The fix-verify pass flagged one CONFIRMED regression in the #1 floor loop: it
pinned any tweaked cableless non-note dest to a hard 0, but two dests have a
non-zero synth idle — filter.cutoff (d(...,0.3)) and tempo.nudge (d(...,0.5)).
So touching the filter knob in zero (or, before this, merely opening OMNI, which
created a neutral tweak) pinned filter.cutoff to 0 → 250 Hz, darkening the lead.
- new DEST_IDLE map; the floor loop seeds destVals[k]=DEST_IDLE[k]||0, so a
neutral/reset knob returns a voice to its idle (filter 0.3), an offset rides
from the idle, and mute still → 0. (verified: reset filter → 0.3 not 0;
drag up → 0.767; mute → 0)
- OMNI update() + knob pointerdown now read tweaks[k] directly instead of tw(k),
so rendering/opening the panel no longer creates phantom neutral tweaks (which
also kept the tweaks map — and saved dimensions — clean). (verified: open OMNI
→ filter has no tweak)
A clean instrument (no tweaks) stays byte-identical; the public house patch was
already masked (filter.cutoff is cabled to sun.speed). Deploy gate: GO.
Console clean, syntax clean, 21 knobs, field healthy.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
fb32d10991