Layer stacking in portrait editing

Question

I have a brief question about a complicated issue that I am hoping someone can help me with as painlessly as possible.  If there are videos, books, tutorials or anything that addresses this matter I will be happy to purchase whatever it takes.  My issue is 99% layers.  I am a headshot photographer and have been using Photoshop for many years.  I’m therefore not a newbie to it or layers but I have some degree of difficulty with the stacking order sometimes.  I also edit for other photographers who know little or nothing about Ps and so the stacking becomes more of an issue for someone who didn’t actually edit the photo.  More to the point, I am trying to learn the best practice for ORDERING things so that it’s always possible to go back and tweak something from a lower layer.  However, my problem is that I frequently need to make a flattened layer (on Mac: shift+opt+cmd-E) and then do something FROM that point forward.  This, of course renders curves adjustments and the like useless insofar as changes you want to make later due to layers above it being baked in going forward.  I am also not entirely clear what adjustments MUST be on pixel layers and which ones can be on a blank layer.  If anyone has advice on this I would greatly appreciate it - OR - can just tell me where I might find instruction on this specific topic.  Lastly, if there is a better forum for me to post this to I will be happy to do so.
Thanks so much for your time and consideration.

Answer 1

Yes, 'stamp visible' is the same thing you called flatten layers (shift+opt+cmd+E); but it makes a new layer as opposed to just flattening what is already there.  Just semantics, we're doing the same thing.  The object is to do as much as you can non-destructively (without stamp visible), and it sounds like you're doing it when you don't have to:
Spot Healing: You can open a new blank layer and use the stamp/clone tool, just make sure to check "sample all layers".  This way the layer only has your edits, not a complete copy of the entire image.  Makes it easier to select changes, and reduces overall size of the document.
Frequency Separation has to have it's own layer(s).  Nothing you can do about it, which is why I do it first.  Just one note: if you are going to do FS, then you should decide whether or not you want to do spot healing before.  In most cases, at least for portraiture, you should do you spot healing in FS.  That's one of the major benefits.  If I'm just cloning out something in the background I won't necessarily do it in FS, since it's quicker outside, but any editing on the subject should be done in FS.  IMHO anyway. 
I'm not sure how you do Dodge and Burn, there are several ways.  But you can absolutely do it non-destructively.  I'd recommend either using a 50% grey method, or just white and black in a soft light or overlay blending mode.  Either methods don't require stamp visible.
Eye and teeth whiting also have several methods, many are non-destructive.  I usually do this in Lightroom when I convert from RAW (I believe you can also do it in the new ACR? in Photoshop).  But if you're using jpg or want to do it in Photoshop it's just as easy, and non-destructive.  Use a combination of curves and hue/saturation adjustment layers.  I prefer using curves for the lightness, as it gives me more control, but the hue/sat has a slider for lightness.  If they have yellow teeth I reduce yellow saturation as well. 
All color and tone should be done non-destructively, through adjustment layers.
Sharpening depends on how you do it.  Again, I do it in RAW conversion.  The only thing I do in Photoshop is either a highpass filter and/or I can do it manually within the high frequency layer in F.S.  I often prefer to do it as a separate high pass layer at the very end, because I don't like too strong of an effect and I get a better idea of how much I want at the end of the edit.
So, to sum it up, really the only stamp visible I use is for frequency separation, and I do it at the start.