MIDI Control Change delay

[elnn]

Hi. I have this problem that appears in all U-he synths I have (Repro, Bazille, Diva) and doesn't repeat itself in Cypher or Alchemy of Logic.

Here's a video of it, using Repro here. CC5 is midi mapped to AmpEnv attack. MIDI slew is OFF

[Viktor [TUC]]

Hi Elnn,
my brain might be a little tired after a long day.
But could you describe with words what's happening and what you expect to happen?
For me the video isn't self-explanatory.

Thanks,
Viktor
u-he team

[elnn]

Each step is to have a different attack time. Step 1 has a long one, step 2 has 0 attack. The problem is, I have to offset the midi change message a lot for it to get in time for the beginning of a note. As you can hear, -14 % (of 8th note @72bpm) start offset is consistent: long-short-long-short. Increase it to be closer to 0 %, it doesn't work at all, Step 1 starts with a way shorter attack and step 2 is not instant attack.

Cypher works ok with -1 % and Alchemy at 0 %.

[Urs]

We smooth out MIDI Learn and host automation to avoid zipper noise.

But if you use the MIDI 1/2 modulation sources (or ModWheel), it'll be as immediate and timely as it gets.

[elnn]

By what margin! 2500+ samples
Yea, though not with parameters without mod inputs. Like Bazille’s ramp or glide time or Diva’s envelopes. Or simply modulating more than 2 parameters of repro rhythmically. A shame:/

Why is it though that MIDI learnt parameters should behave differently? I.e. if I don’t get zipper noise modulating a modulation source with MIDI slew off, why should it be different for MIDI learned parameters? Something about optimization?

[Urs]

Well, yes and no. Precise timing of automation events relies on host, plug-in format and plug-in. As long as VST2 was the weakest link, we never wanted to encourage automation beyond "evolving rather than gating". It also would have been hard on CPU to keep track of potentially hundreds of curves in the early days. So there's a history component in this, I'm sure I'd do it differently if I had to rewrite from scratch.

The reason MIDI Learn ties itself to host automation is exactly that: So you can use MIDI controllers to create host automation tracks rather than MIDI tracks. It never occurred to me to decouple the two since it might be rather confusing. Also because the original concept of VST3 was identical to host automation - any parameter could get a MIDI CC attached to it and the host would deal with MIDI (making MIDI Learn impossible). We abandoned that however, in favour of MIDI Learn.

I've been thinking about adding a preference for automation smoothing as well, but the core really is from 2003 and I have no idea how much trouble we get into when touching it.

[elnn]

Well, yes and no. Precise timing of automation events relies on host, plug-in format and plug-in. As long as VST2 was the weakest link, we never wanted to encourage automation beyond "evolving rather than gating". It also would have been hard on CPU to keep track of potentially hundreds of curves in the early days. So there's a history component in this, I'm sure I'd do it differently if I had to rewrite from scratch.

The reason MIDI Learn ties itself to host automation is exactly that: So you can use MIDI controllers to create host automation tracks rather than MIDI tracks. It never occurred to me to decouple the two since it might be rather confusing. Also because the original concept of VST3 was identical to host automation - any parameter could get a MIDI CC attached to it and the host would deal with MIDI (making MIDI Learn impossible). We abandoned that however, in favour of MIDI Learn.

I've been thinking about adding a preference for automation smoothing as well, but the core really is from 2003 and I have no idea how much trouble we get into when touching it.

I see. The past comes back to haunt us.
I hope you do find the time and resolve to implement it. Incapacity to step sequence envelope times and volume properly disqualifies them from some great uses. Your synths are the only synths I use, as they have that something I couldn't find anywhere else, but e.g. Diva, bass-goddess that she is, all but makes me glum now that I found that step sequencing envelopes is closest to how I think about sound design. And it pains me to see her go, I don't want any affair with anyone else

cheers!