I've tried yesterday a couple of RPI4 4Gb setups (Astroberry, Astropi3 script) and noticed one thing in PA routine. After pressing the Refresh button, I can see this image and it is refreshing slowly. So to be honest it is extremely hard to use it.
I didn't have that issue with Ekos on my MBAir.
Is this caused by RPI4 limitation? Maybe a hint about a camera setting that I've accidentally changed?
PS1. This is not a network issue, I am sitting right next to RPI4 with my iPad and hotspot.
PS2. Yes, my camera had Ha filter that moment and 2000 gain but that wasn't an issue before. Also tried without filter and have exactly the same result.
I suspect that everything on the RPI4 is slower than on the MBAir. I use a QHYCCD QHY5-III-178C for my guide camera and use it do PA on my RPI4 4GB with Astroberry. You have Update PA Error checked which can be very compute intensive so the recommendation I have seen is to use 2x2 binning to reduce the image size. Looks like your camera may be 9M pixels while mine is 6M pixels. I use 2x2 binning with the Update PA Error checked at a 0.5 second refresh and it is usable. Not amazing but usable.
I'm leaning towards to that too.. I do believe that this is a performance issue from RPI. I would definitely try to play with binning and refresh rate to see how it goes.
Refresh uses the star-detection code, which will take some time. I think it is pointless to set refresh to 0.5s on a RPi4. I'd set it to 2 seconds and bin either 2x2 or 3x3, and the performance should be usable. If you have a much faster cpu and usb transfer speeds, you can go to faster parameters. I don't think you lose any accuracy with 2x2 binning. You have to set the binning at the start of the polar-alignment procedure--you can't switch it in the middle.
I’ve tested with 2x2 and it was better. I wanna just to verify , I saw on another thread that you suggest for plate solving the 4.small scale ? In addition to that, will it make any difference if we have installed only the required astrometry files for our FOV? The algorithm runs through all the files, or it goes directly to the correct one?
I think any of those binnings would work, I suppose I'd start with 2x2 if it was quick enough, otherwise 3x3 etc.
Changing binning shouldn't affect the required astrometry files. It's the same field-of-view no matter what binning.
The plate solving is handled by StellarSolver, and it is a "black box" with-respect-to Ekos. You can use Astrometry.net's library, or you can use ASTAP. In any case, I think the scheme used to search files may depend of resources, e.g. does it have enough memory to search in parallel, etc.
If I were you, I'd be generous with the files you load. E.g. if you have a field of view of 1.5 degrees, load at least down to 0.5 degrees or something like that. It seems like the files don't start getting huge until you get down to very small field-of-views.