First to say that the camera module 3 is not supported by the current OpenFlexure software v2. It also does not fit the current v7.0.0-beta2 hardware.
Also you need to be clear that this is likely to break the focusing mechanism, it might break the whole camera.
If you are looking to use the camera with other hardware/software then I found that the lens unscrewed simply by gripping the front black roundish part with a pair of pliers. There was quite a lot of resistance to start with, it is probably glued in. Once started, it unscrewed quite easily. I have only tried on one example, so I might have been lucky.
1 Like
Thank you! It was easy to remove the lens with your advice.
1 Like
I tried the optics module for the picamera3 from your Gitlab fork @WilliamW. Had to add 1 mm to the camera height for clearance, otherwise the design works fine. Removing the lens worked just like you said, a bit of resistance at first but after that it was easy to unscrew.
Here is a picture I took with a 10x objective and libcamera (hence no lens shading/color correction):
And this was taken with the same objective and the camera module 2 with the openflexure microscope:
The picamera 3 image seems a bit sharper to me but with more noise. Also it would need some cropping to remove the black bars (they are from the camera module itself, not from the 3D printed optics part).
The camera module 3 has a noticeably larger field of view, and the image appears a bit sharper but also has more noise. Overall, for a first quick test I’d say the images were quite ok and I’m looking forward to ofm3 and software support for this camera module.
Thank you @rootboi this is very helpful. The image field of view seems to match the difference in sensor size 3.68 × 2.76 mm for v2 and 6.45 × 3.63mm for v3.
What is slightly surprising is that the image is reasonably clear to the edges of the larger sensor. The expected image circle of the lens is matched to the v2 sensor size, as noted in Assembly Instructions . Your images suggest that your objective image circle is bigger. There is a fall-off in image quality, but the edges would still be very useful for stitching. My feeling is that we crop the v3 to square for the preview and normal images, but use the whole sensor image to reduce the image overlap required for reluable stitching.
1 Like
Nice work!
I want to try using the camera directly with the autofocus lens to do close up focus stacked macro shots and get the magnification by reducing the distance from the sensor to the object.
So I should be able to unscrew the built in lens a bit and reduce the focal distance?
How many turns can it make before coming free?
That should work. The focal range of the lens movement becomes very small when you unscrew it for close focus. I don’t know how many turns it is to unscrew, but if it comes out you can just screw it back.
Once again - this may well break the focus mechanism. I was amazed that it still seemed to work after I put the lens back in.
Yes good idea to make use of the blurry edges for image stitching at least.
I was wondering, do you think the pictures would be a bit less blurry in the corners with a more expensive tube lens or is it just the limit of the optics? And a tube lens with a longer parfocal distance would lead to less blurriness, right?
After some fiddling with the the RPi Camera 3 WIDE version I have the following info to report:
-
the lens makes ~2.5 turns before coming free.
-
unscrewed it by 1 full turn.
– The focal range changed to go from ~3.5cm to 7cm
-
unscrewed a further half turn to 1.5 turns total.
– focal ranges changed to ~2.5 - 4.5cm
-
Looks like the minimum focus distance doesn’t change as fast as the max.
-
It can be unscrewed with the camera powered on and the focussing voice coil powered, and it still works afterwards 
Attached is an image captured at 2.8mm distance and focal position 2.0. (not a focus stack).
I generated the code to create a UI that I could use to adjust focus and capture an image using the grok LLM 
1 Like
