fabi/Sylpheed
1
0
Fork 0
Code Issues 9 Pull Requests Actions Packages Projects Releases Wiki Activity

re: a second JP capture closes the transfer -- and corrects my own noise claim

Browse Source 
Last turn I transferred the EN title's capture noise to the JP box and flagged
the gap: build 7 carries ptloop01/02.rat which may animate inside that region
where the EN plate does not, and the era adjudication rests on a single capture.
Took a second, independent capture from a fresh boot in a separate session
(jp_title_session.sh -- sets ja, captures, always restores en; verified back at
language=1).

Within-run stability reproduces: 0 of 138 600 px in the ROI across four
comparisons, with 47k-73k px moving whole-frame as the contrast control.

BETWEEN SESSIONS, inside the box the adjudication uses: 645 of 164 124 px
differ, RMSE 0.3215, against 116 492 px whole-frame -- genuinely different
sessions. And the verdict reproduces to three decimals: stale 58.412 -> 58.413,
fixed 41.690 -> 41.692, margin 16.722 -> 16.721.

The shape is the useful part: capture noise moves both candidates together, so it
nearly cancels in a MARGIN. Absolute scores moved 0.001-0.002 while the margin
moved 0.001 against an in-box noise of 0.32. A margin between two renders scored
on one capture is far more robust than either score is.

CORRECTION to a claim I made earlier today and sent to the port: I said the
settle-vs-rest negative was STRENGTHENED because 1.48 sits below the whole-frame
capture spread of 2.8. Wrong comparison -- the measurement lives in the box, and
in-box between-session noise is 0.32, so 1.48 is well above it. The negative
rests on the render axis alone (1.2, ratio 1.2x), exactly as first stated. I
reached for a number that was to hand rather than the one that applies, which is
the same family as the errors we have both been cataloguing.

METHOD gains: match the noise floor to the quantity, including which noise
applies; and sylpheed-port's point that an instrument which rounds away the thing
being verified cannot verify it (they called a harness reproducible from an RMSE
printed to two decimals when the residual was 0.0565).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Wuu56cE8vJGTBtn1ppsk8v
This commit is contained in:
sylph-decoder
2026-08-30 15:24:56 +00:00
parent f16338bc88
commit 281a058aca
4 changed files with 117 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
docs/re/METHOD.md
View File

@@ -331,6 +331,20 @@ agent's loop prompt, i.e. nowhere durable. See [`README.md`](README.md) for the
noticing that what you are about to look at merely *correlates* with what you
want to know.
* **Match the noise floor to the quantity — including which noise actually
applies.** A margin needs a floor, but the floor must be the one that moves
*that* margin. Two renders scored against one capture share the capture, so
capture noise largely **cancels**: measured on the JP title, the absolute scores
moved 0.001–0.002 between sessions while the **margin** moved 0.001, against an
in-box capture noise of 0.32. Judging that margin against a whole-frame capture
spread — a number that was simply to hand — made a non-decisive result look
decisive, and this corpus published that for part of a day.
⚠️ Relatedly, `sylpheed-port` verified a harness "reproducible" from an RMSE
**printed to two decimals** when the residual was 0.0565: **an instrument that
rounds away the thing being verified cannot verify it.** Check the printed
precision against the quantity before quoting the number, and prefer comparing
*frames* to comparing a statistic about them.
## Runtime / emulator
* **Look at the PNG** — and check its dimensions.

31
docs/re/data/jp-title-at-rest.txt
View File

@@ -43,3 +43,34 @@
# the ptlogo3a/b/c + ptlogo_back2eff* stack, which ui-forced-backdrop and
# the tie-break work record as TRANSPARENT at rest on the English title.
# That is exactly the region the port's drift is localized to.
################################################################################
# SECOND, INDEPENDENT CAPTURE -- 2026-08-30, a fresh boot in a separate session.
# tools/re-capture/jp_title_session.sh (sets ja, captures, ALWAYS restores en).
#
# Why: the five frames above demonstrate the logo stack is at rest WITHIN a run.
# They say nothing about the axis sylpheed-port's drift was on -- BETWEEN runs,
# where a free-running clock lands somewhere else on a fresh boot. Nothing in
# this corpus covered that axis, and the ptlogo_eff3 era adjudication rests on
# a single capture.
#
# Within-run stability REPRODUCES in session 2:
# ROI 350x396 at (405,74): frames 1-4 vs 0, max |d| 0, 0 / 138 600 differing
# whole frame contrast: 47 278 / 51 034 / 58 922 / 72 958 px moving
#
# BETWEEN SESSIONS, inside the 388x423 box the era adjudication uses:
# 645 of 164 124 px differ, max |d| 41, RMSE 0.3215
# whole frame, for contrast: 116 492 px differ, max |d| 51
# -- so the two captures are genuinely from different sessions.
#
# THE VERDICT REPRODUCES TO THREE DECIMALS:
# vs session-1 capture vs session-2 capture
# stale era rest (108,72) 58.412 58.413
# fixed era rest (98,42) 41.690 41.692
# margin 16.722 16.721
#
# 📌 And the shape of that is the useful part: capture noise moves BOTH
# candidates together, so it very nearly cancels in a MARGIN. The absolute
# scores moved 0.001-0.002 between sessions and the margin moved 0.001, against
# an in-box capture noise of 0.32. A margin between two renders scored on one
# capture is far more robust than either score is.

