@DG said:

Things may change in the future, as more and more modelling becomes mainstream, but I think that this is a while away.

I keep saying that here and everywhere: There are only two competitive companies that really knocked my socks off so far: Sample Modeling (and Wallander Instruments). I put the last in brackets, because you have to admitt that it sounds not close to the VSL standard. But my amazement is less about the sound. It is about the usability! Being able to morph between every thinkable nuance of an instrument and use a ribbon controller for pitchbend and all sounds very natural! I just would like to get rid of this approach, where you have to program your music. I want to play it. For certain styles that works with VSL. But Sampling modelling instruments give you the ability to do you almost anything in real time. Just one Instrument loaded and a buch of faders, there you go! Not much editing needed!

This is what I dream of for VSL to become. I think it should be possible, because they already have

of these great, dry samples, MIR ... the rest is software development and ideas. I don't think the future of modelling (or hybrid, like Sample Modeling) is so far away! In my opinon, VIpro 3 just needs 3 vital features:

- a great sounding formant algorythm to make pitchbend sound real

- a convolution engine inside VIpro to morph samples (e.g.: nonVib - Vib)

- a realtime stretch engine

The compination of those would make VSL samples even more flexible ...

Of course, at some point you will be able to modell any instrument performance without using a single sample. But that is indeed something I expect in 20 years or more. In the meantime, I want to be able to play more and program less and I don't think, i will have to wait 20 years for that ...

@DG said:

Unfortunately the DVZ approach doesn't work. Recording violinists two by two does not sound like a violin section when put together.

I should clarify. I didn't mean DVZ in the sense of recording each violin, desk, and so on... but that whatever level of Divisi VSL produces... it should have some flexibility in the performance. Say you have Violin I and II both divisi and all separate performers, etc. That is at least 4 voices. Rather than have separate patches for loose and tight... one could slight a slider... and gradually split the divisi sections up into a more loose performance. A Divisi library would allow for such a simple approach, which saves time rather than having to adjust the performance on our own. As we all know, human performance means imperfection. DVZ's approach at this certainly is more effective, but the recording quality and overall sound is obviously lacking.

Think of my comment geared more at Dimension Brass. A slider to make the 4 trumpets less in time or more in time would have it's uses.. like a timing humanizer... but with Divisi strings, I think such a feature would practically be expected... at least by myself anyway.

Here's how I view the benefits of such a performance adjustment, such as how DVZ has implemented. These are the things I constantly change PER string voice... for a more human performance: Attack, Release, Start-time, and Pitch. VSL has addressed Pitch, but not the other three in an easier to control, more automated way. If I had one adjustment slider to 'stagger' the attack and release amount, and to stagger the start and release times of multiple selected instruments, this would save me a great deal of time in Cubase as it is... If I have a Divisi String library, I wouldn't just be wishing for this anymore, but would expect such a feature. I nearly expect it now... lol Otherwise, each and every last note in Cubase has to be edited VERY tidiously for all of these things to take place. While I'd expect it for Divis strings... I would love to have this now across different midi channels... whether this be a 'timings humanizer' or a slider for sections... or something.

-Sean

@Another User said:

I want to be able to play more and program less

I think that's what most of us want. I don't mind programming for style or shape of the performance, etc. I mind programming just to get what sounds like a real player played it.

-Sean

 I would just like to say that with this simple phrase " I want to be able to play more and program less" you have eloquently summed up what so many of us have been thinking for years. VSL provides the most wonderful resource, however so does Mercedes. The trick is to provide the driver who is not an engineer with a rewarding experience. Not an easy task when so many customers demand to "get under the bonnet" to tweak to their own preferences, and also resist the impulse to "dumb down" the product. Perhaps VSL really are doing all that can be done in this regard, but it would be helful if there was a sign above the door of the office saying " I want to be able to play more and program less". Point well made!

When you take away the ability to program and fix it so it can't be changed, you make it easier to use, but you also remove functionality. Sure, when there are things that can be automated but don't lose functionality, of course, that's why the VSL team keeps releasing software, to improve and expand on these functions.

In this case, VSL is not a mercedes but in reality, it's a modular, complex musical instrument in it's own right. There's nothing wrong with asking the engineers to make existing functionality easier, if that's possible, but there's going to be a limit where you can't tweak something to get a certain result anymore. Point being, for myself I would always in the case of VSL chose functionality over ease of use if there had to be a choice in any particular given example.

I remember a flight to LA, when by chance I sat next to a music developer. He said that orchestral modeling was imminent, and those with bloated, expensive sample libraries would soon be left holding obsolete gigabytes of unusable sound.  

