feat(registry): newsreel animated map pack (5 blocks)

[miguel-heygen]

Summary

Expands the newsreel-style animated maps registry pack to 23 blocks (5 original + 18 new). Original stylized SVG/HTML/GSAP only — inspired by Moving Image Archive Animated Maps aesthetics; not affiliated; no footage or assets included.

Item	Trope
map-route-draw	Route stroke draw-on
map-territory-wipe	Territory fill wipe
map-city-pins	Labeled city pins pop/stagger
map-region-zoom	Grid overlay + camera zoom
map-campaign-arrows	Sequential schematic campaign arrows
map-front-line-crawl	Dashed battle/front line crawl
map-ink-blot-soak	Ink/blot territory soak
map-shipping-lanes	Sea/shipping lane dotted path
map-globe-inset	Inset rotating globe + main map (GSAP rotation, seek-safe)
map-stamp-labels	Stamp/typewriter location labels
map-border-chase	Border highlight chase
map-radar-rings	Concentric radar/range rings
map-boundary-morph	Historical boundary crossfade morph
map-compass-scale	Compass rose + scale bar reveal
map-itinerary-stops	Multi-stop itinerary with numbered markers
map-hotspot-pulse	Heat pulse / hotspot blink
map-parchment-fold	Paper-map fold / parchment aesthetic
map-split-theater	Split-screen two-theater maps
map-arrow-fanout	Arrow swarm / advance fan-out
map-grid-crosshair	Lat-long grid draw + crosshair lock
map-isoline-reveal	Isoline/contour reveal
map-conquest-fill-sequence	Sequential region conquest fill
map-legend-build	Legend panel assemble

Also updates registry/registry.json, registry/CREDITS-animated-maps.md, catalog MDX, docs/docs.json, and docs/public/catalog-index.json.

Test plan

• bun run lint:registry-items with all map-* names — 0 errors
• Spot-check seek/scrub on a few compositions (no rAF / Date.now / Math.random)
• Confirm cohesive sepia/newsreel palette + shared landmass silhouette

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated
hyperframes	🟢 Ready	View Preview	Sep 21, 2026, 10:58 PM

💡 Tip: Enable Automations to automatically generate PRs for you.

[miguel-heygen]

Closing: the animated-maps block pack is not going into the registry for now. Branch kept.

[jrusso1020]

Reviewed at d4ec111dc4f79bb6bb5d4dfb4364274beaa1cde7 (111 files, +16126/-6, base c09594fef).

The pack itself is in good shape — see the verified list at the bottom. Two findings are blocking, and both are in the same place: the provenance story this PR tells does not match the bytes it ships. This is the public OSS repo, so the credits file is the artifact downstream consumers read to decide whether they can reuse a block.

---

1. Blocking — "no footage or assets included" is not true at this head

The PR body says "Original stylized SVG/HTML/GSAP only … not affiliated; no footage or assets included." registry/CREDITS-animated-maps.md:5 repeats it: "Not affiliated; footage not included. Original HyperFrames GSAP/HTML compositions only."

Two of the three PNGs in the diff are photographic frames off the source films. I opened them rather than inferring from the filenames:

Asset	What it is
registry/blocks/map-border-chase/assets/basemap-t0.png	640×480 RGB, 10,867 distinct colours — a colour video frame of a physical relief map captioned ALASKA / FAIRBANKS / CANADA / DAWSON CREEK, with analog horizontal smear across "UNITED STATES"
registry/blocks/map-radar-rings/assets/hold-plate.png	320×240 8-bit grayscale, 220 levels — a grainy b/w film frame of a dotted US broadcast-station map, with film vignetting
registry/blocks/map-shipping-lanes/assets/topo-plate.png	1440×1080 palette, 96 colours — this one is a synthetic hillshade/DEM render, not a frame. No issue here.

So it is 2 of 3, not 3 of 3 — worth stating precisely.

The per-item sections of the same CREDITS file already describe these accurately and in detail: "Source: Moving Image Archive, Network Broadcasting, 1934 (clip bbdd60a1-a3f6-5d22-8953-3419f5a17afd; IA 0809_Network_Broadcasting_02_01_12_00)" and "public/basemap-t0.png (still plate, not MP4)". The disclosure exists. It is the summary line at the top — and the PR body — that is stale from the pre-revert state, when 346b9d4f1 revert(registry): restore code-drawn animated maps (drop footage wrappers) was still the whole story.

Both frames are 640×480 and 320×240 — original video resolutions, not crops — so "still plate, not MP4" is doing real work and is accurate. I am not raising a licensing objection: registry/animated-maps/MEDIA-URLS.md records that the shot pages report Rights: public_domain. The defect is that the headline statement and the per-item notes disagree with each other, and the headline is the one that ships.

Fix: make CREDITS-animated-maps.md:5 and the PR body say what the per-item sections already say — e.g. "three blocks composite a still plate derived from public-domain Moving Image Archive material; see per-item notes. No footage (MP4) is included."

