resolution of midi controller boards

astroman · Post by **astroman** » Fri Jun 02, 2006 12:46 am

On 2006-06-01 18:19, husker wrote:
...It would be a huge change to make Scope itself do anything else, so I wouldn't expect it anytime soon

it is already there, since a couple of years, as you described above

On 2006-06-01 16:56, husker wrote:
What i'm saying is that VST Automation has no relationship with midi whatsoever, so any midi limitation of CC being 7 bit do not apply. Kore uses VST automation, not midi...

Scope works exactly the same way - the internal values are (probably, I didn't count) 24 or 32bit which are mapped to whatever screen resolution and midi CC applies.
For compatibilities sake they support the 7-bit midi standard, but the internal resolution is MUCH higher.

To see it working open an empty Modular, connect a Value slider (0-127) to a Control Smoother to a Value monitor and watch the numbers.
The latter 2 modules are part of the Flexor package, but this doesn't include anything that isn't present is SFP (SDK) - it just arranges things in a supersmart (so to say) way

afaik developers have always used this method - it's just not visible, but sound is about hearing, isn't it

that's why I wrote the problem doesn't exist

cheers, Tom

husker · Post by **husker** » Fri Jun 02, 2006 12:56 am

I'm sorry, but smoothing a 7 bit signal is not the same thing at all as actual 14bit control. Can you tell me how I get more than 7 bit control from an external midi controller into Scope?

Shroomz~> · Post by **Shroomz~>** » Fri Jun 02, 2006 1:09 am

I was talking about the method of coms used between the Nord Mod editor & the hardware. AKAIK it's midi machane code, but I may be wrong. Either way, you say Kore uses no midi, but then go on to say all Scope needs is Nrpn support. Well I agree that would be nice in the from of an NRPNtoCV (advanced MCV) modular module & maybe even possible considering that Scope will recieve, monitor & pass thru Nrpns, but I can't help feeling that you guys are twisting something a little here.

husker · Post by **husker** » Fri Jun 02, 2006 1:32 am

Nah, we're just mixing our metaphors?!?

Kore does it it's own way, which relates directly to the VST automation spec, which in no way applies the way to do it the best way to do it in Scope. Tom claims we don't need it at all

In theory you could feed NRPN into a Modular patch, as it is just 4 x CC messages anyway. Something similar has been done with SynthMaker, which only does CC native, but can be coerced to decompose the NRPN message and get a full 14 bit value.

I am more referring to the standard CC mapping that can be applied to any Scope synth/effect. At the moment you can only configure plain old CC 7bit.

If you look at Live 5, you can map CC or NRPN. You can even do relative mapping (no knob jumping). It would be great if Scope could do this stuff, but it doesn't (sorry Tom). Mostly it doesn't matter, but it does sometimes (and even my ears can hear it...though I am used to smooth as silk 12/14 bit control on my Andromeda

Cheers.

Shroomz~> · Post by **Shroomz~>** » Fri Jun 02, 2006 2:53 am

It's one thing if the host like Live5 is set up with NRPN mapping capabilities, but for Scope to have (and properly use) that function would need SFP itdelf to be re-built in many areas & all SFP apps/devices/plugs to be re-built to use the new upgraded midi spec. That's a lot of work & a thought that I doubt very much CW will relish & jump on in a hurry.

An NRPN translation module for modular would be possible if someone were to see fit to develope it, although it would need to be done by a clever cookie.

Interestingly, there's a few NRPN modules for synthedit (CVtoNRPN, NRPNtoCV etc) but the guys who made them seem to think 12bit is high enough resolution for native land (Cubase etc) without stability problems under heavy multi parameter automaton. IE they think 14 bit would be too much for cubase in certain scenarios.

at0m · Post by **at0m** » Fri Jun 02, 2006 5:44 am

Even my first soundcard, Soundblaster SB64G, used NRPN for its onboard SF2 sampler (4MB memory) and it was pain in the 4ss to work with. Not the bandwidth it used, but just to draw a controlling signal in Cakewalk (at the time, round '94) was near impossible. And there was no other inferface to the card's sampler for automation...

Nowadays NRPN or signals like the Mackie Control uses (another form of MSB/LSB, combining 2 CC# into one) are pretty common, since they're natively supported by the software. But you don't want to mess with these yourself, they're a PITA.

astroman · Post by **astroman** » Fri Jun 02, 2006 9:15 am

On 2006-06-02 01:56, husker wrote:
...Can you tell me how I get more than 7 bit control from an external midi controller into Scope?

for what purpose ?
I didn't mention incremental (and feedback) controllers (the hardware) for nothing

If you load a preset, the internal values are recalled with internal resolution.
You dial towards a new value (let's assume a filter sweep), the it starts at the (hi-res) saved value. There is no effective difference between a smart interpolation that generates the appropriate data flow internally and high resolution 'original' values.
Let alone you'd hardly be able to reproduce any such value mechanically.

Interpolation doesn't mean to simply calculate the arithmetic average - in case of the Flexor ValueSmoother they seem to interprete the positional difference in relation to time and this works very well.

I'd understand the need if you couldn't operate a Scope device smoothly, but that isn't the case as far as my listening tells me.

cheers, Tom

husker · Post by **husker** » Fri Jun 02, 2006 1:50 pm

OK, simple example:

Running some sounds through Interpole, the filter running into self resonance. When I adjust frequency on the filter with a 7 bit CC coltroller, it stair-steps. Sounds crap, can't use it. And I *do* like to do this sort of thing...

