the frame each time), or does it move around?
If the same region is always soft, that tends to point at the optics or sensor alignment. If it shifts, it’s often a focus/stacking artifact.
On the right side of the page you can see the unevenness — some area is blurry, some is very sharp. That’s exactly the thing I’m trying to understand. Why does that happen, and how do I correct it?
Also, a few related questions since I’m trying to work out whether this is fixable for my use case:
If I design and print a C mount for the ARDUCAM 64 and remove the current lens and upper part, would the camera work with a C mount lens? I know autofocus won’t work anymore, but that isn’t an issue for me.
Could this actually be a limitation of the pixel shifting technology? That’s what I understand I’m getting here.
Could it be a memory thing? I’m asking in reference to using the highest resolution vs lowering the resolution.
When you suggest trying again, do you mean using the highest resolution, or lowering the resolution?
Could you also put me in contact with the technical team so they can review 2 of the images and tell me whether this is a known camera limitation or something on my side?
Thanks, I’ve pulled the images and I can see what you mean — the soft and sharp regions within a single frame.
Here’s our read on it:
There does appear to be a virtual focus phenomenon in the shots. It’s worth taking a few more pictures to try and confirm it’s consistent rather than a one-off.
For this kind of symptom there are usually two possibilities:
quality control of the lens, or
quality control of the CMOS.
The reasoning is this: objects sitting on the same focal plane shouldn’t behave this way. In a normal setup, the infinity and the near focus shown in the picture would all be in focus together. When some regions on the same plane are blurry while others are razor sharp, that points at either the lens or the sensor rather than at the scene itself.
As a next test, change the maximum resolution and try again, and see if the same problem shows up. That will help rule out anything resolution/pixel-shifting related.
On your C mount question — yes, people have removed the stock lens and upper assembly to adapt a C mount, and the sensor itself will still capture. Just keep in mind autofocus goes away entirely, and back focus distance becomes your responsibility to get right.
Ok thanks, that’s helpful. To be clear on the resolution question though — when you say “change the maximum resolution and try again,” do you mean go UP to the highest resolution, or drop DOWN to a lower one? I want to make sure I run the test the right way.
And on the QC point — is there any way to tell from my images whether it’s the lens or the CMOS specifically? I’m trying to work out whether a replacement unit would even fix this, or whether every unit has the same uneven focus. I’d rather know before ordering several more.
Good question — for the test, do it both ways: capture a set at the full 64MP maximum resolution, then capture the same scene at a lower resolution. Compare whether the blurry/sharp pattern is identical across both. If the exact same regions are soft at every resolution, that’s hardware (lens or CMOS). If only the high-res/pixel-shifted shots show it, that narrows it toward the pixel-shifting path.
To separate lens from CMOS specifically, one useful thing to try is rotating the camera 90 degrees in its mount and re-shooting the same scene. If the soft region rotates with the lens, it’s the lens. If it stays in the same place on the sensor, it’s the CMOS. I’ve also sent you some additional diagnostic details through private message — please check your messages and give that a try as well.
So the short version of where we are:
The uneven focus you’re seeing matches a virtual focus phenomenon.
Likely cause is either lens QC or CMOS QC, since same-focal-plane objects shouldn’t split into sharp/blurry regions like that.
Please capture a few more images, test at both maximum and reduced resolution, and do the rotate test.
Once you send those back we can be more definitive about whether it’s your unit or a broader limitation. I wouldn’t assume a replacement fixes it until we’ve confirmed which part is at fault.