As I understand it, getting a good approximation of the orchestral "sound" in the digital world has to do with setting the volume (gain) of each of the instances in relation to one another so that their respective dynamic ranges play realistically against one another. For example, a single woodwind should be set at a volume so that, even at it's maximum dynamic level, it never exceeds that of, say, a Trumpet or French Horn. So, the "Natural Volume" feature - again, as I understand it - is there to use as a starting point from which to make adjustments as needed towards the desired balance between all instruments. As far as I know, there's no "scientific" way to do this and it just has to be ultimately a judgement of the ear, taste, and the desired soundscape.
So, my question:
Once the balance in volume between the instruments has been established to the desired amount of realism (so that none of their dynamics is out of place with the other), is it good practice to then set the volumes as a whole to a point where, when the entire orchestra is playing FF tutti, the meter is something like -6db below 0? My worry is that gain reductions in dB are not on an even scale (unless I'm wrong??), and so reducing, say, a woodwind's volume too much to avoid a tutti FF section clipping will result in the woodwind being waaaay too quiet, but if I raise the woodwind, the balance will be out of whack.
Of course, I'll be fiddling with all this and experimenting, but I'd be over the moon if any of you vastly more experienced mixological wizards out there would be willing to impart some of your wisdom :-)