Is this a live orchestra or is it built from samples?

Sounds like it was done with samples, though right where it gets "epic" the sound is so distorted that it's hard to tell. And if it was done with real players then the life was edited out of their performances. The opening sounds way too precise timing-wise to have been played by real players. Also, if you listen to each of the big low brass notes they all sound exactly the same. Listen to the attacks and the tone as they crescendo. To my ears, they're identical.

Is it far-fetched to think that a trailer/cue like this or any other could be done entirely with samples? Not at all. It's done all the time. But rarely do sample-only orchestrations ever sound 100% like a real orchestra. That's not to say that they can't end up sounding good. But these days, "good" and "realistic" can be mutually exclusive. If you can achieve both then that's all the better. But most people don't understand what it takes to get samples to sound real.

In the case of that trailer, the pizzicati in the beginning sounds quite keyboardy and the tone is extremely consistent. My vote? Samples. Now, that's not to say that it doesn't sound cool, but it definitely doesn't sound real, at least not to my ears.

As far as composing that kind of thing, well, it's more than just about composing. It's about orchestration too. And it takes a certain kind of audaciously creative mind to write music like what you hear in that trailer. Imagine writing a piece that builds as it does without changing the key or the bass note for a minute and a half...

Yeah. What he said.

Christian

Isn't it about now that we recommend "The Guide To MIDI Orchestration"?

Thanks for the star e.

That overall sound is as much about the engineering, reverbs, etc. as it is about the samples. I wouldn't be surprised if the composer had some custom libraries as well (though I couldn't tell you with certainty that, for example, the brass came from Library A or the pizz from Library B).

Could it be done with EXS-24? Sure. Not with any of the stock Logic sounds, but there are certainly libraries out there which use EXS-24. Fidelity-wise I find EXS-24 to sound a bit dull compared to other players and sample collections, but nothing that some EQ can't fix. That aside, EXS-24 is a really great sample plug for orchestral stuff because of all of the real-time control you can have over various sound parameters, including aspects of sample playback itself. It's definitely not as complex as Kontakt but it's very easy to program.

On the flip side, some of the higher-end libraries' playback engines offer very little control over sample playback. "Play" comes to mind. Even VSL, despite all of its controls, has an awful sounding filter and no control over the mapping or start time of the samples. Such lack of control can be very limiting at times.

And so, part of the reason that orchestral realizations don't always end up sounding quite right is due to of a combination of all of the above, and then some: lack of available controls in the player; lack of understanding of how to phrase instruments in a realistic way, or, simply not enough time to devote to tweaking the sounds (despite having the knowledge of what to do). Other problems in creating realism have to do with people not having access to multiple libraries. This is an absolute necessity, as no one sample library can serve all styles, tempi, requirements for the size of each section, etc. Just like guitars, every library has a "vibe".

If you listen to the demos from any sample library I can pretty much guarantee that you'll hear something missing and/or wrong with them. And it's not because the people writing them aren't good at what they do. It's just that their hands have been tied by having to use a single library to write the demos. So they end up writing "to the samples" and even then, sometimes the realism of their phrasing fall short.

The key to choosing libraries from online demos is to try your best to listen beyond the compositions and focus on the sonic and emotional qualities of the individual instruments & sections, keeping an ear out for what appeals to you, and not being swayed by the brand name. For example, you listen to percussion section demos from Library "A" and "B". Everything in the "A" demo sounds great except the timps (weak and unfocused) and the xylophones (harsh and "ticky"). However, in "B" the timpani sound perfect and the xylophone sounds just right (full bodied and not too harsh on the attacks). Bingo, you need both libraries if you're planning on writing for percussion section.

Stuff like that.

OMG! What a long answer!! LOL!!

iSchwartz wrote:
Thanks for the star e.

That overall sound is as much about the engineering, reverbs, etc. as it is about the samples. I wouldn't be surprised if the composer had some custom libraries as well (though I couldn't tell you with certainty that, for example, the brass came from Library A or the pizz from Library B).

<snip-snip>

iSchwartz,

That's such great advice to anyone learning MIDI composition.

A few years ago, (more than a few) I heard this incredible demo... I tracked down the composer, an L.A. guy who had been working with MIDI since the late 1980's, he sent me the .MID file to look at. But what really blew me away is the whole demo was done on an SC-55 Roland Soundcanvas, recorded as audio tracks with added EQ, reverb and compression. In the end the sheer musicality overwhelmed any shortcomings of the sample module. As a "whole' the composition worked and for him, that's all that mattered.

pancenter-

Thanks Pancenter. Great story about the SC-55 demo.

