Hi All,
I’ve released a new beta version of the software, v2.10.0b1. To install it, run:
sudo ofm update
then
sudo ofm upgrade --pre
The most noticable change is that I’ve fixed the white balance issues and improved camera calibration, so hopefully this will be useful to lots of people. If you upgrade and it works (or doesn’t) please post here and let me know - it’s the first release I’ve done since Joel moved on, so I’m keen to know I got it right!
The new camera calibration is excellent. I am unable to make it have a colour cast, even in very low light.
The separate button for auto gain and shutter is really useful as well.
Hello,
Newbie here.
Where do I type those commands?
Thank you
Type on the command line on the Raspberry Pi. If you are using the Pi desktop then click on the Terminal icon in the top menu bar (or choose Menu > Accessories > Terminal). A window opens with a black background and some text. You will see the command prompt.
@r.w.bowman did the about tab go from the webapp? It used to have the version and a shutdown button, but I can’t see it on this beta. I might just be blind.
To add to William’s comment, if you are using the microscope over the network, you can also run the commands by SSH: open a terminal on your computer (I use PowerShell on Windows, or Terminal on a mac) and type
ssh pi@microscope.local
The password is openflexure by default. If you’ve changed the microscope’s hostname, or you have an unusual network set-up, you may need to use a different hostname or IP address; this should show up in OpenFlexure Connect in your microscope’s card, before you connect.
It is there now. I had shut down and restarted in between as I needed to move the microscope, I don’t know if that made a difference. It was not a scroll problem as the window is large and the layout adjusts to have the about at the bottom as long as there is room to fit all the buttons.
Probably just a freak occurrence then.
Thank you,
I just tested it.
Are we supposed to be able to toggle between this 2 options?
Currently I can disable it but I can’t enable it again unless I repeat the auto calibration.
PS: will there be a time course option in the capture menu?
Hmm, that looks like it’s still the old interface. If you have definitely updated, you might need to refresh or “force reload” the interface. The calibration buttons should now look like this:
2.10.0b1?
If you are using OpenFlexure Connect, you can force refresh as shown:
There isn’t currently a way to toggle the flat-field correction on and off without re-calibrating. That would be a useful feature, but it would require a change to the way we handle camera settings. If you felt like requesting that feature by raising an issue on the openflexure-microscope-server repository on gitlab, that will ensure we don’t forget about it!
Thank you, after doing force reload I have
Server version: 2.10.0b1
and I have the new calibration interface.
It’s just that I wasn’t sure if it was meant to toggle or not. Is there a documentation for the software describing how each calibration is done, how the stitching is done, etc…? What does Store raw data do (in capture menu)?
Other questions I have are for example whether the auto gain & shutter speed should be performed with or without a sample (idem for the other calibrations).
Also, when I use the set path button in the storage menu nothing seems to happen…am I supposed to be able to create subfolders ?
Maybe a button reset to defaults settings would be useful in the Camera settings menu.
(and a time course option 
)
Also I can confirm that the 8GB model not supported seems to work fine
(screenshot from this page)
Thank you very much!
I started using the openflexure microscope yesterday and I amazed by the whole microscope and software 
Thanks for that, that’s all useful - with the 8Gb Pi, are you running 32 or 64-bit OS? I think the problem is that if you run the 32-bit OS, it might not use the full 8Gb of RAM. So you may not get any advantage over a 4Gb model. However, it is nice to know that it works!
There’s documentation about the internals of the server, the HTTP API, and how to write plugins. However, there’s not really much of a user manual! We should really sort that out… Currently, stitching isn’t done by the OpenFlexure software, but I think folk in the forum have described how they have stitched images with ImageJ. Maybe start a thread about that if you can’t find more info, and mention @dgrosen because he’s made some stunning stitched images!
“store raw data” means that the JPEG image also embeds the raw data - though you currently need to use a different tool to extract it; the gallery only shows the processed JPEG image, even if raw data is present.
Auto gain and shutter speed should work either with or without a sample - and should be fine to re-run whenever the image looks too bright or too dark.
The “set path” button just allows you to choose from locations in the menu above; it’s really only there to allow you to switch between the Raspberry Pi’s SD card, and external USB storage (if you plug in a USB stick, the autostorage extension should make it easy for you to save images there, instead of the SD card). We should definitely make this clearer when we write a software manual…
A button to reset the camera settings would be useful, I think - though it would probably also need to restart the microscope to work properly. I am fairly sure there is a timelapse function in a demo extension somewhere, we should probably enable this by default, though!
Auto gain & shutter speed is useful with the sample in place, if you have a dark sample or something. It retains the white balance and the even brightness across the picture, but adjusts the brightness so that you can see the sample. The other auto ... are calibration and must be done without a sample. They aim to make the whole image flat white.
As rightly pointed out by @William the autocalibration has improved a lot…I used to see dark pink areas directly under the light source because of not using the condenser lens…however with this update I don’t even require a condenser lens.
But the autofocus ability has gone down when performing stack n scan. Its capturing blurry images mixed with focused images in both the sequences.
Also If some help tab be provided for the different types of scan(spiral, raster and snake)- Like where to start on the slide, Placement of slide on the stage and in which direction the scan is performed ?..it would be really helpful.
I’m surprised to hear the autofocus is less reliable; I can’t think of any code changes that should affect that! Do you mean the “fast” autofocus, or one of the older coarse/medium/fine versions? The fast autofocus method is quite sensitive to timing, but I just can’t think of anything that has changed between the beta and the last release. We have upgraded numpy to 1.20, and a few other dependencies, but none of them ought to have changed anything…
It’s the fast one I suppose which comes into action during stack n scan.Scan parameters are as follows.
Step number-15 15 1
Step size- 2000 2000 50
Spiral
Does it have anything to do with the Z step no which I set as one ,because I wanted the scan to complete quickly and previous auto focus was quite good …and I wanted less no of pics for stitching.
EDIT: I had mistakenly run the forced reset in the “OpenFlexure Connect” window not the “Microscope” window. The new features show correctly now. Thanks!
I’ve updated as per the instructions including the forced reset. I’m showing Server version 2.10.0b1 however I’m not getting the new features. Any suggestions?
I guess the trouble with assessing autofocus reliability is that unless you go to some serious lengths to collect statistics, it’s quite hard to really know if it’s become better or worse… @JohemianKnapsody has been working on improving the fast autofocus (especially if you want to acquire a Z stack) but that will take a little while to make it into a beta release, and is currently only tuned for using a 100x lens.
Setting the number of Z steps to 1 will disable Z stacking, but should have no effect on the autofocus routine. If the number of Z steps is 1, the Z step size is ignored.
You’re right, it would be good to document the scan paths somewhere. Again, @JohemianKnapsody did a nice job of visualising scans, so we should figure out how to integrate that into the GUI in a future release. In the meantime, we really ought to start some sort of software manual…