At least, it sounds realistic to me, but I'm not a guitar player.
 
I had a MIDI file for piano.  I ran this through the Strum Acoustic VST.  With surprisingly few MIDI edits, this sounded pretty realistic to me.
 
I then bounced that to an audio track that I ran into Melodyne Editor, selecting the polyphonic algorithm, of course.  There is a menu option to introduce randomness of either pitch or timing.  I chose slight randomness in timing, and didn't do anything with the pitch there.
 
For pitch, I selected the scale for the tune, which was C major.  My thinking was that Strum Acoustic probably used equal tempered scale.  By selecting the correct scale in Melodyne, we could move to just intonation, which would be slightly different.  From there, I used the macro button to correct pitch.  In this case, correcting about 75% sounded the best.  Correcting any more than that sounded very distracting.
 
The end result sounds very authentic to my ears,  Maybe a guitar player wouldn't like it as well, but I think it is pretty neat.  I continue to be amazed by Melodyne.

The recording with this trick is up at
http://bordogni.info/2013/10/250/
 
I'd be curious what others think about the guitar sound.  Could this pass for a 12-string?

I think you did a great job in getting a piano converted to a guitar.. however a guitar is a very hard instrument to emulate and so to me this would only sound ok in a mix with other instruments as a six string.
 
To me sorry (and yes i play some guitar..lol) the 12 string conversion is far from realistic, it sounds more like what it is..a 6 string with something like a chorus effect.
 
Converting from 6-12 string is more than random timing differences, as at best that gets you two 6 string guitars. Things to consider for example, the E,A,D,G strings are doubled in octave, while the B & E are in unison, and the timing would be polarity based, ie upstroke and downstroke and not random, but more a fixed amount related to strum speed.
 
if SA does midi out i would take that output for the 6 string and manufacture the 12 string counterpart in midi..but probably they also have a sample pack with a 12 string.?
 
regards, Ian

Mystic38
the E,A,D,G strings are doubled in octave, while the B & E are in unison,

Ahhhh.  I didn't know that.  I was thinking there was something a little off in depth.  That would do it. 

Mystic38
and the timing would be polarity based, ie upstroke and downstroke and not random, but more a fixed amount related to strum speed.
 
if SA does midi out i would take that output for the 6 string and manufacture the 12 string counterpart in midi..but probably they also have a sample pack with a 12 string.?

I'll look into that the next time around.  Actually I did this as an exercise in using Melodyne.  I had seen all those videos about beefing up vocals and thought I'd try on guitar.  It is still pretty amazing to me that this all works as well as it does without a lot of artifacts.  I mean, those "slow downer" programs have been absolutely horrible over the years, yet Melodyne is able to make much more complex transformations with hardly a trace.
 
Thanks for the input.  Now I'm smart enough to not tell anybody this is supposed to be a 12-string sound.  :)

Mystic38
.but probably they also have a sample pack with a 12 string.?

I don't really see that option at http://www.applied-acoustics.com/strumacousticsession/overview/
 
It looks like the Pro version of the product does everything BUT 12-string.  It seems that 12-string modeling would be a natural for that product.  They are already 90% there with what they have.  They would just need to map the octaves of those 4 strings, give some options for varying the pitch tightness, and model the timing based on strum direction.  Not exactly an overnight project, but still well within their capability.
 
I've always enjoyed the sound of the 12-string.  It is nice to better understand what is going on there.

no 12 string?.. bummer..
 
i completely agree.. it would likely need to be sampled from scratch but i certainly agree it would seem a logical and natural fit for a strumming program!
 

Mystic38
it would likely need to be sampled from scratch

I don't think it WOULD have to be sampled.  I think they could simply model both strings.  Whenever the low E is played, they would additionally model playing the E an octave higher, perhaps introducing a slight pitch variation and slight timing delay (or advance, depending on which direction the strum stroke is going).
 
It is surprising to me this doesn't already exist, but I bet AAS will add that before too long.

One of the attributes of a guitar is that the tone of the same note is dramatically different depending upon where on the fretboard it is played and which string.. so the octave of the low E string (ie low E played at the 12th fret) sounds dramatically different to the open octave'd string on a 12-string.. Then at least on the G string, the octave string is plain wire, not wrapped, which will add to the tonal differences.