The HUD builds place context for its AI summary whenever the camera
settles, and each build paid for up to four reverse geocodes plus a Nearby
Places search, even when no OpenAI key was configured to use them.
- Ask GET /api/openai/hud-summary/status once; with no summary configured,
or after an unconfigured reply, make no place lookups for the page.
- Reverse-geocode at most two points: the view center and one viewport
point.
- Reuse a lookup until the view center moves a quarter of the visible
width, the camera's distance to it doubles or halves, or the camera
crosses a lookup height. The reused points are asked again, so only a
timeout, failure or empty answer spends.
- Repaint the last AI reply when a small pan leaves the context unchanged.
Cached-only hosts and voice context lookups are unchanged.
Reuse retained tile triangles for CCTV's upcoming activation probe, with finite range, exclusions and honest misses. Correct nearest-hit ordering when the ray starts inside a mesh bound.
Fifteen focused surface tests and all full keyless unit/allocation, boundary, format and build gates pass. CCTV renderer integration and existing acceptance gaps remain documented.
Keep shared CCTV logic independent of Cesium, release the original renderer's resources, and guard retired initialization/media work. Preserve native marker, plane, volume, calibration and input behavior.
Record keyless dev/preview validation and retain unresolved shell idle/terrain and photoreal acceptance gaps. Routine QA builds use empty Google and Cesium credentials and block paid maps before navigation.
Apply the accepted findings from the fourth three-model French review:
« Fermez d'abord cet outil » in the pointer-busy notices, the colon form
« Outil actif : {name} » for the directions notice, « FOV CONSTRUCTEUR »,
« RAPPORT » for the USGS trigger chip, « CONTEXTE GLOBAL DÉSACTIVÉ », and
wording fixes for the radio, recent imagery, bundle preview and draw label
strings. The guide gains glossary rows for the datasheet FOV and for
rapport versus relevé.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Record keyless launch instructions without changing app defaults.
Block Google tile and token fallback requests in routine CCTV checks;
add a guarded startup check for both engines.
Catalog: refresh/updates wording, received SDR aircraft, playing states,
move-start button, impersonal layer note, live-data descriptions,
arrival steps, retry timing, tracking spelling, military chip, dollar
amounts, label: value statuses, upstream source, globe map, stage
recovery, comparison signs as words, ellipses and earlier round-2 fixes.
Code: FIRMS cell cards use the plural entry for their count (FIRES
pruned), translated locator pin titles are isolated in RTL, legend
stops keep their sign, voice timeout minutes agree with the number,
CCTV badges are translated, first-run and referenced-item alignment.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Apply the accepted findings from the third three-model French review:
tool-name agreement (« Outil actif : {name} »), elision before ADS-B,
« GORGE AVAL », « CLASSIQUES », « EN SURVOL », one name for the VIIRS
daily overview, U+202F thousands separator, and wording fixes for the
scene editor, directions, radio tuner and status strings.
The CCTV HUD summary now shows its CAL badge, source kind and view band
through existing catalog entries, so French reads « CAL ESTIM. INITIALE ·
SRC AMONT · CONTEXTE RUE »; English output and unknown codes are
unchanged. presentation.js matches the pt branch's round-3 version byte
for byte, so the two branches merge cleanly.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Catalog fixes from the third review round: booster flight count
({count}回目の飛行), 晴れ for PARTLY CLOUDY, 配信元 for upstream,
部分的 for the PARTIAL feed state, または between provider names,
POS kept as a HUD code, product names in their own case (YouTube,
Facebook, Radio Browser, ChatGPT OAuth, Googleストリートビュー), Esc,
bracketed button names, en-dash ranges, 土砂ダム湖 for both lake
labels and smaller wording fixes. Guide: range rule and new glossary
rows.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Move native gizmo entities and input into an owned renderer. Restore the
captured camera-input flag, handle cancellation and duplicate presses, and
reject foreign handle identities. Preserve native geometry and picking.
Avoid no-op sync-chip mutations and let unchanged CCTV polls complete
without publishing scene changes. Keep health/UI polling and real geometry,
auto-hop and media updates. Extend browser idle samples across two polls.
Validated full Node24 unit/allocation, boundaries, scoped format, production,
panel and QA builds; native legacy parity, real media/GPU picks on dev and
preview, engine contracts and production shell. Retain failed evidence.
Three-model review (astra, sol, fable), round 3: 17 of 18 findings
applied (some in own wording, one in part); "Missões espac." rejected.
- pt: layer cards «{label}: camada ativada/desativada», pointer and
Directions toasts name «A ferramenta "{owner}"», counters after a
colon where the count can be 1, traffic «sem dados de fluxo» and
«Fluxo: {percent}% cob», «limite de requisições atingido», FECHAR,
ANT., PREDEF. VISUAIS, quoted "Nível da rua", infinitive tooltips,
and more (see docs/translations/pt.md).
- CCTV summary line localizes its CAL, SRC and CONTEXT values through
existing catalog keys (all locales); codes stay underneath.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Catalog fixes from the second review round (squawk 7600/7700 names,
transit occupancy, radio category, cyclone advisory wording, Orbital Watch,
validation and hint texts), cockpit distances keep "km" in Japanese as in
Chinese, the CCTV context line uses the translated view band, and the
Japanese layer status line no longer splits "1分以内".
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Prove native geometry, real image and video pixels on dev and preview before renderer extraction. Scope fixture interception to API routes so geometry workers remain live.
Separate native scene ownership from shared ALPR source and selection state,
and add bounded Navara sprites with the existing direction/card overlay.
Preserve marker artwork, picking, floor retries and source attribution.
Fix lost post-render wakeups and redundant mobile rail mutations, and keep
HUD typewriter dimensions stable so static views settle without extra frames.
Add parser-fed dev/preview and legacy evidence, including retained failures.
Common W2 terrain and combined performance gates remain open.
Sameh Khamis