Replace the fixed 60s idle timer with video-completion pacing: when the base loop
video finishes one playthrough (the timeupdate wrap point, or the ended safety
net), drift one altitude deeper (descend, wraps) — gated by the default-on toggle
and deferred during a manual drag or a playing morph. Removes the whole
AUTO_ADVANCE_MS/scheduleAutoAdvance/bump timer machinery and its interaction
resets. loop-recovery.spec disables the toggle to test the loop-back path in
isolation. auto-advance.spec reworked to the completion-driven model.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Replace the fixed 5-min rotation with end-driven shuffle: each track plays through,
and a short crossfade a few seconds before it ends rolls a random OTHER track
(pickNextIndex, no immediate repeat) — so Music is always playing while selected.
An 'ended' handler is the safety net (short track / missed timeupdate) and a
watchdog resumes an unexpectedly-paused active track. Crossfade injectable via
window.__hefMusicXfadeMs. music-playlist.spec.ts reworked to the end-driven model.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The opening screen only had Volume + Video speed — no way to pick the sound mode
before launching. Add a welcome Audio dropdown (None/Soundtrack/Music) mirroring
the panel; applyWelcomeToPanel carries the choice into the session on launch, None
greys the welcome Volume, and ?audio= presets it on the welcome screen too. +4 e2e
(welcome-audio.spec.ts).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Replace the single ethereal loop with the operator's chosen set — Music for
Manatees, Dewdrop Fantasy, Eastminster, Overheat, Concentration (Kevin MacLeod,
CC BY 4.0, drumless). The Music source now plays one track and every 5 min
crossfades to a random OTHER track (pure HEFScrub.pickNextIndex, never an
immediate repeat), altitude-independent, reusing the two audio elements; both
primed in-gesture for Safari. Intervals injectable via window.__hefMusicRotateMs
/ __hefMusicXfadeMs.
Media loudness-normalized + transcoded under the 25MB LFS limit (build_audio_media
MUSIC), committed via LFS; ethereal loop retired. CC-BY attribution added to
credits.html. Docs updated (music-candidate-pool = final set; audio-candidate-pool
ethereal section superseded).
+2 node unit (pickNextIndex), +2 e2e (music-playlist), audio-source spec updated.
40 unit + 49 e2e green; live probe confirms playback + rotation.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
?audio=none|soundtrack|music (and ?vol=0..10) presets the Audio source/volume on
load, so a link opens straight into a chosen sound mode — e.g. share …/?audio=music
with someone who'd prefer music, …/?audio=soundtrack for the ambience. Invalid
values ignored (default Soundtrack stands). +5 e2e (url-params.spec.ts).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
11 CC0 stereo binaural/entrainment candidates (verified on freesound) for
evolving the Music option into a crossfading stress-relief set. Durable record
in docs/binaural-candidate-pool.md; interactive audition gallery at
simulator/static/review_binaural.html (gitignored local-only, like the other
review galleries — streams the CC0 previews, tick-to-keep + copy picks).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
After the dial sits idle for 60s the altitude drifts one step deeper (descend;
wraps past the deepest back to the top), like a slow kiosk tour. Any manual
interaction (drag/wheel/arrow-key/knob) restarts the idle countdown; a mid-morph
or mid-drag beat is skipped and retried. A default-on 'Auto-advance' toggle in
the Altitude panel disables it. Interval is injectable for tests via
window.__hefAutoAdvanceMs. i18n for 4 langs. +3 e2e (auto-advance.spec.ts).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Widen the panel (280->300px), keep control labels on one line (white-space:
nowrap), and let the range shrink (min-width:0) instead of pushing the label
to a second row. +1 e2e (panel-layout.spec.ts).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The mood grade was dropped while a morph played (busy): the WebGL canvas set
filter 'none' and the fallback set #vid's filter 'none' on morph load, so the
transition showed ungraded and the grade snapped back only on landing. Now the
grade rides through the morph on both paths (the painterly DREAM stays off mid-
morph, as before, to keep it crisp). +1 e2e (mood-grade.spec.ts).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The Music-option asset produced by build_audio_media.py: deadrobotmusic
'Ambient F Sharp Minor Ethereal Choir Pad' (CC0), sourced from the freesound
hq preview and looped/loudness-normalized to ethereal.loop.mp3 (3:34). Both
the raw source and the served loop are committed via git-LFS, matching the
per-scale soundtracks; served at /media/audio/music/ethereal.loop.mp3.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Split the single 0–10 'Audio' slider into two controls:
- Audio — a dropdown selecting the source: None / Soundtrack / Music
- Volume — the renamed 0–10 slider (master gain for the active source)
Music is one ethereal, relaxing ambient loop played the same at every
altitude (no per-altitude crossfade), per the innerflo.me brief. Sourced
CC0: deadrobotmusic 'Ambient F Sharp Minor Ethereal Choir Pad' (freesound
#808032); added to build_audio_media.py, the candidate-pool doc, and the
public credits colophon. Media stays gitignored; run build_audio_media.py.