Not to go on too long about realism, but I found an online demo that's a good one to critique. Go to the following link and listen to demo number "4.2 String Orchestra Descen" of this Mirsolav Vitous string library:

http://www.bigfishaudio.com/detail.html?4;16;1:115::511988:::::G511988;A115::511 935

The purpose here is not to tear this demo to shreds. I rather like it and I think the sounds are really nice. My purpose is just to illustrate a few of the finer points to be aware of when doing orchestral realizations using samples.

As the demo progresses you'll hear a triplet figure in the violins. Listen closely and notice how the melody kind of ducks out of the picture during the second two notes of the triplet. The phrase sounds "mushy" because the attack of the samples (or perhaps the attack setting on the player) is too slow to articulate those notes clearly at the speed they're played.

After the pause you'll hear the triplet played a second time. The overall articulation of the notes is awkward there as well. In neither case do those phrases sound like they're played one one bow (legato). They don't sound like they're played detached either. It's somewhere in-between, and it's not an authentic sound.

How to fix it? One solution there would be to articulate that phrase with another instrument which provides an authentic legato sound. It could be from the same library or a different one. Whatever works. This part could be an overdub, leaving the MV sound in the background (which might lend some consistency to the overall sound). Or, the more authentic legato instrument could be used to replace the notes entirely in just that phrase. What works best is purely a matter of experimentation.

Next... listen for how some of the notes ring out in the last chord played before the pause. There's an Ab below middle C in particular that seems to end later than the other players in the ensemble. This makes the ensemble sound "loose", unconducted, as though the violas (?) weren't listening to the rest of the players.

How to fix it? One approach would be to go into the piano roll editor and edit the length of the notes so the result sounds like all players ended the phrase at the same time. That doesn't mean that you should make all notes the same length. A little "slop" on the end of a phrase is OK. But that viola (?) note's length should definitely have been made shorter. Easy fix.

Overall, the release time sounds a little too long. While a long release time helps to connect notes and play legato phrases, if it's set too long the overall sound starts to become mushy. Like reverb but without the characteristics of a room. I'd have shortened the release time a bit in the player, and then added some reverb to help connect the notes rather than depend on the longer release time to "glue" the section together.

So anyway, those are just some of the finer points that, to me, make the difference between, say, a sketch (in which it's fine to throw up sounds, compose, and tweak later) and a "final".

iSchwartz wrote:
Thanks Pancenter. Great story about the SC-55 demo.

Not to go on too long about realism, but I found an online demo that's a good one to critique. Go to the following link and listen to demo number "4.2 String Orchestra Descen" of this Mirsolav Vitous string library:

That pretty funky (as in not so great) for a product demo don't you think?

Sounds are ok but it's really mushy, almost like it was being played live by a couple of musicians... reading! I wouldn't mistake that one for the real thing.

The evaded cadence at the end is pretty cool, sounds like it's going to resolve to Cmi but ends up going to a type of minor iv (Abmi6/9?) which is similar to the preceding chord's upper structure.

pancenter-

@ pancenter -- agreed, probably not the best demo in terms of showing a polished result, but I thought it showed off the sounds in an honest way. I liked that chord progression too. Something about those samples I really like. They have a quality that I don't have in my other collections.

@ eMagnus -- depending on the type of project you're working on, you may or may not be required to be the composer AND the person who does the finished orchestral realization. Lots of scenarios. Here are two extremes:

• you're hired to score a low budget film which calls for an orchestral score. But there's so little money in it that you can't afford to hire a mockup guy to finesse your sketches. In that case it's all up to you. Same thing if you're an orchestral composer, just starting out, and you need to make a demo reel for yourself. You don't have any money to hire real musicians or even a mockup person, so it's all on you to learn the skills necessary to realize your music using samples.

• again, you're a film score composer working with a real budget but you're on a tight deadline. You write sketches and hand them off to an orchestrator who then hands off his charts to a mockup guy to create the demos for presentation to the director.

But the level to which you need to impress a potential client (or a director) with real-sounding orchestral music can vary wildly. For example, I recently pitched for a gig by writing three orchestral pieces. Made them sound as real as I could, very emotional and interesting, blah blah blah. The reaction from the potential client was a bit astonishing: "sounds too much like John Williams". I wasn't sure if I should be flattered or annoyed (since the style of my music is nothing like John Williams's). So if your client isn't sophisticated enough to understand the difference between John Williams's orchestral music and any other, it's likely that the degree of realism you put into your mockups won't count for very much.

Still, my feeling is that when it comes to expressing your music that you need to be true to yourself first. And if there's a great deal of feeling in your music, that alone can sometimes transcend whatever skills (or sample libraries) you might lack. The story that pancenter related is a great example of having skill in writing moving music without the benefit of a variety of sample libraries. On the flip side, there are guys/gals out there who have tons of libraries but write boring-sounding music. And everything in-between.

