How do you make virtual instruments sound ok ?

I think you will get a lot of good responses to this, maybe.  Though maybe post it in the X2 forum or even a non-SONAR forum since it is more of a universal question and not specific to 8.5.

There are a ton of things I do, I could probably write a book.... if I had to pick one thing I would say write to the sample, not to the instrument it's pretending to be.  That is to say, if you're working with a violin ensemble, don't write for a real violin ensemble and then try to make the sample emulate it.  Instead, imagine that the violin ensemble SAMPLE is it's own instrument, like no other instrument out there, and write so that the SAMPLE sounds good.  The patch/sample has it's own characteristics and certain things that it does well... utilize those and if it's not working for what you want then maybe it's not the right instrument, try a different patch/sample.  In my experience, every sample has a certain way to be played where it sounds the best... so play it that way and don't try to force it into something else.  Not sure if that makes sense but it's a quick answer...

Here is an example of composition using EWQLSO and Garritan. This is the 1st mvt of my new Concerto for Euphonium and Concert band. Instruments used : Euphonium solo (Garritan) Picc (EW) 2 Flûtes (EW) 1 Ob (EW) 1 EH (EW) 1 Bsn (EW) 1 Eb Cl (Gar) 1st cl players (Gar) 2nd cl players (Gar) 3rd cl players (Gar) Eb Alto Cl (Gar) Bass clar (EW) 1st Alto Sax (Gar) 2nd Alto Sax (Gar) Tenor Sax (Gar) Bari Sax (Gar) 1st Trp (EW) 2nd & 3rd Trp (EW) 1st and 3rd Fr.Hn (EW) 2nd and 4th Fr.Hn (EW) 1st & 2nd Trb (EW) Bass Trb (EW) Euphonium section (Gar) Bass Tuba (EW) Dbl bass (EW) Harp (EW) All perc (EW) What do you think of the sound ? What would you change in the mix to make it sound better ? Thanks Alex (from France) http://www.alexandrecarli...erto-mvt1-finalmix.mp3

Conventional mixing isn't the main thing with classical music. It's more about musical arrangements and articulations than the mixing techniques usually discussed here. As mixers, our job is really about letting the instruments speak for themselves.

Long before digital audio or even electronic audio, composers and conductors were the mix engineers. Panning and levels were set by where each instrument was physically located on the stage, with the conductor serving the role that automation plays in modern productions. 
In particular, Beethoven probably contributed the most in this regard, and his stage setup has not changed much since. That's why most orchestral libraries follow the classical arrangement of instruments in the panorama, making them very easy to mix. 
When there was no "sweetening" - no effects, no post-production automation - it was up to the players and the conductor to add interest and dynamics. When composing electronic music in a classical vein, we usually follow the same philosophy, relying on different articulations to keep it interesting. 

I am not an EWQLSO user, but every sample library offers multiple articulations and some way to switch between them, usually using keyswitches, keys that don't play a note but rather enable a different sample set. This allows a violin, for example, to quickly switch between legato and staccato. This may be the added touch of realism you're looking for.

I agree with both Daves and maybe what is wrong is you are manually entering data. 
Do you have the skill to acctually "play" the notes on a keyboard? 
I think that would bring out the dynamics and articulation needed. 
I could never get anything to sound right to me if all I did was draw it on a screen, I have to play it.