Routing lives in a pure, node-tested HEFScrub.audioPlan(source, level, frac);
app.js's restAudio/blendAudio consult it. None greys out Volume. The welcome
screen's mirror level slider is relabeled Volume too. i18n for all 4 langs.
Spec: docs/superpowers/specs/2026-07-01-audio-music-option-design.md
Tests: +5 node unit (audioPlan), +4 e2e (audio-source.spec.ts); 32 related
e2e green, 38 node unit green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Build default app-path → /human-machine-strata; deploy.sh APP_PATH default too.
_write_redirects now emits a permanent redirect /human-experience-simulator/* →
/human-machine-strata/:splat so old links still land. Media (R2 bucket) + Pages
project names are unchanged (not in the benstull.art URL). Updated the build
test (asserts the legacy redirect) + static-build e2e path + README paths.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Page <title>, on-page <h1> (app.title — kept as the proper title, same across
all 4 languages), the About + Credits headings/prose, and the author dev tool.
"Strata" = the layered cosmos→abyss descent; the slash = the human↔machine
membrane the work is about.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
On mobile the panel wraps below the (sticky) stage; putting the Altitude knob
first means it sits right under the video instead of behind Output + knobs.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
With reduce motion on, selecting a new altitude jumps (autoScrub's reduced
branch → setPos) and skips the segment that would re-roll the destination clip,
so activeClipId stayed on the previous altitude's clip and the video didn't
change. Self-heal at the landing point: if the landed altitude's pool doesn't
contain activeClipId, pickPoolClip a fresh one. TDD: new altitude-lock e2e.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per operator: Think/Feel/Mood group sits above the Altitude knob; remove the
'Turn the knob…' altitude hint (and its dead i18n key); 'Experience knobs (0–4)'
→ 'Experience knobs' (all 4 langs).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Same root cause that retired reverse playback: the resting loop clips aren't
all-intra, so playbackRate >2x makes Safari stutter on inter-frame seeks. Narrow
both sliders (welcome + panel) + the tick datalist to 0–2x; default still 1x.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
These were accidentally committed (tracked on main); they are regenerated on
every run and shouldn't be in git.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per operator: Speed only controls the resting loop, so reduce-motion is still
needed to turn off the morph/transition videos. Re-add a visible 'Reduce motion'
toggle on BOTH the welcome screen and the panel (mirrored, default from OS
prefers-reduced-motion, remembers the choice). When ON, altitude changes are
instant cuts instead of an animated morph tween (autoScrub's reduced branch);
the resting loop is unaffected — it follows the Speed slider (0x freezes it).
Decouples the loop from reduce-motion (was: main froze both videos + disabled
the slider under OS-reduced). playLoop()/landing/applyPlaySpeed no longer check
isReduced(); isReduced() now only gates the morph tween. Welcome controls are
now Language / Speed / Audio / Reduce motion. i18n: restore rm.label (4 langs),
reword warn.body to point at the on-screen control.
Tests: a11y Task 5 reworked (speed independent of reduce-motion; new instant-cut
behavior test); welcome defaults assert reduce-motion off. node 35, e2e 32,
static 2, pytest 301 (3 pre-existing ffmpeg/cv2 env fails).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Real-browser eyeball: below-0x reverse jittered because the base loop
clips aren't all-intra (the flagged risk). Narrow the slider to 0x–4x
forward (still continuous, step=any); 0x still freezes; OS reduced-motion
behavior unchanged. Removes the manual rAF reverse loop and the
HEFScrub.reverseStep helper + its unit tests.
