re: F5 -- A snaps the plate, and does NOT advance the artwork's clock

Three captures, aligned by the sweep's position rather than by frame number.
The plate's alpha goes 23,46,69,92,115,139 over ~11 frames with no input,
and 255 in a single frame when A is pressed: zero intermediate values
against eleven. A cuts, it does not accelerate -- the human's prior holds.

The control is what makes it readable. Four elements ramp out right around
the press and I could have reported that as the effect; at sweep x=-1.42 all
three runs agree quad for quad, so those exits are the ordinary build-in.

Refutes clock:"shared" on F4's own discriminator: the artwork keeps
animating across the press, frame for frame identical to the control. Reach
stated -- that half rests on a 5-frame window, since everything else on
screen is already at 255 and cannot discriminate.

Also records the first attempt's miss: blind wall-clock timing put the press
after the settle, where A is accepted instead. The probe now gates on the
sweep appearing, which is the parent's declared t=70..100 gate.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Jc4pciRArGHfxGGhEbwp5t

tools/re-capture/f5_snap_or_accelerate.sh

@@ -42,15 +42,27 @@ echo "logging confirmed at ${SECONDS}s (${sz} bytes)"
 
 # --- F5: a SECOND A during the title build-in ---------------------------
 # Discriminator (brief): a snap shows no intermediate alphas, an acceleration
-# does. Press 2 must land while the build-in is still running -- after the
-# sweep is up (title t>=100) and before the settle (~t=236). Blind timing is
-# the weak point, so the press instant is RECORDED and checked post hoc; a
-# press that lands late is reported as such, not silently reinterpreted.
-PRESS2="${PRESS2_DELAY:-16}"
-sleep "$PRESS2"
-echo "PRESS2 at ${SECONDS}s  (log frame marker below)"
-grep -c '^--- frame' "$OUT"/xenia_re_ui_draws_*.log | tail -1 | sed 's/^/frames at press2: /'
+# does. Press 2 must land while the build-in is still running.
+#
+# ⚠️ BLIND WALL-CLOCK TIMING DOES NOT WORK HERE and the first run proved it:
+# +16 s put the press at log frame ~1217 when the title had already settled at
+# ~653, so A was accepted instead of accelerating anything. The build-in is
+# only ~180 log frames wide and run pacing varies 2x. So GATE ON AN OBSERVABLE:
+# wait for the sweep's texture page to appear (that IS title t=70..100, the
+# parent's declared gate opening), then let a fixed number of frames pass.
+LOG=$(ls "$OUT"/xenia_re_ui_draws_*.log 2>/dev/null | head -1)
+SWEEP=8154424FFC48FE61
+for _ in $(seq 1 900); do
+  grep -q "$SWEEP" "$LOG" 2>/dev/null && break
+  sleep 0.2
+done
+grep -q "$SWEEP" "$LOG" 2>/dev/null || { echo "FATAL: sweep never appeared -- title build-in not seen"; pkill -x xenia_canary; exit 1; }
+n0=$(grep -c '^--- frame' "$LOG")
+echo "sweep up at log frame ${n0} (title t~70..100) at ${SECONDS}s"
+target=$((n0 + ${PRESS2_FRAMES:-40}))
+until [ "$(grep -c '^--- frame' "$LOG")" -ge "$target" ]; do sleep 0.2; done
+echo "PRESS2 at log frame ~${target}, ${SECONDS}s"
 pad "press=A"; sleep 0.4; pad ""
-sleep 90
+sleep 60
 pkill -x xenia_canary
 echo "done at ${SECONDS}s"; ls -la "$OUT"/*.log 2>/dev/null