2. Blocking — #msl-credit can never render

registry/blocks/map-shipping-lanes/map-shipping-lanes.html suppresses its on-screen credit three ways over:

• :243 — inline style="display: none" on the element
• :267 — gsap.set("#msl-credit", { opacity: 0 });
• the timeline never restores it. grep -n 'msl-credit' returns exactly three hits in the file — 49 (CSS), 242 (markup), 267 (the set). Nothing fades it back in.

:272 says why: /* credit hidden for judge - ceiling */. That is a QC-process artifact that shipped.

The attribution for this block now survives only as a data-credit attribute on a display:none div and in the repo-side CREDITS file. Its two sibling archive-derived blocks both render theirs normally, with no suppression — map-border-chase.html:148 and map-radar-rings.html:120-134. So the one block whose on-screen credit is hidden is one of the three that composites archive-derived material, which is the opposite of the intended direction.

Fix: drop the display: none and the opacity: 0 set, or if the credit is deliberately not wanted on screen, delete the element and the CSS rule rather than shipping a hidden one with the text still in the DOM.

---

3. Non-blocking — what the head commit actually did

d4ec111dc is titled fix(registry): align map metadata with supported controls. Alongside that, it removed the attribution surface from precisely the three archive-derived blocks:

• Deleted the credit variable and the credit param from all three. They are now the only three map blocks at variables=0, params=0; the other twenty carry 3-8 each. The credit text is still hardcoded in the HTML, so it is now unoverridable rather than absent.
• Rewrote their description to drop "Moving Image Archive … not affiliated". They are now the only three whose description has no MIA mention; the other twenty retain it.
• Deleted the "Sample verified shots" table from MEDIA-URLS.md — five clips with IDs and archive.org links (NARA / Prelinger / FedFlix).

To be fair to the change: the description rewrite is a real improvement on its own terms. The old strings carried internal process noise into user-facing catalog copy ("QC_PASS 2026-09-21", "Alaska Highway reverse: static basemap plate…"), and the new ones read like product copy. If the runtime genuinely has no credit control for these, removing the param is exactly what the commit title says it is.

It is the combination that is worth a second look: the one commit that strips MIA references from shipped metadata is the head you are asking to approve, and it lands on the three blocks where the attribution matters most. If the credit param was removed because the control is unsupported, the credit should at least be correct where it is hardcoded — and on map-border-chase.html:148 it currently reads MIA - map-border-chase reverse - miguel-heygen, which is internal shorthand naming the engineer instead of the film. The other two name the title and year properly.

4. Nit — stale title

Title says newsreel animated map pack (5 blocks); the body and the diff are 23 (5 original + 18 new). registry.json carries all 23.

---

Verified, no issue found

Publishing these so the scope of what I actually checked is clear:

• Seek-safety across all 23 blocks — clean. No requestAnimationFrame, Date.now, Math.random, performance.now or new Date( in executable code. My first sweep flagged map-hotspot-pulse.html:180; that line is the comment // seek-safe blink via timeline opacity pulses (no Math.random / rAF), and the blink underneath it is a GSAP yoyo/repeat tween on the paused timeline. Stripping comments clears all 23. This covers the whole pack, not the spot-check the test plan proposed.
• lint:registry-items — zero map-* errors. The run prints two failures, both pre-existing and untouched here (blocks/vfx-text-cursor, components/caption-emoji-pop, both font_family_without_font_face). Test-plan item 1 checks out.
• registry.json completeness — 23 map-* entries, names matching the PR table and the CREDITS list exactly, no strays.
• Asset wiring — each PNG-bearing block references its asset at a real relative path and declares it in registry-item.json files[] as type: hyperframes:asset. No orphans, no undeclared files. map-radar-rings also declares assets/measured-data.json.
• registry/catalog-artifact/ — the local-vectors.bin/.json regeneration follows the established tracked-artifact pattern; those paths were last touched the same way by #4056. Not a stray binary.
• .gitignore — one scoped line, .mia/, for the local scrape cache.

Could not run: check:docs-catalog and check:catalog-drift both need node_modules (ERR_MODULE_NOT_FOUND: typescript) and this worktree has none. So docs/catalog payload drift is unverified by me — flagging rather than implying it passed.

Could not verify: the Slack note says "Current-head delta approved; both major findings closed." This PR has zero reviews and zero review comments on GitHub (pulls/4237/reviews → 0, pulls/4237/comments → 0), and the lane thread carries no prior #4237 findings, so I had nothing to check the closure against. If those findings were recorded somewhere else, point me at it and I will confirm them specifically.

Gate mechanic: the PR is a draft (isDraft: true, mergeStateStatus: UNKNOWN). GitHub will not merge a draft, so no approval at this head opens the gate until it is marked ready for review.

One aside, out of diff: scripts/lint-registry-items.mjs exits 0 even when it prints 2 registry item(s) have lint errors (measured directly, not through a pipe). Since the test plan cites that command as its evidence, it is worth knowing it cannot fail a gate as written.

— Rames