Tests: node unit 10/10; a11y e2e 12/12 (Task 5 now 3 slider tests).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Continuous -2x..2x slider drives ONLY the resting altitude loop video
(#vid-loop); the morph/transition video is scrub-driven and untouched.
Forward = native playbackRate; ~0x = freeze; below 0x = a manual rAF
reverse loop (native negative playbackRate is unreliable). Reverse math
is the pure, node-tested HEFScrub.reverseStep helper.
Reduced-motion accessibility is preserved but now derives live from the
OS prefers-reduced-motion setting (no visible toggle): when active the
loop stays frozen and the slider is disabled. Retires hef.reduceMotion.
i18n: speed.label (en/es/fr/ja); warn.body reworded to point at the OS
setting.
Tests: +4 node unit cases for reverseStep (14/14 green); a11y e2e Task 5
rewritten for the slider (13/13 green); static-build e2e green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Per docs/superpowers/specs/2026-06-30-welcome-screen-design.md. Fold the bare
#loading splash, the #motion-warning modal, and the #run-sim button into one
#welcome overlay — the universal entry gate:
- State A (.welcoming): heads-up notice + four output controls (Video ON,
Reduce motion OFF, Audio 2, Language) + 'Launch Simulator' button, with the
'Loading Universe' progress pinned at the bottom (media preloads behind it).
- State B (.loading): pressing Launch while still loading hides the messaging +
button, keeps the controls, and centers the loader; entry happens automatically
once phase-1 finishes. Launch when already loaded enters directly.
- The right .panel is hidden until entry (body:has(#welcome)); welcome control
values carry into the panel inputs on entry. Language + Reduce motion drive the
shared state live (both selects/checkboxes mirror); Video + Audio apply on entry.
- Launch is the single audio-unlock gesture (beginExperience runs in-gesture even
when entry is deferred). Retire WARN_KEY (welcome shows every load); entry uses
the chosen Audio level rather than forcing 2.
- i18n: new welcome.launch key + es/fr/ja for the heads-up (warn.title/body),
reworded to point at the on-screen Reduce-motion control; drop dead run.button.
Tests: new welcome.spec.ts (defaults, live language re-render, State B
deferred-entry, straight-in when loaded, control carry-over, panel hidden);
update altitude-lock/a11y/i18n/loop-recovery/static-build e2e + the Python
audio/credits e2e to the welcome flow; add a window.__hefReady diagnostic seam.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A drag is a way of grabbing the altitude dial, not a free-standing encoder
detent. Until now onDialUp held the live blend wherever the knob stopped (the
"no auto-complete" model), so a grab-and-release could leave the experience
parked mid-morph between two altitudes.
Now a real turn (moved >= 6) auto-scrubs the short way to Math.round(pos) and
settles there (frac 0 locks) — a grab-and-release always lands on a discrete
altitude, exactly like a tap or label click. The tap path (moved < 6 ->
jumpToScale) is unchanged, and reduced motion is handled by autoScrub (jump to
target, no tween).
Adds an altitude-lock e2e asserting a partial drag settles on an integer
altitude after release. Full altitude-lock (14) + a11y + i18n suites green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Gate the universe on one clip per altitude + connecting morphs (~126 MB, ~10-20s),
then grow the pool in the background. The random pick (pickPoolClip/pickRandomMember)
is restricted to fully-loaded clips — base + connecting morphs both ways cached
(HEFPreload.eligibleDestinations) — so a transition never stalls on un-downloaded
media and new clips become reachable only once fully loaded. Pure planner in
preload.js (+ node tests).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Integrates the fully-static Cloudflare publish line (live at
benstull.art/human-experience-simulator) into main: the static build (Pages + R2,
manifest-only media tree, baked API JSON), the "Run simulation" start button, the
public CC-BY/BY-SA credits page with the "License & reuse" CC BY-SA 4.0 derivative
declaration (Option A + D1), the repeatable Cloudflare deploy script, and assorted
load/audio fixes (progressive boot, scaleAudioUrl via mediaBase).