Do the same thing on my Andromeda with 12 bit control and no stairs, smooth as...

128 values is not a lot to control a range of frequency.

Flexor may have some magic, but that doesn't help elsewhere in Scope.

JP76 · Post by **JP76** » Fri Jun 02, 2006 4:00 pm

quote : in case of the Flexor ValueSmoother they seem to interprete the positional difference in relation to time and this works very well.

I'd understand the need if you couldn't operate a Scope device smoothly, but that isn't the case as far as my listening tells me.

cheers, Tom
[/quote]

So the Flexor ValueSmoother can smooth a midi control sequence once it has been recorded in 7-bit to increase the resolution to 32-bit?

astroman · Post by **astroman** » Fri Jun 02, 2006 5:22 pm

it translates to a 10 digits number between -1 and +1, you can set the 'intensity' of the process and even modulate it, so that the more you approach a value the more smooth it gets (or vice versa)

btw even CWA's regular controls have a similiar behaviour:
if you operate a control on screen (say a slider in modular), then it jumps in 127 steps from min to max if you drag exactly from the center of the control.

But if you first move the mouse away from the center and then start to drag (in the direction the control is supposed to operate), the screen values will still display 127 steps, while the precision of the 'effective' value of the control is increased according to the longer distance on screen.

cheers, Tom

husker · Post by **husker** » Sat Jun 03, 2006 12:44 am

Yes, the on screen control works very well. We are talking about external contollers via CC here though

astroman · Post by **astroman** » Sat Jun 03, 2006 1:25 am

On 2006-06-02 14:50, husker wrote:
OK, simple example:

Running some sounds through Interpole, the filter running into self resonance. When I adjust frequency on the filter with a 7 bit CC coltroller, it stair-steps. ...

that's correct, a very slow sweep of the self-oscillating filter makes it noticable.

On the isolated sound - I'm not sure if one could detect it (unless explicetely pointed to it) even in a 'minimal' mix.
Maybe they considered it sufficient, as in 3 years noone complained about it, but the afforementioned 'graphical smoothing' is used in Interpole, too.

In any case the CC translation depends on the implemetation of the respective device.
Afaik Stephen Hummel modified CWA's old Bluesynth (or Inferno?) with smoother parameters and since then all developers took care of this.

It's absolutely not my point to 'defend the platform', but it makes few sense to rewrite half a million lines of code for an academic advantage.
The smoothing of the Interpole parameters is fixable (imho), maybe CWA is open for suggestions...

On the other hand even completely analog gear has parameter jumps if the dial is hopping accordingly

cheers, tom

astroman · Post by **astroman** » Sat Jun 03, 2006 1:30 am

On 2006-06-03 01:44, husker wrote:
Yes, the on screen control works very well. We are talking about external contollers via CC here though

oops, should have looked before I sent the post above (it was on screen for a while)...
both (effective) values are identical - I stepped it with the pageup/down keys and counted the 127 steps

cheers, tom

at0m · Post by **at0m** » Sat Jun 03, 2006 7:14 am

Arrow Left/Right would give the smallest resolution, as set by the creator of the device. PgUp/Dn makes coarse adjustments...

astroman · Post by **astroman** » Sat Jun 03, 2006 7:36 am

it increments/decrements the midi cc values by 1, so it served me as a fake external controller

the arrow keys are a much better example for the internal resolution - rtf (me, ouch...) or ask a question on planetz - tnx at0m

cheers, Tom

<font size=-1>[ This Message was edited by: astroman on 2006-06-03 08:43 ]</font>

at0m · Post by **at0m** » Sat Jun 03, 2006 8:49 am

Still.

That depends on so many low level factors. Different controls on different devices could still be different, as you gave an example with WL's stuff.

Then there's lots low level SDK/DP stuff that can make more differences, between MIDI, GUI, and actual DSP sound math...

FrancisHarmany · Post by **FrancisHarmany** » Mon Jun 05, 2006 7:57 am

I use control smoother!

This module smooths out any control signal fed to it. It is ideal for smoothing steps from low-resolution midi controllers, or for the creation of glides between notes generated by note/control sequencer, and it can also round the edges of a square or saw shaped LFOs. This module is optimized for non audio-rate, control signals.

<a href="http://adern.com/manuals/flexor/referen ... her.htm</a>

valis · Post by **valis** » Mon Jun 05, 2006 9:03 am

On 2006-06-02 02:32, husker wrote:
Nah, we're just mixing our metaphors?!?

Kore does it it's own way, which relates directly to the VST automation spec, which in no way applies the way to do it the best way to do it in Scope. Tom claims we don't need it at all

KORE actually uses OSC to communicate with the host PC, which is an open source spec that NI did NOT create, nor did Steinberg (re: 'relates directly to the VST automation spec'). In fact you'll see no Steinberg technology or product listed (yet):
http://www.cnmat.berkeley.edu/OpenSoundControl/

husker · Post by **husker** » Mon Jun 05, 2006 1:20 pm

OSC for the comms to the controller. Great! It would be good if Creamware supported that one day.

If course Kore (the software bit) does use VST and VST automation, that what it is for...hosting and managing presets for VSTs, and providing automation of parameters. If'm sure there is some Steinberg technology in there somewhere

valis · Post by **valis** » Mon Jun 05, 2006 6:59 pm

Using the 'vst automation' in software doesn't help much with hardware control of scope though does it?

Anyway OSC is still only just gaining market acceptance. It's a worthy addon to the Creamware 'wish-list' but I wouldn't expect it to get integrated into current versions of Scope.