Your observation about the problems with high string registers is very perceptive and on the mark. EWQL has multiple microphones, but what's the use of a room mic that captured the high strings at forte? I STILL think that the best high strings (and low strings) I've encountered are the old East West Ultimate Strings Collection. I converted the Roland format disk to soundfont. It doesn't have the 2000 different articulations, release tails and legato scripting but I don't care about that stuff as much.

VSL is the best.

What you recommend when i need frech horns without vibrato ? i just learned miroslav has always vibarto in french horns.:( and i heared GPO french horn not very common type sound.:(

Another thing to consider is that in a quick mock-up all instruments attacks start at the exact same moment. In a real orchestra each player's playing style will be ever so slightly different from everyone else's. In addition, one reason that a "synthy" sound is such a problem is that the samples are based on a section, not the individual players of a section. To truly emulate an orchestra string section, each player would need to be sampled separately, and then a separate midi track would need to be created for each player. It would then be possible to weave indivual instruments within the section, and do edits in a way that at different points, different players within the section would dominate. One can only guess the cost of such a library and the computer power needed to run it.

Curiously, GPO is the one library that allows for that sort of ensemble building on a small scale. With VSL or EWQL, layering say a solo violin/chamber violin section/full orchestra section, offsetting the attacks ever so slightly and other tricks can help with achieving realism. My hope is that when Garritan releases GPO 2/Orchestra Strings 2 the quality of the samples will be much higher, and the ensemble building possibilities greatly expanded.

For a truly high-end strings library, Audio Impressions DVZ Strings is seeking to allow for greater flexibility, but the estimated cost for the library and the computers to run it is around $25,000. Obviously, that library is aimed at a very small maket niche.

Some of the budget library samples I found to be far from adequate. For example, as a strings player, I found the attacks on most of Kirk Hunter's strings to be very over exaggerated, and they ended up reminding me all too much of someone with overall poor technique (except when such hard attacks are desired for a specific reason).

Overall, I would agree that strings are the weakest link in sample libraries. There are so many suble nuances between bow/bow position on the string/amount of rosin used/bow pressure (via arm weight)/bow speed that a good player will vary based on interpretation of any particular note, that capturing it all is nearly impossible (and this doesn't even begin to discuss portamento, particularly the many cases where it must occur when shifting while executing slurs with the bow, etc., vibrato techniques, etc., etc., etc.).