Uh oh, another lengthy post. Gonna go now.

I'd like to second the original request for recommendations of good sample libraries for strings (or perhaps techniques for getting the built-in orchestra ESX samples to come to life).

I've had two tracks in the last 6 months I've produced for friends that ended up needing realistic string sounds. The first one was a metal opus that needed huge sounding orchestra stuff. I managed to get through that with just the logic built-in samples, by adding huge reverbs and compression. In the end all the metal madness drowned out a lot of the really obvious "hi I'm some fake strings" tells in the samples.

I'm not having so much luck with my current project which is a more intimate singer/songwriter thing that needs some real exposed, organic sounding cello textures and some string quartet stuff in places.

I downloaded NI Session Strings while it was half off last week. That seems to have some really nice articulations that the built-in ESX stuff doesn't have, so that's helping.

But there are two things that are frustrating the heck out of me, and I wonder if some of the gurus here might have some suggestions.

A) score view won't show more than one track
I have everything separated on separate tracks. Two cello tracks, 2 violins a viola and also a clarinet track.

Score view will show only the selected track. Which is tough when you're composing, and want to line things up, or have a look at how you're putting together chords and stuff.

Is there a way to get score view to show multiple tracks simultaneously?

Back when I was in college (like almost 20 years ago now), we had a toaster mac midi'd up to a DX7. We ran a program called EA Music Construction Kit (they hadn't even started with the sports video games yet). That thing was SO handy for composing multi-instrument stuff because it threw everything on the screen at once and you could reach right in and pull just one instrument's notes around and visually see it in relation to the other instruments, and you could audition it right then and there. In all other respects, it was brutally primitive though.

B) endlessly dinking around in midi piano-roll view
Going in and manually set the veolcity on each and every note is just about the only way to get the instrument to "breathe". Setting individual articulations on each note in a quick phrase is about the only way to make the samples sound plausible. This takes FOREVER.

I swear to God, at this point, I'm really thinking to myself that these lines I'm writing aren't that complicated, and it might be quicker to find a music store in town that will rent me a cello and a violin and just learn how to play them enough to squeak out this real basic stuff in front of a mic.

ahicox wrote:
A) score view won't show more than one track

I have everything separated on separate tracks. Two cello tracks, 2 violins a viola and also a clarinet track.

Score view will show only the selected track. Which is tough when you're composing, and want to line things up, or have a look at how you're putting together chords and stuff.

Close all existing score editor windows. Select the regions for the different instruments you want to see in a score. Now open the score editor and Logic will have automatically created a "score set" showing you the instruments of the selected regions. Give that score set a name, and make sure that when you have a score window open that that score set is selected.

B) endlessly dinking around in midi piano-roll view
Going in and manually set the veolcity on each and every note is just about the only way to get the instrument to "breathe". Setting individual articulations on each note in a quick phrase is about the only way to make the samples sound plausible. This takes FOREVER.

Yes, it takes forever. That's a given when doing this kind of thing. If you want realism from sampled instruments the amount of tweaking time can really start to add up. The only thing that can save you time is spending more time: getting to know your samples better, and learning how to play them so that the amount of editing you have to do after-the-fact is minimized.

But I would suggest that adjusting velocities isn't going to make your music breathe. Rather, it's going to set the initial volume (and/or tone) for each note. Velocity is a static parameter. Once that note is played and held, the velocity can't then be made to change to impart dynamics to that note.

So in terms of tips on the velocity front: adjusting velocities can be helpful to make a line sound more consistent level-wise from note to note. And yes, varying velocities can help simulate a crescendo or decrescendo on fast passages (notes gradually increase or decrease in velocity with each new note played). But again, once that note is played, the effect of velocity is done with.

That said, not, all sample libraries are going to respond to velocity by making notes louder/brighter or softer/duller. Some libraries use velocity to control the volume of notes AND switch samples. So in cases like this you have very little range to adjust velocities to control the volume level of the note. Outside of that range you run the risk of switching to a sample that might be too out-of-character for the line you're writing.

And some libraries don't control volume or tone at all. Take, for example, Garritan Personal Orchestra. Here's a library where the dynamics are created by moving the modwheel up or down while you hold a note. Velocity, quite strangely, controls how fast or slow the artificial (enveloped) attack time is. That aside, because of the what the modwheel does, you can hold a note and impart actual dynamic phrasing to that single, held note.

Anyway, there's some food for thought.

Edit to above: where I wrote "And some libraries don't control volume or tone at all" I meant to say that some libraries don't let you control volume or tone +with velocity+.