It was 1995. 

I've bought a few now-obsolete sample libraries since then. But they were supplanted by better sample libraries like VSL. Orchestral modeling (synthesis) remains a work in progress. 

I am by no means against extensive tweakability at the cost of ease of use! It is just the philosophy some VSTi's I really digg (e.g. sample modeling). You have one patch, where everything is at you command with some controllers. Also, every aspect of articulations can be morphed via CC to another, whereever the real instrument could do just that.

In the end, only a very detailed physical modell could render all possible output of the real thing! But, until that will become possible, I would like to have access to this kind of technology. Even, if it is not yet without any limits. I share plowmans opinion, that it will still take a very long time, until we will be there (if we ever will). But that doesn't mean, we can't have something close to complete in the meantime. SM has proven that IMO. VSL could make something like that possible without even having to record a single sample more. This is what I wish for in the future in VIpro.

@mpower88 said:

VSL is not a mercedes but in reality, it's a modular, complex musical instrument in it's own right

That is a good idea, and very true about VSL.  The thing that makes it not become obsolete like libraries of the past or the quickie symphobia type stuff is that it is based upon building orchestral sound from the ground up.  The methodical approach and carefully, analytically constructed edifice of samples that VSL pioneered are based not upon clever technology, but upon fundamental musical notes and expressions.  That is why it will not become obsolete. It is authentic musical sound, and the system they have developed allows for continuing additions and improvements without invalidating the very first samples recorded ten years ago. They sounded great back then and still sound great, but many hundreds of thousands more have been incorporated into the instrument, making it immensely more expressive.  New programming approaches of Vienna Ensemble and Instruments and Dimension Brass allow new uses of these basic musical building blocks, but it is the intense focus upon pure, true musical sound that makes VSL so great.

VSL can indeed be easier to program while maintaining the same level of flexibility.

I and others have come up with great ideas regarding how. We have stated these on here, the VSL DAW thread, and others. Before I purchased VSL I'm sure others did before me, and so on. The point isn't whether VSL can make improvements, but concerns of user demand (which I think we all agree is prevelant) and developement (that darned money eater).

Humanize is a step in the right direction, but is more random than anything. It's not really random as one has degrees of control, but while reading on Numerical Sound (Ernest Cholakis's site) regarding his DNA Groove Templates... I realized that the humanize feature isn't where it could potentially be. Playing patters of individual players would be a more effective tool than what VI-Pro currently has. I have suggested a timing humanizer, that would adjust the attack, decay, start, and end times of each note played just so slightly as to create a humanized performance.

Vienna Instruments Pro, if radically upgraded (Like 3.0 or something) could have an entire screen tab full of humanize options. On the VSL DAW thread, many of us referred to ways that a VSL DAW or VSL Notation editor (or a combination of both, like Notion, etc.) could have features that would save 'programming time'. We even came up with ideas of how those could work. So again, I don't think the problem is flexibility vs ease of use. I think it's just a matter of whether VSL finds enough market in this issue and what they do about it. I absolutely think there is market. I think someone would be a fool to say otherwise. VSL has shown some more committement in the promise of Vienna Instruments Pro 2.0 but we just have to wait. Perhaps that will solve a great deal of this frustration. Perhaps only parts of it; maybe we'll even see more from VSL in the future. That future could be upgrades in VE, VIP, or by other means. We'll just have to wait and hope for the best.

I thought that Cubase has scripting abilities and possibly even plugins (not VST, but actual program feature or gui plugins). Am I wrong? Has anyone made any decent ones to save time on such issues? I know that I heard or read Hans Zimmer saying that Klaus Badelt had created several scripts for saving time. I have often wondered how and what and what I can do to save time myself. Perhaps some of us (myself included) just spend too much time commenting on here. [;)]

-Sean

@iscorefilm said:

I thought that Cubase has scripting abilities and possibly even plugins (not VST, but actual program feature or gui plugins). Am I wrong? Has anyone made any decent ones to save time on such issues? I know that I heard or read Hans Zimmer saying that Klaus Badelt had created several scripts for saving time. I have often wondered how and what and what I can do to save time myself.

Hi iscorefilm,

Just inspired by your post I have done some investigation in Cubase 6. Indeed there are some possibilties to implement a customizable amount of random values in whatever CC with the transformer feature.

I have included a very quickly made wmv with a nice impression of what can be done. I have randomized attack, release, humanize and filter settings here. You will hear a SE+ flute collection of 6 simple patches: staccato, portato, fp, sustain, rep_legato and rep_staccato. You can enjoy a beautiful ballet of four (!) sliders going up and down.

