Files
gods-eye-view/intro-sequence/data/SOURCES.md
T
Claude c9421a3f33 feat(intro): add the First Light intro sequence
A deterministic WebGL film of God's Eye View coming online: a seed on Null
Island grows into a culture that draws the planet, indexes the surface
tile by tile, ingests the live feeds and outgrows its dish, ending on the
logo. Renders 16:9 and 9:16 frame by frame in headless Chromium; the
procedural score is locked to the same clock. Data and imagery are
fetched by the scripts in tools/ and are not committed.

Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01KqP857pippaUs6konSDfe8
2026-10-18 03:21:30 +00:00

11 KiB
Raw Blame History

Intro-sequence data snapshot: sources

All files are real downloads from public keyless endpoints, fetched on 2026-10-18 between 02:24 and 02:42 UTC. Nothing here is synthesized. The scripts only filter, convert units and round. Helper scripts are in ../tools/fetch/ (fetch_basic.sh, process_basic.py, process_tle.py, process_vessels.py, fetch_wind.py, fetch_bonus.py, cctv_fetch.py, cctv_select.py).

JSON files are compact. Tabular layers use {"cols":[...],"rows":[[...],...]}, and each file's cols lists its columns. Coordinates are [lon, lat] in WGS84.

File Rows Source URL(s) Fetched (UTC) Notes / licence
flights.json 5,550 airborne aircraft https://opensky-network.org/api/states/all 02:24:18 (t=1792290258) OpenSky Network state vectors, anonymous tier. Airborne only (on_ground=false), non-null positions. alt_m is geometric altitude, or barometric when geometric is missing. Licence: OpenSky Network, non-commercial/research terms; credit "The OpenSky Network, https://opensky-network.org".
military.json 45 airborne aircraft https://api.adsb.lol/v2/mil, plus https://opendata.adsb.fi/api/v2/mil (merged and de-duplicated by ICAO hex; adsb.lol wins) 02:24 Includes the type column (ICAO type code, e.g. C17, K35R). alt ft→m, gs kt→m/s. country is null because these feeds do not provide it. The count is low because the snapshot was taken at a quiet hour. A re-poll at 02:42 gave fewer (30 and 31 airborne). adsb.lol data is ODbL 1.0; adsb.fi open data. airplanes.live /v2/mil returned 403 and was not used.
tle.txt 15,928 satellites (3-line TLE) https://celestrak.org/NORAD/elements/gp.php?GROUP=active&FORMAT=3le 02:26 CelesTrak "active" group. The FORMAT=tle request kept failing with connection resets after about 11 s through the proxy, while FORMAT=3le answered in 4 s. Element epochs range from day 264 to 293 of 2026. A descriptive User-Agent is required (as in server/providers/space/celestrak.js). Credit: CelesTrak / Dr T.S. Kelso. Data originates from 18th SDS / space-track.org.
tle_groups.json stations 17, starlink 10,651, gps-ops 32, geo 567, weather 72 gp.php?GROUP={stations,starlink,gps-ops,geo,weather}&FORMAT=tle 02:24–02:27 groups_norad: NORAD ids per group, to join against tle.txt (columns 3–7 of line 1). notable: name→group for the first entries of each group (ISS (ZARYA), CSS (TIANHE), GOES, NOAA, GPS, …).
quakes.json 10,120 events https://earthquake.usgs.gov/earthquakes/feed/v1.0/summary/all_month.geojson 02:24 (t=generated ms) USGS ANSS ComCat. Covers 2026-09-18 02:26 to 2026-10-18 02:20 UTC, max M7.7. Sorted by time. Public domain (USGS).
fires.json 246,411 hotspots https://firms.modaps.eosdis.nasa.gov/data/active_fire/suomi-npp-viirs-c2/csv/SUOMI_VIIRS_C2_Global_24h.csv (111,551); https://firms.modaps.eosdis.nasa.gov/data/active_fire/noaa-20-viirs-c2/csv/J1_VIIRS_C2_Global_24h.csv (118,991); https://firms.modaps.eosdis.nasa.gov/data/active_fire/modis-c6.1/csv/MODIS_C6_1_Global_24h.csv (15,869) 02:24 Rows are [lon, lat, frp_MW, brightness_K]. Brightness is VIIRS bright_ti4 or MODIS brightness. Despite the "24h" names, the acquisitions run from 2026-10-16 00:11 to 2026-10-17 23:56 UTC. SNPP and NOAA-20 overlap, so the same fires appear twice. NASA FIRMS (LANCE); free use with credit "NASA FIRMS".