43
docs/re/structures/ui-resting-pose.md
View File

@@ -93,16 +93,47 @@ the two captures differ by **83 496 pixels** whole-frame (max |d| 174) — they
genuinely different grabs — and by **0** inside the box. The screen's free-running
element is the plate, which lies outside the logo region the adjudication uses.
### ✅ And a second JP capture, from a fresh session, closes the transfer
The paragraph above transferred the *EN* title's capture noise to the *JP* box and
flagged that as a limit. A second independent capture was taken —
[`jp_title_session.sh`](../../tools/re-capture/jp_title_session.sh), fresh boot,
separate session, locale set and restored —
[`live-title-jp-at-rest-run2.png`](../captures/title-builds/live-title-jp-at-rest-run2.png).
Within-run stability **reproduces** (0 of 138 600 in the ROI, 47k–73k px moving
whole-frame as the contrast control). Between sessions, inside the box this
adjudication uses: **645 of 164 124 px differ, RMSE 0.3215**, against 116 492 px
whole-frame — genuinely different sessions.
**The verdict reproduces to three decimals:**
| | vs session-1 capture | vs session-2 capture |
|---|---|---|
| stale era rest `(108,72)` | 58.412 | 58.413 |
| fixed era rest `(98,42)` | 41.690 | 41.692 |
| **margin** | **16.722** | **16.721** |
📌 **Capture noise moves both candidates together, so it nearly cancels in a
margin.** The absolute scores moved 0.001–0.002 between sessions while the margin
moved 0.001, against an in-box capture noise of 0.32. A margin between two renders
scored on one capture is far more robust than either score is — which is why the
render axis, not the capture axis, is the binding constraint here.
**So the margins re-stated against every noise floor now measured:**
| comparison | margin | vs render noise (1.2) | vs capture noise (0.0 in-box, 2.8 whole-frame) |
| comparison | margin | render noise (1.2) | in-box capture noise (0.32) |
|---|---|---|---|
| stale vs fixed era | **16.7** | 14× | ≥5.9× |
| settle vs rest | **1.5** | 1.2× | below whole-frame noise |
| stale vs fixed era | **16.72** | **14×** — decisive | 52× — decisive |
| settle vs rest | **1.48** | **1.2×** — NOT decisive | 4.6× |
The era result survives on both axes. The settle-vs-rest negative is **stronger**
than first stated: 1.5 is not merely inside the render plateau's flatness, it is
below the whole-frame capture spread as well.
🔴 **Correction to a claim made earlier the same day.** This page briefly said the
settle-vs-rest negative was *strengthened* because 1.48 sits below the
**whole-frame** capture spread of 2.8. That was the wrong comparison: the
measurement lives in the box, and the in-box between-session noise is **0.32**, so
1.48 is well above it. The negative rests on the **render** axis alone, exactly as
first stated — the "strengthening" was reaching for a number that was to hand
rather than the one that applies.
⚠️ **Reach of the transfer.** This is the *EN* title's capture noise applied to the
*JP* title's box. Build 7 carries `ptloop01/02.rat`, which may animate inside that

35
tools/re-capture/jp_title_session.sh Executable file
View File

@@ -0,0 +1,35 @@
#!/usr/bin/env bash
# Drive a JAPANESE title capture end to end, and ALWAYS put the locale back.
#
# The first JP capture (2026-08-30) demonstrated the logo stack is at rest
# WITHIN a run: five frames ~1.5 s apart, byte-identical over the port's ROI,
# against a whole-frame contrast control showing 5-8 % of the screen moving.
# That does not test the axis sylpheed-port's drift was on -- BETWEEN runs, where
# a free-running clock lands somewhere else on a fresh boot. This script exists
# to take a second, independent capture so that axis can be measured.
#
# Usage: jp_title_session.sh OUTDIR [wait_s]
set -u
export HOME=/sylph-home/re SDL_AUDIODRIVER=dummy DISPLAY=:98
SD="$(cd "$(dirname "$0")" && pwd)"
OUT="${1:-/sylph-home/re/jp2}"; WAIT="${2:-700}"
mkdir -p "$OUT"
restore() {
echo "-- restoring locale --"
python3 "$SD/set_console_language.py" en || echo "RESTORE FAILED -- check by hand"
python3 "$SD/set_console_language.py" --show
}
trap restore EXIT INT TERM
ps -o pid= -C xenia_canary | xargs -r kill -9 # kill by NAME: `pkill -f` matches this shell
python3 "$SD/set_console_language.py" ja
python3 "$SD/set_console_language.py" --show
( cd "$OUT" && nohup run-canary --mem_watch=false \
--logged_profile_slot_0_xuid=B13EBABEBABEBABE \
>"$OUT/canary.stdout" 2>"$OUT/canary.stderr" & )
sleep 8
timeout $((WAIT + 120)) python3 "$SD/jp_title_capture.py" "$OUT" "$WAIT"
echo "CAPTURE STEP DONE"
ls -la "$OUT"/*.png 2>/dev/null | head