I have included three different SE+ instrument mp3 files with a better bitrate (320 kbps) for the flute, violin and bassoon. Just for better sound results and more variety in instruments.

Furthermore two images with the transformer setup for filter and release settings.

I had to add a controller lane with continuous changing values to keep the script going on btw. Not so much work (copy-paste): it just won't work with only one value at the start.

It is possible to set random absolute or relative (to what you have indicated yourself in a controller lane) values. Maximum of four parameters.

Personnaly I think and hope that these features will be presented within the VSL sampler itself, but for the time being it is quite OK if you want to have a more personalized VSL.

I hope you enjoy the very charming video.......[:D] and the rest btw.

Best,
Erik

lol, I love the ballet sliders! Thanks for looking into this. I'm not at my workstation to check if Cubase 5 has it... but I hope so. Nice feature. And yes, I agree that it would be nice to see this kind of thing in VI Pro.

-Sean

There are already a lot of controllable humanize functions in VI Pro.  But humanize functions are less effective than the randomness that comes into a performance during actual playing.  If you don't believe that, then A/B a keyed-in, manually CCed MIDI performance with a humanized notation file.   

What I really wonder about is the original subject of this thread.  If for instance, one took the exact same recording setup as Appassionata violins - same players, same positions on the recording stage, same microphones, etc. - and recorded 4 passes of 4-player desks (or however many the numbers would work out to be) for four divisi sections without changing anything, would it be possible to get a sound like the Appassionata full ensemble by combining them?  I know that it is impossible to do that with individual solo players.  But what about multi-player desks carefully recorded and combined without moving microphones or seating, etc?  It certainly works with the Dimension Brass, which sound exactly like solo vs. ensemble.  I have a feeling that this is what VSL is working on...  and if the automation of splitting can be done as well as it is done with the brass, you have the ultimate divisi strings.  However, as some people have complained about, there need to be an enormous amount of recordings done in order to duplicate enough articulations for full expressiveness.  If you have to choose between many different expressive possibilities without divisi, or divisi with only a few expressions, what will you do?  Decisions, decisions... 

@Another User said:

From my personal point of view I would speak more about an expansion than of an improvement.

I believe the question should be:

How could we come even closer to a real performance? If we ask this question we need...

1. some more articulations (min 4 layers)
2. most of the articulations should be available either with Vib or without Vib - if possible with "Sweet Vib" and  "Normal Vib" 
3. 4 levels with the legato samples (just as sus4Levl-perf - it allows nice X-fading sequences) 
4. more levels with the dynamic samples (3-4 for dim/cresc, 2(-3) for sfz, etc.)
5. Two different Violins
6. Two different Cellos
7. ...

Although I think Dimension would mean an improved blend and ANY new articulations and/or more bowing flexibility would be GREATLY welcomed... I think his list pretty well sums up the most important aspects of a future VSL string library. Perhaps, like their other products, and unlike their Dimension Brass... we will see a Dimension Strings expanded library. The more able to sound realistic, and more capable this library is... the more of a "I need this library more than any other" people will react to it. So in light of 'those other libraries', I doubt VSL is thinking much differently than us users are about what this library needs to be. And ultimately the point I wanted to make in that is that more divisi AND more expressive possibility are inseparable; most users seem to want both and will keep asking for it until both are had.

-Sean

Although I would welcome either additions to the library, using the 'horse behind the cart argument', I'd rather see a Violins II product first, since any Divisi derivatives would have to be developed separately for the two sections, otherwise the divisi between Vs.I and Vs.II would sound the very same and would need electronic tweaking (and they would still be the same, distorted). It's not like with the brass section which is not doubled in conventional orchestral writing.

Violin II would be nice. I'd welcome it and buy it. But Divisi offers more than just 'more voices'... but more voices for a less perfect playback. A dimension library would mean a better blend (depending) and so on. Violin II should have been out LONG before now if it was going to be done. Note, that the competition already has Violin II.... AND Divisi.... so at this point, while I'd welcome it. I'd also think that VSL would have made a seriously wrong move in doing so. They are already playing catchup here. I think users expect VSL to match and exceed the current expectations with anything new they produce. Vienna Imperial and Dimension Brass are absolutely that... So I just doubt they'd simply do a Violin II after release those two products.

-Sean

@iscorefilm said:

Violin II should have been out LONG before now if it was going to be done. Note, that the competition already has Violin II.... AND Divisi....

"Should" have been?  According to whom?  Zeus?  (Actually I would do it if he commanded.)