Clean auto-merge with main's per-clip affect/feelings + i18n work — only app.js and
altitude-lock.spec.ts overlapped, both auto-resolved. Verified green on the merged
tree: node 29, pytest 315 passed / 2 skipped.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Clamp each per-altitude step to the WCAG 2.3.1 floor (no-op at 1200ms, guards
a future lowering). Audit: #black fade + audio crossfade can't exceed 3/sec, no
clamp needed. boot() now seeds the gate-dismissed flag (returning visitor) so
the run-sim test isn't blocked by the new interstitial.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Affect was authored per SCALE — every clip in an altitude shared the same 4
emotion words. Now every clip carries its OWN feelings, read from that specific
footage (mood, motion, light), so e.g. the violent Tarantula nebula evokes "a
violent ferment" while the dying-star Southern Ring evokes "a star's slow dying /
a luminous farewell," and a drowsing seal haul-out evokes "a sun-warmed drowse."
Adds AFFECT_CLIP (per-clip override; falls back to the scale register) covering
all 41 clips — 41 distinct feeling-sets, one per clip. Keeps each scale's
emotional family as a base and swaps in 1-2 footage-specific feelings, reusing the
existing feel.* palette where it fits and adding 19 new nuance keys (sublime,
tenderness, turbulence, radiance, mortality, poignancy, eeriness, purity,
solitude, verdancy, freshness, repose, hush, poise, flow, intricacy, spectral,
alienness, strangeness) with full es/fr/ja. 42 distinct feeling keys now in use.
Drafted by viewing each clip's footage (per-scale passes); first-pass reads that
want operator by-eye review. Data-only (build_pool_manifest.py + manifest + i18n
catalogs). Verified: 34/34 catalog+i18n+build pytest, 16/16 node unit, 4-language
parity for every clip, live render shows distinct feelings per clip (incl. new
keys) with Think/Feel overlap still 0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The hold-last-frame settle gated playLoop() behind a canplay event that doesn't
re-fire on an already-loaded element → the next altitude's video never started.
Reverted to the simple swap, and gate the universe on a full preload (every base +
every morph) so navigation is smooth with no on-demand stall. Adds a ?debug on-screen
media-state overlay for diagnosing real-browser playback.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
role=slider + tabindex on the dial; Arrow/Home/End stepping; labels become
Enter/Space buttons; aria-valuenow/valuetext track the committed altitude via
renderDial. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Single guard in playLoop() holds a still frame; autoScrub jumps instantly;
applyReduceMotion pauses/resumes. Default from prefers-reduced-motion, then
the choice persists. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
s0031 removed "About this work" to avoid duplicating it on the separate About
page, but that section WAS the D1 compliance — it declares the altered work is a
derivative offered under CC BY-SA 4.0, which is what satisfies the ShareAlike
obligation for the BY-SA source clips. Without it the live site credited the
BY-SA authors but never declared the derivative share-alike.
Restore the declaration on the credits page, under the heading "License & reuse"
instead of "About this work" so it isn't confused with the separate About page
(which covers the project's intent). Tests assert the declaration is present and
that the heading is NOT "About this work".
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Adds the Reduce-motion toggle markup + i18n label in the same Output
fieldset (wired in a later task). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The CC BY-SA work-license declaration ("About this work") was on the credits
page, but the About page (built on the parallel feat/accessibility-pass) will
own that content; keeping it in both places would duplicate it. Remove the
section from credits.html and make the footage note self-contained (it still
states the footage is altered — the per-clip "indicate modifications" cue stays
with the attributions). Tests updated to assert the declaration is intentionally
no longer on the credits page. The .declaration CSS rule is left in place for the
About page to reuse.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The audio layer built the soundtrack url as a literal /media/audio/<f>, which 404s
on the static R2 build (audio err 4, silent). Route it through mediaBase() like the
video path: dev → /media/ (unchanged), static → https://static.benstull.art/. Verified
live: aud loads 200 from R2 and plays.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Denser left labels (now >=3/clip) plus the always-4 right affect words made the
two HUD layers collide. renderAffect now nudges each Feel word vertically off the
Think chip plates (and off already-placed words + the ANALYSIS status tag),
staying on-screen, keeping the least-overlapping spot if nothing is fully clear.
