I looked through the code and I'm quite certain that you're right, the values given by the internal guider don't take binning into account.
Will fix at some point in the next couple weeks if someone doesn't beat me to it.
I made the same observation. With 2x2 binning I had RMS values "too good to be true". But for 2x2 binning to work well, I also had to set the proportional gain much higher than recommended (2x - 4x). Could that be affected as well?
Yes, Guido. Makes sense. The 2x2 binning guiding had error calculations that were too low, by a factor of 2.
This is now fixed in the source code and in the nightlies. phabricator.kde.org/D29426
If you had come to like a certain proportional gain with binning greater than 1x1, you'll likely want to change that once you start using this fix.
It actually confused and frustrated me quite a lot, that I had those incredible good results with a QHY5 at 2x2 binning, but could not get an ASI 224mc at 1x1 to guide at all. It just went all over the place and aborted. Now it's clear that the proportional gain was much too high, as I was using the settings from the 2x2 binning.
I only got suspicious recently when I started to play around with PHD2 guiding. There QHY5 and ASI224mc worked fine as expected. With this experience I got back to INDI's internal guider and used the default recommended proportional gain values and it worked with both cameras at roughly the same RMS performance only 2x higher than I was experiencing with 2x2 binning.
As a side note: PHD2 performs a bit better than the internal guider for me, so I am using this for now.