6f87ecc4db
The port reproduced the floor exactly (159) once the predicate was named, and counted an independent capture from a different session: 753 against this run's 714, a ratio of 4.7x against 4.6x. 'Never goes off' is no longer single-run. Two METHOD entries. A detector that can fire on a single frame will fire on the wrong one. The A/B's first pair was void because the title detector tested one frame against a glyph threshold and the intro movie throws sub-second green flashes of 1298..5433. The presses were real and skipped the movie, so both legs returned a clean, symmetric, meaningless result -- a void test that looks like it ran is worse than one that errors. Same shape the corpus already recorded for screen_id.py calling the SQUARE ENIX logo 'title'. Twice paid for. The rule is that a screen detector matches a signature over time, and a broken run's own series is the cheapest control for its replacement. A demand for reproducibility can surface a defect that is not the one demanded. The literal answer to 'your figures are unverifiable' was 'here is the predicate', after which they verified exactly -- but writing the method down is what exposed the cross-geometry floor comparison, which nobody was looking for. And both sides were wrong at once: the challenger's counts were the wrong measurement AND the published figure had a real flaw. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Wuu56cE8vJGTBtn1ppsk8v
173 lines
8.5 KiB
Markdown
173 lines
8.5 KiB
Markdown
# The `PRESS Ⓐ` plate **pulses**, continuously — it does not blink once
|
||
|
||
**Classification: measured.** Xenia Canary, 2026-08-30, one no-input boot, two
|
||
separate title windows. Answers the port agent's ask #1, and the answer **deletes
|
||
nothing** — it says the port is currently wrong on the boot's end state.
|
||
|
||
## The question, and why it mattered
|
||
|
||
The port had removed a speculative `looping_focus_records` pulse, leaving a
|
||
renderer that *"shows a flash and nothing after"* — reasoning from `ptbtn00`, whose
|
||
keyframes make it opaque for 8 units (t=236–238, gone by 244). It asked for the
|
||
observed shape, saying this was the one ask that could **delete** an authored entry
|
||
rather than confirm one.
|
||
|
||
It confirms one. **Keep the pulse.**
|
||
|
||
## What the running game does
|
||
|
||
Held at the title with **no input at all**, the plate oscillates continuously for as
|
||
long as the title is up — 58 s in one window and 57 s in another, ~23 cycles each,
|
||
with no decay and no settling.
|
||
|
||
| | window | duration | glyph px | period |
|
||
|---|---|---|---|---|
|
||
| run 1 | t = 254…312 s | 58 s | **714 … 1520** | **2.530 s** |
|
||
| run 2 | t = 83…140 s | 57 s | **714 … 1520** | **2.540 s** |
|
||
|
||
Two windows of one boot, agreeing to **0.4 %**. Evidence:
|
||
[`../data/plate-pulse-timeseries.txt`](../data/plate-pulse-timeseries.txt) ·
|
||
[peak frame](../captures/title-builds/live-title-plate-pulse-peak.png) (1520) ·
|
||
[trough frame](../captures/title-builds/live-title-plate-pulse-trough.png) (714).
|
||
|
||
## 🔴 What exactly is being counted — the port could not check this page, and was right to say so
|
||
|
||
This page quoted 159 / 714 / 1520 without naming either the **region** or the
|
||
**threshold**. The port agent has `live-title-build4-no-plate.png` and still could
|
||
not reproduce any of them: counting `green > 150…200` over a 513×48 plate box it
|
||
got 528 / 4 908, 3–5× these numbers at every threshold. **A published figure that
|
||
looks checkable and is not is worse than one that is obviously incomplete.**
|
||
|
||
Stated properly, so it can be checked:
|
||
|
||
* **Region: the whole 1280×720 frame.** Not a plate crop.
|
||
* **Predicate: `is_title.py`'s**, byte-identical —
|
||
`(g > 130) & (g − r > 45) & (g − b > 45)` — a *three-channel* test, which is why
|
||
it counts far fewer pixels than a bare `green > N`.
|
||
|
||
Re-run against the committed frames, this reproduces 1520 / 714 / 159 exactly.
|
||
|
||
### 🔴 And fixing that exposed a real weakness in the floor
|
||
|
||
The **159** came from `live-title-build4-no-plate.png`, which is **1279×675** — the
|
||
game surface — while the pulse frames are **1280×720**, the whole display. Those are
|
||
not the same crop, so the floor was being compared across geometries.
|
||
|
||
✅ **Replaced with a same-run, same-geometry floor.** In run 1 the counter sits flat
|
||
at **154** for ~2 s (t = 253.0…254.8) immediately before the plate ramps in at
|
||
t = 255.0 — the title art alone, in the very frames the pulse was measured in.
|
||
|
||
| | count | provenance |
|
||
|---|---|---|
|
||
| title, plate absent | **154** | same run, same geometry, 2 s before the ramp |
|
||
| pulse floor | **714** | 4.6× the plate-absent level |
|
||
| pulse peak | **1520** | |
|
||
|
||
**"It never goes off" now rests on one run in one geometry**, which is what it
|
||
should have rested on from the start.
|
||
|
||
### ✅ …and it now has a witness that is not this run
|
||
|
||
With the predicate named, the port agent reproduced the floor **exactly** and
|
||
counted an independent capture from a different session:
|
||
|
||
| | this run | the port, from `live-title-press-a.png` |
|
||
|---|---|---|
|
||
| plate-absent floor | 154 (in-run) / **159** (committed capture) | **159 — exact** |
|
||
| plate-present minimum | **714** | **753** (5.5 % apart) |
|
||
| ratio | **4.6×** | **4.7×** |
|
||
|
||
The second row is the one that matters: a capture taken in a different session,
|
||
counted by someone else, lands within 5.5 % of this run's pulse floor and gives the
|
||
same ratio to one part in fifty. **The load-bearing claim is no longer single-run.** ⚠️ The port's independent ratio — about
|
||
**1 : 10.4–10.9** between plate-absent and plate-present over its own crop, stable
|
||
across a wide threshold band — brackets this page's 1 : 9.6 and is the part that is
|
||
robust to how anyone counts.