Feelings are scene-level (soft position) so we move THEM, never a chip off its
subject. Obstacles are computed from the manifest, not the DOM, and a tracked
label's path is sampled across the loop — so the clearance holds for the whole
loop even though the per-frame track re-render keeps moving the chips (no per-frame
re-layout, no jitter). New diagnostic seam window.__hefNoDeconflict (sibling of
__hefState) A/B-toggles it.
Verified across all 6 scales at max knobs: authored overlaps drop to 0 with none
introduced (cosmos 3->0), and a tracked abyss clip stays at 0 overlaps sampled
over loop time as its chip drifts. No affect-position test exists to break;
16/16 node unit + frontend pytest green.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Boot becomes ready on the first clip and streams the rest in the background
(was: blocked on the full ~2.3 GB before 'Run simulation' appeared). During nav,
the visible morph and landing base only swap to cached blobs — otherwise the
current frame holds (settleGen guards a stale settle) — so rapid altitude-clicking
never blanks the stage.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
CC BY / CC BY-SA footage needs its author + license shown to the viewer, and the
installation alters every frame (a derivative). Per-clip license/source already
shipped in clips.json but only surfaced in the dev panel and internal review
pages — the public experience showed nothing, so the static benstull.art site
was not attribution-compliant.
Option A — add a manifest-driven credits/colophon (credits.html + credits.js):
the page fetches the same clips the experience uses (baked clips.json in static,
the live API otherwise) and renders one attribution block per clip (title,
license, source), HTML-escaped, sorted, never silently dropping a clip. An
"ⓘ Credits & licenses" link in the header (localised en/es/fr/ja) makes it
reachable from the work — what CC's "reasonable to the medium" calls for. Audio
is all PD/CC0, credited as a courtesy block.
Option D1 — the colophon declares the altered work itself is offered under
CC BY-SA 4.0 and states the footage is modified, satisfying the ShareAlike +
"indicate changes" obligations for the BY-SA sources. Non-commercial.
Ships in the static build (added to PUBLIC_ASSETS). Verified: 41/41 clips
credited incl. all 7 CC-BY/BY-SA clips, 0 missing a license. Tests: node units
for creditEntries/creditsListHtml + escaping + page structure/declaration;
pytest build-output assertion; Playwright e2e (page lists attributions, header
links through).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Every clip now carries at least 3 factual LEFT annotations to match the 4 RIGHT
(Feel) affect words — fixes the "many clips missing Think labels" asymmetry
(right read "everywhere", left felt sparse). Adds one researched factual label to
each of the 23 thin clips (orbit_bluemarble had only 1; 22 had 2), with full
es/fr/ja parity via the catalogs, and converts the 4 still-static abyss creature
subjects (octopus, bigfin squid, dandelion siphonophore, sea pig) from fixed
boxes to 3-keyframe motion-tracked labels (tracked clips 7 -> 11).
Boxes/tracks are a first pass sampled by eye from base footage (t=0.05/0.5/0.95);
they want operator by-eye review in author mode to fine-tune positions. Remaining
static clips are landscapes (correctly static) plus 2 diffuse fish schools and 2
surface-animal scenes, available for author-mode tracking if wanted.
Data-only change (build_pool_manifest.py LABELS + manifest + i18n catalogs); no
app/engine code touched. Verified: 43/43 manifest+catalog+i18n pytest, 16/16 node
unit, live render shows 3 chips/clip in all 4 languages with no JS errors. (The 3
failing pytest are pre-existing cv2/ffmpeg-absent env limits, unrelated.)
Note: the reported "audio not playing" was a false alarm — exhaustive repro
(Chromium+WebKit, strict autoplay, both servers, RMS-verified non-silent files)
found the audio engine correct; the operator's Safari output was muted.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The experience boots silent and waits for a gesture (browser autoplay blocks
an un-gestured audio play()). Until now that gesture was an unlabelled Video
toggle — not obvious to a first-time viewer. Add a "Run simulation" button,
revealed over the black stage once media has preloaded; clicking it turns
Video on, snaps the audio dial to a gentle 2/10, dismisses the prompt, and
starts the soundtrack in that one gesture. Turning Video on directly still
begins the experience too (couples audio in, dismisses the button) via a
shared beginExperience(). Localised en/es/fr/ja.
E2E coverage updated (Playwright + pytest-playwright): button shown after
load, button starts video+audio+dismisses, and direct-Video-on path.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
BenStullsBets