Orchestral sample libraries are not exactly like the real thing. There, I said it! They are snapshots of the real thing, like the difference between an attractive person and a picture of an attractive person.
However, they may be all we have to create orchestral style music, they are great learning tools, and sometimes, while there are things they will never do well that real players can, they can be made to do things real players either cannot or would not.
So assuming that your goal is to make your MIDI-based orchestration composition stand out, and unless you are trying to mockup a Classical work, your focus in my humble opinion should be to do whatever you can to make it sound great, even if sometimes it makes it sound less real.
This is highly subjective and some of my ideas are certainly open to disagreement, so I put the question to members of a couple of forums that I participate in, and got some interesting responses. Here are my favorites:
Nils Neumann: "Don't copy and paste, especially if you are layering instruments"
I already had that as one I was going to include, as it is an article of faith for me. Additionally if you must do that because of a time crunch, use a "humanize " feature if your DAW has one. Otherwise, you end up with your sampled orchestra sounding like a giant accordion.
Rodney Money: "To trick the ear that the music is quieter, use smaller sections such as horn a2 for "quieter moments" and larger sections such as horn a6 for louder ones."
In all honesty, that had not occurred to me. With East West Hollywood Brass for instance, there are 6 French Horn patches and 2 French Horn patches, as well as solo French horn patches. In my template I have the 6 French horns and the solo, but not the two. I am going to rethink that.
Dave Connor: "A sustained tone should always be either rising or descending in volume. The rare exception being a deliberately held chord such as in a loud or soft brass chord where the idea is that it should sustain at a particular volume."
This is especially true with samples, when their volume is static, they just have no life. Woody Allen has a joke in “Annie Hall”: “A relationship is like a shark, it has to be constantly moving. What we have here is a dead shark.”
Don’t let your samples be dead sharks.
Karmarghh wrote: “A big turning point for my compositions was when I really started to pay attention to variation in the tempo track. In the right context adding a dip or increase in tempo at the end of a phrase can completely change a piece.”
If your DAW allows it, I find that automating the tempo slightly in real time gives it a human quality that makes a g difference as well.
I like this one from NoamL: Line up your transients. Bounce them to audio if you're not sure. Every developer seems to have a different philosophy about exactly what "attack" means.
Storyteller echoed something I have written many times. When using EQ, take the Hippocratic Oath and “first do no harm.” He wrote: "Don’t turn knobs because you think you are supposed to. Just like in studio recording, if the engineer has done his/her job and has used the right signal chain (e.g., preamps, cables, microphones, A/D converters), very little to no EQ will be needed. The same goes for professional sample libraries. Assume little to no EQ is needed unless it is intended to serve a purpose such as to create a specific sound (such as "bombastic" percussion) or to match sample libraries. The exception to this rule is the use of the low-cut filter (also known as the high-pass filter) to remove unwanted buildup of the lower end frequencies of the noise floor."
I do, however, think that people also overdo the cutting of the low frequencies.
When it comes to editing, here is some good advice from pmcrockett: “There are no bar lines; there is no grid. Edits should be made based on how they sound, not how they look.”
I like jsg’s comments about balance and transparency: Balance: Does the texture as a whole work, is any frequency range too loud or too soft? Do the instruments added together create only volume, or do they also create a pleasing tone color?
Transparency: Even in a dense, fully orchestrated passage, the orchestration, voice-leading and counterpoint should be able to be heard, in other words, the melodic motions of both inner and outer voices should be clear.
There are some situations where this may not be desirable, but usually it is.
The great Hans Zimmer said this: "Don't even think about doing the drums until you've written everything else! Fit them around the music. Once you cover everything in drums, there is no more room for inventive writing and your bass end will become problematic... I always put the percussion in last, but I write and orchestrate everything else knowing what percussion I want to use and intimately knowing my sounds—and what I want the percussion to do.
Try doing things like doubling your digital sounds with analog sounds, like a synth patch or a library like Solid State Symphony form Indiginus. I actually did this back in the early 90s with real string players scoring the TV series “Zorro”, doubling them with a Memorymoog because it made the smallish section sound bigger and warmer.
Composer Craig Sharmat aka Scoredog taught me this one, and I paraphrase: ”If a pre-recorded run does not have the notes you need but you want the “sweep” that you cannot get playing in your own, play in your own and mix the pre-recorded run softly under it. In a full composition you won’t really hear that the notes are not correct and it will give you the sweep you want.”
This works great for woodwinds, strings and even harp and I have used it many times. Add processing plug-ins that a real orchestral mixer might not. Go for it. If you don’t like the end result, unlike those recording and mixing sessions with the real guys, it isn’t over until you decide it is over.