this still has to do with the old issue of clipped recordings. when they're clipped the dynamic range is compressed = the rms goes way up. this means that two sounds identical excluding dynamic compression, played at the same volume level, will sound like one is louder than the other (then one with less dynamic range/one with compression/one with clipping). once the dynamic range is shrank, its gone. compression is easy and common, but they're are no good expansion algorithms out there.
another key point is what aztec, and most everyone else, have such a hard time expressing. that they're are a lot of minute cues within the dynamics of sound, which humans have come to rely on. probably the most important of these is spatial information.
lastly, the footsteps are most likely related to this dynamic compression also. when they're are two sounds at the same time with enough differing in their "loudness", the lesser one simply isn't heard. mp3's and other lossy encodings take full advantage of this (among other psychoacoustic effects) by eliminating the lesser volume sound entirely.
the solution: as I've said before, records sounds at a low enough level as to ensure no clipping. then normalize them all (individually) with software. after that, take a listen and use minor dynamic compression on those softer than the "loudest" one of the group to make them all sound similar in volume, while retaining as much dynamic level as possible.
one method of alleviating a lot of this postproduction work is to standardize your recording method find a distance and placement that has proper tonal characteristics (gas/noises vs. barrel resonance, etc) and stick with it through all the guns, especially the enviornment. the easiest way to contain all these variables is to setup the mic and gun on separate stands/mounts, position them, and simply switch out the guns throughout recording.
good luck.
