I've been using MMultiAnalyzer a lot to EQ my tracks for better separation, and what I've found is that the most common areas it's collision detection feature alerts me to are the areas around 500Hz and 1kHz. Give or take a hundred Hz or so, these are the regions which seem to require the most attention. Is there any particular reason why this should be? Has anyone else noticed this? Like I find I'm frequency cutting 500-600Hz in instruments to make the bass cut through a little more.

The 500 to 1KHz region is important because it is usually where the vocal or melody resides.
It can help to remove a little of that frequency range from the instruments to keep the melody on top.
It depends on the song and arrangement. Sometimes, removing that frequency range from the instruments doesn't sound good. In that case the arrangement needs to change. Example: Change the piano to either a lower or higher octave so it doesn't clash with the melody/vocal, or scrap the piano part completely and make a new part.
 
I have had a similar problem with direct recorded bass guitar. The mix sounded great on big speakers but the bass was weak on small speakers. The solution was to add a 500 Hz bell curve to the bass. Adding 500 to the bass sounded better than removing a little 500 from the instruments. It depends on the instrumentation and arrangement. Try it both ways and let your ears decide what is best. 
 
 
 

This isn't exactly a scientific answer, but pseudo-scientific...

Every now and then I rediscover this handy little chart. It's been floating around for a decade, so perhaps you've seen it:
http://www.audio-issues.c...ation-youll-ever-need/

Lots of instruments have their fundamental frequencies in the mid hundreds - some of them you might not even expect, like cymbals.

If my memory is serving me correctly, many instruments (eg brass) also carry much of their acoustic power in their second harmonic. Bass really relies on second harmonic to cut through, in my experience. And that's right there in the mid hundreds too.

Once you start throwing in guitar amp distortion, you get all kinds of sidebands, too, and the mids get very crowded indeed.

For me, it's more likely to be 400 Hz and 2 KHz, but the principle's the same. As Mike notes above, the phenomenon is simply due to many tracks contributing to the same 300-500 Hz range. 
 
Consider acoustic guitars and pianos, for example. Both were as designed to be full-range solo instruments. Which is fine if you're playing guitar around the campfire or piano in the parlor, but when thrown into a dense rock mix you have no choice but to thin them out to avoid spectral crowding.
 

Yeah I find I have to cut a lot out with pianos. When detecting frequency collisions with bass there's always a very dense collision around the 500-600Hz mark and I always feel there's some good stuff in the bass there so I cut some out of the other instruments. I think it makes the bass "nose through" the mix a little more. 

only cut what sounds bad.
 
maybe those frequencies actually sound GOOD on certain parts,
and a blanket rule simply whimpifies some of your best tracks.
 
'collision' detection....
there's a real mind phuck of a term.......

batsbrew
only cut what sounds bad.
 
maybe those frequencies actually sound GOOD on certain parts,
and a blanket rule simply whimpifies some of your best tracks.
 
'collision' detection....
there's a real mind phuck of a term.......

I only use Collision Detection on tracks I feel could use some separation, and if it tells me that there are a lot of collisions around 500Hz I will make a cut around there on one of the tracks, and it definitely helps distinguish them both. I wouldn't make the cut in the same place on both tracks.

For me it is more like 300 Hz and 2KHz and above.  But yes there is a lot of information around 1KHz sometimes.
 
But we are still not addressing the real issues.  These overlapping problems tend to exist because the parts are over playing.  If two parts are always overlapping each other then the better place to sort these types of problems out are in the arrangement itself.  Why not ensure some parts are not playing when other parts are.
 
Parts can weave in and out of each other much more so.  They don't have to overlap.  Listen to Steely Dan for the ultimate in economical parts. (Even in the most complex of their tunes you can still walk right around in the mix and touch the black backdrop behind the music!)  Then when you get this happening you don't have to EQ the crap out of something to make it work.  In fact you can leave it alone and allow it to sound fuller and better.  Even add to the sound by making it sound fuller again and then after that you can turn it down a little. (Maximum illusion, minimum voltage)
 
Several parts can come together to make an identical statement too and that is powerful as well.  Then it is more about balancing them (for that section only) volume wise and not so much eq cutting.
 
When there is too much going on and too may parts are playing at the same time then more of this EQ cutting crap has to used in order to make it sound OK but then you end up with a whole lot of thin sounding parts.  And the result is still unclear.  (because there is still too much going on !)
 
What many of you don't realise is that if any of you had the opportunity to work with a really great producer on one of your own songs what they would do first is cut half of the stuff you guys have tracked right out leaving space and only the real important stuff behind.  People are too scared to cut stuff out and leave space behind. They think it does not sound full and is weak but the opposite is true.  It sounds fat, strong and clear.

Thanks, Jeff, for another very enlightening post. I really enjoy your detailed to-the-point explanations.
 
Having the arrangement take care of the mix is a great, but unfortunately takes a lot of time/pre-production and also convincing people. Just last night I had a drummer eventually realizing (after the 3rd round of "downsizing" his groove) that if he completely leaves out the ghost notes on that verse, it doesn't get thinner but the opposite is happening as now the subtle rhythmic throbbing of the synths was no longer masked by his ghosts and snare rattle and the whole section developed different dynamics and feel ... so I reckon it applies across all instruments and not only to the studio as this was during a rehearsal session for a live show.

sharke

batsbrew
only cut what sounds bad.
 
maybe those frequencies actually sound GOOD on certain parts,
and a blanket rule simply whimpifies some of your best tracks.
 
'collision' detection....
there's a real mind phuck of a term.......

I only use Collision Detection on tracks I feel could use some separation, and if it tells me that there are a lot of collisions around 500Hz I will make a cut around there on one of the tracks, and it definitely helps distinguish them both. I wouldn't make the cut in the same place on both tracks.

 
sharke, i kinda figured that,
but thought i'd put it out there for anybody who hasn't really gone there before...
 
there are so many dumb ideas being floated out in the interwebs, some people don't do their homework,
but this is actually a GREAT idea when used properly, exactly like the way you described.