|
||
|
||
## The instrument, and its controls
|
||
|
||
`tools/re-capture/plate_timeseries.py` counts the green Ⓐ-glyph pixels — the same
|
||
counter as `is_title.py`, byte-identical. **Controls run before it was pointed at
|
||
anything unknown**, on committed captures:
|
||
|
||
| capture | counted | documented |
|
||
|---|---|---|
|
||
| `live-title-press-a.png` | **753** | 753 ✅ |
|
||
| `live-main-menu.png` | **327** | 327 ✅ |
|
||
| `live-title-build4-no-plate.png` | **159** | — (⚠️ 1279×675, a different crop — see above) |
|
||
|
||
⚠️ That third one is the reason the result is readable at all: **the plate-absent
|
||
floor is non-zero, not 0** — 154 in-run, 159 on the differently-cropped capture — the title art carries green pixels of its own. During the
|
||
pulse the count never goes below **714**, four and a half times the floor, so *the
|
||
plate is never absent*. It dims and brightens; it does not blink off.
|
||
|
||
### 🔴 One estimator replicates and one does not — and the failing one says so
|
||
|
||
| | run 1 | run 2 | agrees? |
|
||
|---|---|---|---|
|
||
| upward mid-crossings | **2.530 s** | **2.540 s** | ✅ 0.4 % |
|
||
| single-sinusoid least squares | 2.553 s | 2.413 s | ❌ 5.8 % |
|
||
|
||
Both were controlled on synthetic sinusoids at 2.24 / 2.55 / 3.10 s laid on the
|
||
**actual sample timestamps**, and both recovered every one exactly — so neither is
|
||
broken. The sinusoid fit is **misspecified**: the waveform is a fast rise and slow
|
||
decay, not a sinusoid, and its own variance-explained ($r^2$ = 0.468 and 0.228) is
|
||
the tell. **Take 2.535 s**, from the estimator that does not assume a shape.
|
||
|
||
⚠️ Indexing by sample position rather than timestamp is the documented trap here
|
||
([`ui-clock-freezes-at-settle.md`](ui-clock-freezes-at-settle.md), where it returned
|
||
48 against a true 51); both estimators use the timestamps.
|
||
|
||
## 🟡 The wall-clock is 13 % longer than the corpus's, and that is expected
|
||
|
||
[`ui-record-loop-length.md`](ui-record-loop-length.md) decodes the plate's cycle as
|
||
**120 units declared**, and earlier runs measured 2.12 / 2.19 / 2.34 / 2.31 s
|
||
(mean **2.24**). This run gives **2.535**.
|
||
|
||
| | implied units→seconds factor |
|
||
|---|---|
|
||
| declared 120 units at a nominal 60 units/s | 2.000 s |
|
||
| corpus's earlier runs | ×1.12 |
|
||
| **this run** | **×1.27** |
|
||
|
||
Same declared number, different emulator pacing — this container was under load.
|
||
✅ **This corroborates the existing instruction rather than disturbing it: author
|
||
the 120 units, never a second count.** Any wall-clock figure from this container is
|
||
pacing-dependent and is not comparable across runs without an independently
|
||
measured pacing factor, which this run does not have.
|
||
|
||
## Reach
|
||
|
||
⚠️ **Boot title or attract title — this does not say.** The first window opens at
|
||
t≈255 s from launch, and Q9's no-input baseline puts the title at ~193 s, so run 1
|
||
may already be an attract-loop title rather than the boot's first. Both windows are
|
||
"the title, held, with no input", which is what was asked; neither is proof about
|
||
the *first* appearance specifically.
|
||
|
||
## ✅ Why the floor is 714 and not 159 — the port's mechanism, and it fits
|
||
|
||
The port agent implemented the pulse and reports the shape that reproduces these
|
||
two levels: a **steady base plus a pulsing glow**, not a glow alone.
|
||
|
||
* `ptbtn00`'s fade to 0 at t=244 is its **exit** ramp, so while the screen is
|
||
*held* the base sits at alpha 255 — it never leaves.
|
||
* `ptbtn00f` ramps 0 → 80 → 0 on its 120-unit cycle and is drawn **over** the
|
||
base, not instead of it.
|
||
|
||
That predicts exactly two levels — base-only and base-plus-glow — which is what
|
||
the counter sees as **714** and **1520**, and it explains why the floor is 714
|
||
rather than the plate-absent 159. ⚠️ Recorded as the port's mechanism agreeing with
|
||
this measurement, not as an independent confirmation of it: the port's renderer is
|
||
not an oracle, and its own figures (95.68 / 115.52 in the plate region) are its
|
||
render's, not the game's. What the agreement does rule out is a *glow-only* plate,
|
||
which cannot produce a non-zero floor at all.
|
||
|
||
⚠️ **The glyph count is a thresholded pixel count, not an alpha.** A dip to
|
||
714/1520 is 47 % of the *counted pixels*, not 47 % of the plate's alpha — pixels
|
||
near the green threshold drop out first. Do not read a duty cycle or an alpha ramp
|
||
off these numbers; the shape (periodic, non-decaying, never absent) is what is
|
||
measured.
|
||
|
||
⚠️ **One boot.** Two windows inside it are not two boots, and the pacing factor is a
|
||
property of the run.
|