Testing all canon camera setup for alice

[Baeist]

Trying out a setup with only canon cameras instead of the 5 raspberry pi's. Goals of this is to check if its more robust and to implement the extra canons cameras with the entomology gui.

So far the results are varied and there are a lot of issues. Things to note: -Implementing new cameras with the gui is relatively easy.

• When running one camera from the other pinned insect stations this can still cause the camera to disconnect itself. So there is some inbuilt instability with the cameras- this is with software provided by canon(it is not our software anyway). If this is indeed an issue that will continue to pop up with each canon then having an only canon camera setup would risk being somewhat disruptive throughout a full day.
• Adding too many cameras (more than 2) to one usb card controller through the macs ports (2 usb c in front share a card, 4 usb c in back share a card and 2 usb a in the back share a card), appears to cause the connection and photo shooting to become less robust.
• Connecting the cameras works fine when they have not been connected before after being switched on. After the first connection it become a lot more uncertain what happens. Fex. if for some reason the entomology gui is shutdown and reopened and the cameras are then being reconnected they are unlikely to connect immediately and require some reconnection attempts from within the gui menu. Sometimes this can only be solved by turning the camera off and on again before reconnecting.
• When attempting to take photos with the cameras there are some real issues. Only 3 out of the 6 that have been tested reliably takes photos currently. The 3 are the 2 original cameras and the only other R5 that have both a spare memory card and a spare lens. There can be a multiple reasons for this. Cameras are not exactly the same model(EOS R5 and EOS Mark IV). We are missing a lens for the last R5. We are missing the same type of memory card that the original 2 canons have. This can be tested further by moving the hardware around between the cameras, but it wont allow for a full test. It is worth noting that Mark IV cameras save their images as cr2 files instead of cr3. These issues persist even when not using the gui, but the cli(command line interface) provided by gphoto2(software library) which when operated manually only tries with one camera. It gives a PTP error 110 and often the image is not there even if the output says it has been saved. In the gui this just makes the whole shoot fail.
• When shooting with many canon cameras even though they fail it takes up to 15 seconds for all of them to finish shooting(they shoot sequentially, should be in parallel, and the last one appears to lag behind significantly). The lagging behind could also be due to the camera and the time it uses for saving the image locally. This appears to be a software/gui issue and fixing it would require some rework on the guis core functionality.
• Most of the testing was done with 2 raspberry pi's connected through the usb a ports which were both performing perfectly. Perhaps with the exception of the image files generated by them missing some of the exif data we would want.
• Adding both barcode and digitizer directly to the exif data (hidden data belonging to an image) from the gui looks to be possible. Would need to be built into the guid. This also means if there are any other data we would like to add before we take the photos we can.
• The gui is built with pyqt(python library). It is a fairly inflexible to adjust too much with the way the initial setup of the code is. Adding new buttons, views and potentially a barcode scan field.

Current ideas for more testing includes checking if the actual camera settings could be causing the issues with the photos not correctly being saved. Switching the hardware around to check if that is causing the failure(memory card and camera lens). Testing a 4 raspberry pi, 3 canon camera setup or 3 raspberry pi, 4 canon setup. Addding the 7th canon camera to the test- would likely want another lens available for this since borrowing from the second pinned insect station isnt ideal.

@bhsi-snm @PipBrewer @Gomismis @chelseagraham @ThomasAlscher1991 @k-zamzam

[PipBrewer]

I would suggest throughly exploring the all-canons set up for now and resolving issues, so we have a full proper test and baseline, before complicating the situation further (e.g., mixing pis and canons). We could test same memory cards, lenses and models (including hiring for a day or two).

@Gomismis in conjunction with @Baeist perhaps you could reach out to NHM (London)? I think they have all canons in their ALICE set up. What are their solutions for e.g., USB controller. Have they encountered these issues before?

