Maybe Cherry Audio can solve steppy controllers?

[wwjdx]

I believe it is possible within programming knob movements to reduce stepping in physical knob turns that mostly only get 128 positions over MIDI.
IE the jumps tween single positions of knob movements

I've done no programming in decades, so all I have is potential functioning ideas that could be good or bad, or a catalyst to brain storming toward something better. In other words I have no idea how to accomplish any of this, all based on my best guess of theories about how software synths are programmed and operate.

The goal: spread midi 128 controller physical knob/slider steps SMOOTHLY across a softwares 0-1024, or 65,000 steps as found under mouse manipulation, based on speed of movement.

It's all about algorithms and timing.
And, of course, the feature would have to be able to be switched off also.

So, you turn a filter knob paired to a soft synth from 50 to 49, there is already super low latency as it jumps down the line: knob > internal converter > midi send > received > understood > sent to synth as a change (No clue how long that takes, not important)
the SYNTH jumps from lets say 512 (assuming 0-1024 range in the synth) to 451* with a noticeable audiable jump. From Bang to Bing. Very obvious stepping.
*ignore my math please, just winging it for an example

Now, if there was a "movement algorithm" involved, it could SMOOTHLY step from knob 50 to 49 passing all the extra steps just like the mouse can.
If it is a QUICK turn, it skips many steps to get to the end result quicker. If it's a slower turn, it rolls a smoother numeric step down, like the mouse can do on a virtual knob.
If the knob turn was really fast say, 90% down to 20%, same thing: fast turn skips many intermediaries the might not be audiably perceptable. Test and tweaking the speed would be required, OR maybe even a user setting choice via hidden knob.

The slope mapping tween the 50 and 49 would not be consistant, but based on speed of the change. Now thinking about it, one step is not enough to determine the SPEED of the change, but two might be. Any case, the smooth change would need to ride the latency as it adjusts to the new amount, instead of snapping strait to it. I feel the change would lag behind the knob just a little bit but still enough to be musically in the pocket.

Hence, there WOULD be tiniest bit of latency (I'm not sure how fast things operate in that chain above) and would need to give the user the option to switch it out as desired.
However, the possibility of using a 128 stepped knob to SMOOTHLY sweep a filter via the wonderful full range of the software could be game changing.
I have not seen too many, okay I have not seen any controllers (havent dug too deep) that allow 1024 or higher steps. It's a shame because all this great software gets neutered by external controllers that way.

Anyway, just an idea, take it or leave it

[poetix]

The module implementer may (and often should) apply a smoothing function to the knob value to avoid glitches and crackles when it's turned. There's even a handy utility class, SmoothValue, which does just this.

On the other hand, if a parameter is meant to support audio-rate modulation via a CV input, you don't necessarily want to be smoothing out that input. So it's a bit of a case-by-case thing. Something like a filter frequency knob should definitely be smoothed, imho, but do bear in mind that this incurs a very slight performance cost.

[wwjdx]

right. I could see the performance cost being factored in:

a. you can always switch the feature off for fully live
b. you accept that turning the knob with have the slightest catch up delay while it decides to slide slow or fast for you.

I could work with that. generally fast filter moves from 0-127 aren't that noticeable anyway, but the slower one, you would feel it tracking the knob, but be much smoother than steppy. just thinking out loud.