I'm having similar problems with music versus movies (mp3/m4a versus mkv) ... e.g., comfortable volume levels for my AVR are approximately -40db versus -20db respectively (i.e., movies are more attenuated).
The mystery is ... if I apply volume leveling for both (presumably by analyzing audio for both music and movies, and enabling DSP volume leveling), why are not the volumes more alike? Is there a way make the volume adjustment different with respect to format?
There are a few issues here. Firstly, you have Adaptive Volume set to Night Mode, which is a dynamic adjustment and counteracts any Volume Leveling.
You need Adaptive Volume set to Peak Level—or preferably disabled—for leveling to work correctly. This will be significantly quieter than you currently have, but it is required for good leveling.
Secondly, films often have a greater dynamic range than music, which often exceeds the headroom that volume leveling has with the current target of -23 LUFS. For films to be properly leveled, you would need at least -30 LUFS.
Thirdly, album-based leveling was changed to use the loudest track on an album, rather than the average level for albums—this means that
most albums sound too quiet now, and are no longer level when they are part of a mixed playlist.
Fourthly, when downmixing multichannel audio (which is common with films) to stereo—though it seems that you are not—Media Center can only do a very rough approximation of the volume level.
I feel that Media Center really needs to have an option to create a downmixed analysis for videos, because it's just not very even right now. It's better than nothing though.