Thanks a lot for the logs! I've only now got to properly look at them due to recently fighting a nasty flu-like illness and general lack of time.

Surprisingly I haven't found any direct evidence what might be wrong - the logs are pretty similar to the output I get on my Jolla, but there are more tiles on the screen (1080p vs 720p) and the actions generally seems to take less time (faster CPU - race condition potential ?).

It also seems that the tiles model is working correctly - there is the correct number of tiles in the model and the web Mercator coordinates of the map corners also generally look sane. There are even some tile debugging calls in one of the log files that confirm the corresponding tile images do exist and are being loaded - they are just not shown for some reason!

I've also checked and the issue is not just due to the 1080p resolution - I've created a 3000x3000 pixel local VNC server screen and run modRana in 1080p (both portrait and landscape) on it and the issue did not manifest. Also as mentioned earlier, it does not happen on the Jolla Tablet, which has even higher resolution.

So I'll continue studying the logs and will probably add a few more commits to that debugging branch, mostly adding more logging and maybe also trying some stuff, like:
- setting allowed window orientations, there are some error messages about that (I don't have those warnings on Jolla C)
- logging tile screen coordinates (to check the tile coordinates are correctly translated to screen coordinates)
- disabling asynchronous tile model updates (to remove the possibility they are causing this)

I will certainly let you know when I'm that far & will again need your help with running that branch again on your affected devices. Thanks in advance!