docs(notes): OPAL-26 listening pass — PLN's floor feedback, cross-checked

Appends the OPAL Festival 2026 section to the archivist's log: floor reactions
per track, transitions, and the sound/technical notes, each paired with what the
measurement said.

The pairing is the point. Where PLN's ear and the meter agree, that is a
calibration datum:
  "not loud enough, i had to set max volume"  -> -18.2 LUFS, 4.2 dB under target
  "MAFIA bass goes too high for a few secs"   -> +24.5 dB over local median,
                                                60-250 Hz, at 1:02:10 (he filed
                                                it at 1:03:10 — one digit off)
And where they disagree, that is more interesting still: Mafia got a strong
floor reaction while being the NARROWEST (width 0.05) and most bass-dominant
(78% in 60-250 Hz) track of the night. Width is not what makes these land.

Near-miss worth recording: I first wrote this file with Write, having been told
twice by a stale tool cache that it did not exist. It did — 319 lines including
the entire Take89/Montreuil listening pass. 314 lines were destroyed and
recovered from HEAD. The corpus CLAUDE.md calls a documentary source survived
because it was committed.

armada/tide-table/performance_notes.md

@@ -317,3 +317,83 @@ the code or the SynthDef, and **none of them were mix problems**:
 rules came directly out of this session (PV014, PV011-as-error) and immediately found
 the same defect class in `mafia_sans_serif` — which was already carrying PLN's own note
 *"FIXME ACID WEIRD NOT GOOD"* with no known cause.
+
+---
+
+## OPAL Festival 2026 — "Sunset Forest" — 2026-08-08, Scène Cabane
+
+Timecodes are in **master time** (`Opal26_Master_Raw.flac`, t=0 at the first
+recorded frame). The set proper runs `bombe_dj` → `REVOLUTION`; the 9:45 before
+the recording is soundcheck, and `quand_on_decolle` after `REVOLUTION` was an
+ambient transition out, not a set piece.
+
+### Floor reactions
+
+- `0:36:51` · **Piment Bresilien** · `[+]` **PEAK DANCEFLOOR.** "so hot, people
+  danced bang, i let it last" — PLN extended it live to 8:26, the longest track
+  of the night. Measured: loudest track of the set (−16.2 LUFS), joint-widest
+  (0.37), 32% sub / 49% bass, centroid 297 Hz. *The one to study: this is what
+  the arc should aim at.*
+- `1:01:22` · **Mafia** · `[+]` "got a real strong floor reaction too <3".
+  Measured: near-MONO (width 0.05, narrowest of the night) and the most
+  bass-dominant track (78% in 60–250 Hz, only 10% sub). Floor impact clearly does
+  not require width — worth remembering before "fixing" it.
+- `0:29:54` · **Take five Drops** · `[+]` "perfect vibe match" at sunset; PLN has
+  a video of a dancer during it. Landed at the golden hour by luck of ordering.
+- `1:04:33` · **Vague de CRIME** · `[-]` "mixed fb compared to my hopes."
+  Measured: densest track of the set (crest 14.8 dB, LRA 9.3 — the most
+  compressed), and the thinnest low end so far at 7.6% sub.
+- `1:09:12` · **Desire** · `[-]` "uncooked and imo not great." Measured −21.6
+  LUFS, 3.5 dB below the track before it. Its bass had been repaired from a
+  non-compiling state only 5 days earlier (J-5) and never played to a floor.
+- `1:15:47` · **REVOLUTION** · `[-]` "a fair but not bang conclusion."
+  Measured: quietest track of the set (−23.7 LUFS) and almost no low end at all
+  — **1.9% sub** against 30–50% for most of the set. Highest centroid (422 Hz).
+- Mid-set the crowd thinned because **the festival kitchen called dinner**, not
+  because of the music. PLN judged the quieter back half as matching the actual
+  floor energy. Real, but he still endorses the criticism for future arcs.
+
+### Transitions
+
+- `0:18:28 → 0:25:41` · **do_it_right → You My Sunshine** · `[+]` "blend of voice
+  into sunshine was perfect... at least one 'magic vocal transition across
+  tracks'." The good case to reproduce deliberately.
+- `0:25:41` · **You My Sunshine** · `[-]` an **arp lingering from the previous
+  track** audible through the whole track, gone once `take_5_drops` starts.
+  `[!]` To remove in the master. *(Orbit not yet pinned: the RMS-duty scan is
+  too coarse — d10 triples its duty during Sunshine and dies in take_5, but
+  Sunshine declares d10, so it is not a declaration-level ghost. Needs a
+  pitched/spectral lens, not a level lens.)*
+- `1:15:36` · 34 s of **full digital silence** before REVOLUTION — this was a
+  **clap / encore moment**, not a dropout. With no floor mic it recorded as dead
+  air. `[!]` Trim to ~4 s in the master.
+- PLN "forgot quand_on_decolle" as the intro and "ran head first into diams
+  techno" — the set opened on `bombe_dj` instead of the planned ambient intro.
+
+### Sound / technical
+
+- `0:00:00–0:00:06` · **intro** · `[!]` cut to 0:06 — first *diams* note.
+- **intro** · `[-]` **the régie son reported the diams intro was saturated.**
+  An outside ear, on the PA, at the very start. `[?]` not yet reproduced in the
+  recording — the master's own true peak is −1.0 dBFS with zero flat samples, so
+  if it was real it happened downstream of our capture or in the PA.
+- `1:02:10` · **Mafia** · `[-]` "bass goes too high for a few secs."
+  PLN filed this at 1:03:10; measurement puts it at **1:02:10.6 (+24.5 dB over
+  local median in 60–250 Hz)** and **1:02:13.2–1:02:15.5 (+15.0 dB)** — one digit
+  off, same event. Found only with a band-limited lens; broadband RMS saw
+  nothing. This is what motivated `tidal-ears master saturation --band`.
+- `0:08:50` · **WAP** · `[?]` "bass is very wobbly, lets confirm and cleanup if
+  we can." Not yet cross-checked.
+- Whole set · `[+]` "honestly it works great on jack no sound drop/worries."
+  Measured: **17 xruns in the whole 1h27**, in two events (+16 at 3:25, +1 at
+  76:10). Against 1676 in a 10h48 rehearsal, this was a clean run — despite the
+  package hitting 100 °C with sustained throttling.
+- Master · PLN's own first pass was "a quick 80/20 audacity jab." He judged it
+  **"not loud enough — like i had to set to max volume or closeby."**
+  Measured −18.2 LUFS: ~4 dB under the −14 streaming target. His ear and the
+  meter agreed exactly.
+
+### Execution
+
+- `[+]` **15 of 15 tracks played in the exact planned order** (backlog.md
+  setlist vs the gig log's track events). No improvised reordering, no skips.