I sometimes hear the same type of phase issue with Garritan CFX, and other VIs based on recorded samples of a piano. It can be particularly obvious playing scales with headphones (on certain sections of the keyboard); a few neighbouring notes seem to shift radically between the L & R earphones. But looking at the individual waveforms the amplitude change is not so obvious. And some users I have spoken with don't notice this at all.

I spoke to the Garritan techs and they provided an answer similar to that of Philippe.

Different mics and positions helped me address that "phase issue" in Garritan CFX (although frankly I haven't noticed the phase issue in PianoTeq). Using speakers rather than headphones helps as well. Using different headphones can help also. Or just using a real acoustic piano.

I would also like to take this opportunity to note how slick PianoTeq's "Delay Compensation" button is (mics section).

Joule- Are you using a subwoofer, or is there anything unusual about your speaker positioning?

I'm mainly using headphones (DT-1770 or HD-650), but it's not so much a matter of listening environment in this case. The effect comes from the source which can be verified with a phase correlation meter, or by just collapsing the signal to mono and listening for cancellations. As Philippe pointed out, it's a realistic effect.

Anyway, I've tried some M/S setups in PTQ now. It's interesting to use for a different kind of stereo, but sounds kind of boxy.

PS. Maybe there are other cool tricks as well - like having two different mono pianos slightly panned L and R .. or doing an M/S mic setup by using two instances of PTQ and an M/S matrix.....

You are perfectly right. You will find on page 47 of the manual four standard mic configurations. If you use the MS technique (M+S and M-S), you will obtain the required minus sign on S via a right click on the corresponding mix matrix cell and setting "Reversed polarity" to 1. (Note that you do not need two instances of Pianoteq, you can do that directly in the Pianoteq Standard mic Recording section.)

I will experiment more with mic placement, and possibly lowering the S channel in post-processing. My aim being to make it a bit less wide without losing too much at specific notes.

PS. Maybe there are other cool tricks as well - like having two different mono pianos slightly panned L and R .. or doing an M/S mic setup by using two instances of PTQ and an M/S matrix.....

The stereo correlation is uniquely a matter of mic position: the closer they are to each other, the closer to 1 will be the stereo correlation of the notes. This happens both in the real world and in Pianoteq. Once the mics are at a certain distance from each other, say more than 20 cm, you cannot avoid having some notes in phase whereas some other notes are out of phase, unless you change the mic position for each note. But that would not be desirable, as you would hear the notes jumping from place to place in the stereo image. The fact that some notes are out of phase is part of the natural physics, and in our piano recordings that serve us as reference, we have sometimes some notes which are so much out of phase that their fundamental almost vanishes. Of course we do not attempt to reproduce those particular notes, I'm just telling that to explain that this happens also in the real life.

In short, if you want to increase the stereo correlation in a given preset, and if you have the Standard version, go in the Sound Recording section, and move the two main mics closer to each other. Note that having a correlation close to 1 is mainly desirable when you want to have mono compatibility, that is, summing the two channels provides a good result, without vanishing overtones. Note also that the mono output provides a clean mono.

An example of the problem:

1) Load the YC5 home preset and stereophonic output.
2) Disable the delay and reverb just to make sure that these effects aren't interfering.
3) Bring up a goniometer (phase meter)
4) B-4 will have a stereo correlation of app -0.8
5) D-5 will have a stereo correlation of app +0.7 or more.

You will notice this by collapsing the signal to mono. The fundamental (?) of the B-4 key will pretty much disappear, while the D-5 key stays more intact.

Main point is the following: If the stereo correlation was at least consistent, it would be sufficient to just adjust the S content to taste.

PS. I know that some ears are very sensitive to phasing (I'm not one of them), while others will find the exaggerated stereo compelling. When it comes to production and mixing - negative stereo correlation is de facto bad practice.

I’m of the opininon that Joule actually makes a very valid observation, pointing to an issue that, in my view, definitely merits attention and further investigation ...
So, I'm with Joule: the stereo integrity of PTQ's output (for some of the models) is an aspect of the sound that can be improved.

Thank you, Piet, for adding your comments to the discussion.

And admiration for your compositions and Pianoteq demos.

More from this for me later when the storm passes, no doubt.

With other piano models, such as the model B, there is much less tendency to have the "floating sound" problem, even with the same microphone set ups.

It is interesting to see how we can have differents feelings.
I play only with headphones and find the Grotrian Player preset perfect for that. Also I have the same feeling with the YC5 Home preset (which is a player's perspective). I feel immersed in the sound, and like that very much.
Did you try the YC5 Home preset? Is that feel what you call "floating sound"?

Of course, everyone is entitled to his opinion.

Please note, however, that this comment about PTQ6 sounding "incredibly bad" comes from someone making his very first post to this forum.  This lone comment comes in light of dozens to hundreds of accolades to date.

Sounds fishy to me.  Just saying.

The OP never said that PT6 is sounding incredibly bad, he said that the stereo cohesion in many of the models is.
To immediately brand him a ‘fishy contributor' with suspicious intentions for failing to join in the current wave of applause and having the nerve to say something critical instead, is a most unkind and even rather nasty welcome to a new member of the PTQ community, if you don’t mind me saying so.

The more so because, I’m of the opininon that Joule actually makes a very valid observation, pointing to an issue that, in my view, definitely merits attention and further investigation. The sometimes erratic stereo-mono compatibility of many Pianoteq instruments is a chronic problem of the software (one that has been discussed in more than one beta-test) and PTQ’s stereo output is indeed not always the most solid. (The sound quality of many of its stereo patches can get severely compromised when rendered to mono, which is always an indication that there's something not quite right with the stereo to begin with).

And it’s got absolutely nothing to do with ‘impressions’, or ‘ears’, or with the way the brain processes sound or any of that. It’s something, I guess, in the way the sound is generated that, with certain of the instruments, produces a distinctly fragile stereo signal. Simple as that. There’s something about frequencies being-not-quite-in-or-out-of-phase (between both sides of PTQ’s stereo output) that doesn’t always translate very well to mono.

It’s already much better than it used to be though. I still remember, in the pre-v5 days (v5 being the version where the first improvements to this particular problem got implemented) struggling enormously with trying to mix earlier versions of some of the PTQ-instruments, and it was always because of the exact same reason: weak stereo. I can’t count the number of times I had to resort to third-party stereo tools to make PTQ’s (stereo) output behave in a mix.

So, I'm with Joule: the stereo integrity of PTQ's output (for some of the models) is an aspect of the sound that can be improved.

_

Greg.