Autocalibration Inconsistencies

Request Help
#1

Hi! We are running the high-powered non-motorized version of the OpenFlexure microscope with the Raspberry Pi camera module and a microscope objective (10x for these images) and are having some inconsistencies when running the autocalibration part of the software. We used to get a pure white screen after calibration, but now it varies. Attached are images taken immediately after calibration without any slides in the microscope. We ran calibration a number of times in a row without any external changes (e.g. in ambient lighting) and got different results each time. Ran the rescue code and tried swapping out camera modules and lenses with no luck.

Anyone encountered anything similar?

postcalib10x_1
postcalib10x_1832×624 21.4 KB

#2

Here’s a few more images, I could only attach one above.

postcalib10x_2
postcalib10x_2832×624 15.4 KB

#3

Here’s a few more images, I could only attach one above.

postcalib10x_4
postcalib10x_4832×624 21.1 KB

#4

Welcome to the Forum @mpearson .
This is an issue that is known and a fix will be coming in a future release of the Server software. Sometimes it is so extreme it gives a deep green background. It is not an issue with the camera but with the way the calibration currently works.

Until the new Openflexureserver release there are some ways to get the calibration to work. First, if the calibration is not far off then simply re-running the calibration can get it correct. Sometimes it can take three or four runs. Second, sudo ofm reset sudo ofm restart will clear most of the calibration from the server and give the calibration a fresh start. Third, the issue can be reasonably reliably induced by reducing the lighting, so make sure that your lighting is well aligned so that the camera has plenty of light to work with. With plenty of light and a fresh server the calibration is pretty reliable.

Y a t-il des français pour m'aider?
#5

For some more details there is an issue on Gitlab, and a fix in a merge request

#6

Hi @mpearson, as William says this is something we’ve spotted, and I believe I’ve fixed it - if you are comfortable pulling down the new version from gitlab, you can test it out today; if not, I will need to have a bit of a merge-fest then make a new release. That might take me a little while, as Joel used to do this and has now moved on, so it will be my first attempt…

2 Likes
#7

Thanks for the help! I haven’t been able to successfully pull down the new version from gitlab; I’m not too familiar with git.

I ran the reset and restart commands and made sure there is plenty of light. I’ll check with some other lab members to see if anyone is more familiar with git and would be able to pull the new version.

#8

sure - in order to do that you would need to run ofm develop to switch from the released version to the master branch. You’d then need to switch from the master branch to our calibration branch, and manually rebuild the web application that forms the interface.

I suspect you may be better off awaiting the release with the calibration information - I will try to do this soon!

1 Like