As many other piano players I have a amassed a collection of piano VST's, most of them which I never use and are no good. The notable exception was Garritan CFX which has been my workhorse. Being a fan of the Steinway D sound I was very excited when the VSL Seinway D was released...the specs on paper looked exceptional so I jumped in and got the full version.
The BEST: This is the best sounding piano sample set I have ever heard, of any piano, period. The different mic perspectives and ambience are phenomenal but the quality and character of instrument itself is just gorgeous and the numerous velocity layers capture the dynamics and nuances of the instrument in a way that I have never heard a sample based piano do before. Which is why it pains me to note...
The WORST: Sound is not everything in a piano VST that is meant to be played live with a keyboard(not just sequenced). The simulated instrument should mimic how the real instrument reacts, and this instrument fails, almost as if a gross error or simply no attention was paid to the MIDI velocity -> velocity layer map. As others have noted the instrument mapping is either aggressively biased to the ff side or the velocity layers sampled were very skewed perhaps due to the mapping used for the 'robot'. Now, this would usually not be a problem as every other tool on the face of the earth has velocity curve editors, but that critical feature does not seem to be present in Vienna Ensemble, except for the very limited 'MIDI Sensitivity'.
Can anyone suggest a workaround to this problem? I use a Kawai Novus NV-10 which has limited control over velocity curves, however this has not been a problem with other VST's like Garritan and the limited control is plenty to make any adjustments. With the VSL Steinway it seems the adjustment needs to be pretty significant and non-linear.