Violin II is a nice idea in solo strings (irrelevant in any ensemble string ensembles since you already have it unless you are too much of a moron to know it). If the solo  is slightly different in quality.  Such as a more relaxed vibrato and darker timbre. Then it would be worth sampling rather than merely pitch shifting and transposing (though as I know from having done it you DO get a difference in timbre by those means that is very effective).

However, the "competition" you mention is pathetic to be honest.  The depth of VSL recording is so far beyond any of the competition there is truly no comparison.  Even one library of VSL is beyond any of the others. And they have four!  What the other companies are all doing is trying to sucker people into buying their stuff with easy, quickie tricks.  Like it is "easier to do."  But all the "easier" things end up being lower quality in the long run. 

I do listen to these new libraries, and they are not comparable to VSL.  You need to realize one thing - VSL has been doing this with incredible focus for a decade now.  Every Johnny-come-lately looks cool - until you see their feebly limited articulations and hear their sound.   But many people are looking for instantaneous and crude sounds.  So they won't be interested in the difference. 

Nevertheless, the divisi strings would be an important addition to the VSL arsenal.  And new programming is making things easier to do all the time.  The technology behind dimension Brass is obviously similar to what is needed and can be used to develop a powerful divisi library.

@William said:

However, the "competition" you mention is pathetic to be honest.  The depth of VSL recording is so far beyond any of the competition there is truly no comparison.  Even one library of VSL is beyond any of the others. And they have four!  What the other companies are all doing is trying to sucker people into buying their stuff with easy, quickie tricks.

Look, I think Hollywood Strings mostly sounds like crap compared to VSL... imo anyway. I also agree that VSL is beyond everything else out there. Basic VSL is better than other libraries at their best, imo. But I'm not foolish enough to discount them as not being competition. I only make the comment in reference to what we'll likely be expecting from VSL. I simply doubt that it would be a Violin II, that's all.

-Sean

I'd die for a dimension strings library!!! 

If it would be available, it could finally be possible to control how "amateurish" the strings are playing, with modular internal tuning and timing on each player. It would be a huge step towards more realistic strings!

Another thing I have on my wishlist is to have the ability to xfade samples to silence with real recorded long release samples. Now I always have to combine velocity xfade with expression to make smooth releases. Anyone having any better tricks? I mean, the pfp-samples have fantastic real smooth attack and release, but no ability to play intervals. 

My third thing on my wishlist for strings is to have sul-samples for each string. For example for violins it would be sul-g, sul-d, sul-a and sul-e, with at least fast/slow legato and staccato samples.

I understand that a dimension strings library would be at least 10 times larger that the current strings library and eat tons of ram. But I know it would be worth it! 

//Felix

Seems to me like having divisi sections would require new full section samples, too. 

While I think LASS was keen in doing this, they did it from the ground up. Let's say I buy $5k in VSL string samples today...not buying a whole other $5k collection for divis sections. 

I think it would be a cooler "feature" given VSL's extensive sample collection to implement a faux autodivisi feature using maniuplations of volume and maybe timbre of existing content. Otherwise, everyone will have to be using the new sample sets to getthe divisi feature to work. And, if someone has thousands invested...divisi really is a subtlety, in the grand scheme.

I'm betting some of the tips in this thread could be rolled into the engine...take the lower of two notes played and use a sample or two DOWN and transpose UP (which would thin the sound naturally a bit) and low the volume of both by whatever percentage...even just make that an option, right? Whether you want the higher or lower voice "deemphasized".

Now, for the solo strings, which I own...they really COULD do a sample set of a second player/instrument of each. Same room, same mics. It isn't really even supposed to exactly match--that's the point, right? And that wouldn't be that expensive a session for them to have. It could be a "plus" add on for existing solo strings owners.

But, full divisi session means throwing away their current sample content. Not something I'd think is financially appealing to anyone. 

@popmann said:

implement a faux autodivisi feature using maniuplations of volume and maybe timbre of existing content. Otherwise, everyone will have to be using the new sample sets to getthe divisi feature to work.

First- I think recording and audio engineering principles make this impossible to make a reality.

Second- VSL wouldn't necessarily have to throw out the old in order to add divisi. I have thought of arrangements in which VSL could add a single dimension library that could both add a Chamber divisi and Violin II, or a Orchestral Divisi and Violin II (each with appropriate sized sections with both Chamber/Orchestral, while using the same exact patches, just organized differently per library). This would add Divisi and Violin II.

The only think I worry about is that other libraries are coming out. With what other libraries are doing, and VSL's history of pushing the limits, I think people expect VSL to not only have a Dimension string library... but one that has even more than what we do now. (Mute Spiccato... eh?) lol

-Sean