Pitch quantisation on touch and glide

[thetechnobear]

This seems to have been changed… (possibly 2.x?)

in the past, I remember there was a combination of glissando/vibrato settings that would mean you’d get a un-quantised touch, and also glide.

I cant remember exactly, something like only using vibrato (or glissando?), and turning down all the sliders…

it was similar to init rounding/rounding = 0 on continuum

but Ive noticed the settings have changed a bit, and now no combo works.

ok, not quite true, you can with glissando get it to some degree, but its still feels like there a large ‘centre’ zone thats quantised too, and also slow slides show its still snapping to the grid.

I raised a ‘bug’ saying what I wanted…

but I trying to figure out, has it really changed?

I do half remember,
the glissando was always a timed slew… that tallies with observation above.
so, that’s why you did it with vibrato, but without a ‘return’, but thats does not work anymore.
as return = 0, means it doesn’t even start to the glide, at least if you move slowly.

ever time Ive looked at this, I wonder, why they made is to complex, when the continuum, can do it with just 2 rounding parameters

[thetechnobear]

got a couple of interesting posts on this in discord…

@embodme

full un-quantised pitch for first touch and also glides…
→ Set Tune Loc in pad, with In Tune Width 0

I pushed back, said I was using pad / in-tune=0/smooth=0, and was still hearing quantisation…

@embodme

There is no snapping to our knowledge, what keygrid size are you using? The only way you’d have a snapping is if. you are using 2x2 keygrid, then you’d hear the quantization due to the limit of FSR cells interpolation. With 3x3 there is no snapping. If i’m still not catching your issue, please send a video. Thanks

this is very interesting…
when Id been testing this, I tried every combo of setting, and thats how I’d determined it was ‘sometimes’’ kind of working with pad / in-tune=0/smooth=0
but other times, it clearly was not…

so, its possible, I was testing on different layouts 2x2, 3x3 and 4x4… so was seeing different results - so more testing needed…

I am intrigued though, why 2x2 does not work? or really how the whole interpolation works at all.

I assume interpolation was always between two cells, so that the pressure gives a bias between them…
now im starting to wonder if often your finger covers 3 cells,
and, that potentially the pressure differential on 2 cells (after filtering, taking into account cell variation) is not quite enough to (confidently) determine the bias left or right.
but the 3rd cell, however small the pressure, gives that extra bias.

that said, if this is the case… Id say, its likely, that there are other scenarios, when moving slowly, its effectively got a 2 cell ‘resolution’ for the x axis.
i.e. it might be quantising to a note, but it might be quantising to (C#/D) / 2

Im kind of ok with this… as Ive said before, Im fine with technical limitations, I just wish they were made clearer.

in practical terms this means, if you want un-quantised, then you will want to use larger note spacing (or for microtonal, use a quantised grid with microtonal scales)

all that said, Im now very interested to use the API to see if we can see this cell quantisation, or its it more an inference from the pitch / scale functionality. if the data is really daw in the api, there may be other interesting options