[Baeist]

Anders came by and suggested we try an opensource app called digicamcontrol that runs on windows(written in C#). So i tried that and it works really well with the canons. Downside to this would be that it wont work at all with any raspberry pi´s out of the box. We could build a module that would allow this. It has some pretty great features such as inbuilt barcode scanner that allows you to scan and then press one button and all cameras shoot and has barcode added to their filename. Liveview of the cameras you specify and also easily configurable interface for choosing cameras in general. You can change pretty much all the settings on the cameras. It is automatically set up to run session based shooting where you can specify additional information. The UI itself is much nicer to look at than the entomology gui. It can do stacked images(bracketed) and it does not appear to be especially slow. There is reasonable documentation for how to use it.

[Gomismis]

That is amazing news, @Baeist! As we were talking about yesterday, I read the articles that NHM/London made about their ALICE setup. Their cameras are tethered in parallel using DSLR Remote Pro Multi-Camera v. 1.9 (https://www.breezesys.com/solutions/breeze-multi-camera-non-event/ used for their setup). This system costs $490 and can connect up to 100 cameras. They also mentioned that DigiCamControl v.2.1.0 was tested briefly and found to be a sustainable, free, multi-camera tethering alternative.

[Gomismis]

@PipBrewer, should I still contact NHM/London, or should I wait until we are further along in the physical development process? I have already noted several questions regarding their setup/development, that I would love ask them.

[Baeist]

Tested the entomology gui with 7 canons connected and 6 that were able to shoot (missing one lens for the last camera, and the last pinned insect was in use all day so couldnt borrow the last lens). This worked fine when cameras were newly switched on.

Discovered that the mac/cameras sometimes changes the ports assigned to the cameras seemingly at random but almost always when a command is given through the ghoto2 cli (fex port change from usb:001,002 to usb:001,001). This caused some very strange behavior. Trying to get the camera names(the ones we have set and uses to identify each canon with) i had multiple cases where one camera name was given twice or more. So what i think happens is that the port is switched or changed around between the calls to get the names and this somehow made this possible (the calls has to happen sequentially even with a shell script). I also think this might be the cause of cameras sometimes disconnecting or becoming unresponsive until they are turned off/on again.

[Baeist]

Borrowed the last lens from the other pinned insect station. Ran a test of automatically shooting a full set of 7 images every 30ish seconds 60 times. This concluded successfully.

Before that there was an issue with the dassco0063 camera that gave an internal communication error 70 (happened maybe 5-6 times while running pre tests). This is a known issue with the canon EOS R5 camera and if it persists we will have to send the camera to a canon shop.

There are still issues with the port numbers changing and the cameras can have problems connecting to the gui after the gui has been restarted or the camera has been reconnected without being turned off/on.

[Gomismis]

@Baeist is the above mentioned test only done with the GUI or also with DigiCamControl?

[PipBrewer]

@PipBrewer, should I still contact NHM/London, or should I wait until we are further along in the physical development process? I have already noted several questions regarding their setup/development, that I would love ask them.

You are more than welcome to contact them and ask lots of questions, at multiple steps along the way, no need to ask

[bhsi-snm]

@Gomismis test was done with both, but Digicam control app needs a windows pc which is not available upstairs(you could basically use your own laptop) but the whole thing is set up with Mac for testing with Entomologui for now.

[Baeist]

Created another branch 4pis3canons that allows the guid to display 4 pi-eyes and 3 canons. The setting of the settings inside the gui were not updated so you have to be careful not to choose cameras/pis that are not actually connected when creating camera settings. Current setup works for top, side and dassco 0024 canon cameras and beetle, cicada, dragonfly and earwig raspberry pis cameras.

[Baeist]

Created another branch Test6Canons that displays 6 canons, one of which is the R10 that i think we have borrowed from somewhere. I "accidentally" changed the names of two cameras formerly known as Top and Side to their dassco names. This can be undone by connecting only that camera to the pc and using: sudo gphoto2 --set-config ownername=name , use: sudo gphoto2 --get-config ownername , to see if it worked. If you want to use any of the other branches again changing back to Top and Side as names is needed.