The recordings used in the Avantgrand that are of the CFIIIS which are great by the way, are recordings made in 2000 for which Yamaha spent a small fortune to record. They did a very small adendum in 2010 to add some more color but the core sound is the same. They are very high quality recordings which are then converted to fit the sound they look for in each model. The piano sound in the Motif/S90/EsXS piano plugin card of the CFIIIS is the same session. The sustain was recorded even longer than what they used originally and that is what you get on the AG when the CFIIIS is the piano you hear. Yamaha never will record a CFIIIS from scratch for any new instrument ever because they don't need to. It does not make financial sense. Just remember that even though it is an old recording, it is still mastered with newer equipment making it sound more full with a longer decay.
From 1986 to 2009 Roland researched and developed modeling and crescendoed with the V-Piano. The technology in the V-Piano is leaps ahead of anything 1986...
Please quote your source for the information about Yamaha's sampling so we can all make an assessment of the credibility.
I would not describe the AG as having much of a longer sustain. It does have longer attack samples but the sustain/decay behaviour is still poor due to the looping (which Yamaha has never done well). The AG sounds tonally different to any Clavinova I have played or heard. More mellow on Piano 1 but lacking a degree of timbral change according to velocity. And more metallic on Piano 2 but giving believable timbral change. I have played Yamaha DPs for years and never heard these pianos. Yes, they are all from the CFIII-S. But I hear something different with the AG. As an aside, the GranTouch also had a unique 88 key Yamaha sample in my opinion, which was substantially different to the Clavinovas of the day (including your CLP-990).
Yes, overall the V-Piano is more advanced (and effective) than what they were doing in 1986. But frankly, relative to what else is/was out there in the market place at the time, what they did in 1986 was literally ground-breaking. The V-Piano is an update of the same technology and principles. Furthermore, I'm bound to point out that the RD-1000 had other (fully modelled) voices, not just piano. And it had a full wooden action. And it was adopted by numerous pro players. Somewhat ironically in some of those respects the V-Piano is a regressive step.
You're correct in that the V-Piano in some ways could have been better by having a real wooden key action and also it could have used some other voices besides the standard midi gm to layer for good measure.
Also, you mistake me for someone with an agenda, or someone making things up to convince people to pick one company over another. That is very far from the truth. I have no reason to make anything up, nor would I accept money to promote one company over another.
The truth of the matter is...
I do have my sources for the information about the recording sessions Yamaha did in 2000 of their CFIII-S. They've used that same recording in almost every digital piano in one way shape or form. Like I pointed out, the differences in sound that you hear when you play different DP's by Yamaha is a combination of how compressed the sample along with how many layers used. Resonance and sounboard samples were taken seperately and in models like the CLP990, are layered for the piano 1 sound. In all future Clavinova's the sounboard sample was mixed together with the main piano to save memory space. Also, certain notes were chosen to be stretched which in essence more than half the notes from the all 88 sampled piano go missing while all the remaining notes are stretched. I know you do not believe me, but your AG does have some new recordings in 2010, along with the originals from 2000. I know for a fact they no longer use anything older than the 2000 recordings of the CFIII-S. I wish Yamaha would release the unflitered and uncleaned sound of their original recording in all its glory as a software sampler, but they never will. The closest to the original would be on the CLP990. Just so you understand, Yamaha used many mics for the recording and the AG flagship uses these recordings in ways not possible on older Clavinova's.
Whether you believe me or not does not matter to me. It is also not my place to bring in my sources either and this information I have is common knowledge within my circle of friends and honestly is not a secret. I do not know much about the 2010 recordings of the CFIII-S that were needed but I can assure you the difference in sound you hear has to be from those samples and how they were used. But make no mistake, the original recordings are the main source of the classic and excellent CFIII-S. Furthermore I prefer the sound of the CFIII-S over the CFX recordings which I heard on the NU1. But that is personal preference.
Yamaha motto: If it aint broke don't fix it and definately don't spend any more money on it than we have to. Newer isn't always better as well. Consider yourself lucky that the new CFX recordings weren't used on the original AG line as that recording session was not as well done in my opinion as the one in 2000 based solely on what I hear in comparison.