vessels.json 4,385 vessels https://kystdatahuset.kystverket.no/ws/api/ais/realtime/geojson (Norway, 3,272 with a position); https://meri.digitraffic.fi/api/ais/v1/locations?from= + https://meri.digitraffic.fi/api/ais/v1/vessels for names (Finland/Baltic, 1,113 after de-duplication against Norway) 02:28–02:29 Not global. Coverage is the Norwegian coast and EEZ including Svalbard (Kystverket Kystdatahuset open API; NLOD licence) plus Finnish waters and the northern Baltic (Fintraffic Digitraffic; CC BY 4.0). Kystverket returns short trails, and the script uses the last vertex as the current position, checked against COG. sog_kn/cog_deg are null when AIS sends "not available". No keyless global AIS source was found. AISStream (the repo's source) needs a key, BarentsWatch needs OAuth (401), the Danish DMA host has an expired TLS certificate, and Kystverket's raw TCP NMEA feed cannot pass the HTTPS-only proxy.
wind.json 360×181 grid (65,160 cells) https://noaa-gfs-bdp-pds.s3.amazonaws.com/gfs.20261017/18/atmos/gfs.t18z.pgrb2.1p00.f006 (byte ranges from its .idx, UGRD/VGRD at "10 m above ground") 02:34 NOAA GFS 1.0° run 2026-10-17 18Z, +6 h forecast, valid 2026-10-18 00Z. This is the same AWS Open Data bucket and .idx range approach as server/providers/wind/gfs.js, decoded with ecCodes (pip eccodes). The grid is re-indexed so lo1=-180, la1=90, dx=dy=1, row-major from north to south, m/s. The 00Z run of 10-18 was not yet published. Public domain (NOAA).
cables.json 1,948 polylines (732 cables) https://www.submarinecablemap.com/api/v3/cable/cable-geo.json 02:24 Each LineString part is a separate polyline, rounded to 2 decimals. © TeleGeography, Submarine Cable Map (CC BY-SA 4.0); attribute TeleGeography.
coastline.json 1,429 lines, 60,392 vertices https://raw.githubusercontent.com/nvkelso/natural-earth-vector/master/geojson/ne_50m_coastline.geojson 02:24 Natural Earth 1:50m coastline, rounded to 2 decimals with consecutive duplicate points removed. Public domain.
borders.json 288 rings https://raw.githubusercontent.com/nvkelso/natural-earth-vector/master/geojson/ne_110m_admin_0_countries.geojson 02:24 Natural Earth 1:110m countries, each polygon ring as a line. Public domain.
cities.json 3,000 https://raw.githubusercontent.com/nvkelso/natural-earth-vector/master/geojson/ne_10m_populated_places_simple.geojson 02:24 Top 3,000 by pop_max. Natural Earth, public domain.
airports.json (bonus) 5,280 (1,176 large, 4,104 medium) https://davidmegginson.github.io/ourairports-data/airports.csv 02:24 OurAirports large and medium airports. iata falls back to the ICAO ident when there is no IATA code. size is L or M. Public domain.
radio.json (bonus) 12,983 stations https://de1.api.radio-browser.info/json/stations/search?has_geo_info=true&hidebroken=true&limit=100000 02:40 Radio Browser community directory (the repo's radio provider). Stations with geo coordinates only. Public domain / open.
bikes.json (bonus) 31,176 docks 18 GBFS station_information.json feeds (Citi Bike, Divvy, Bay Wheels, Capital Bikeshare via gbfs.lyft.com; Bluebikes; Toronto, Barcelona, Madrid, São Paulo, Rio, Santiago, Bogotá via publicbikesystem.net; BIXI; Ecobici; Vélib'; Indego and LA Metro via bcycle; Oslo via urbansharing) + https://api.nextbike.net/maps/nextbike-live.json (all nextbike cities, 16,926 stations) 02:41 Per-system counts are in systems. Operator open-data terms apply (mostly GBFS open licences).
launch_sites.json (bonus) 249 pads https://ll.thespacedevs.com/2.3.0/pads/?limit=100&mode=list (paged) 02:42 Launch Library 2 (The Space Devs), sorted by total launches. Free tier.
power_plants.json (bonus) 34,936 plants https://raw.githubusercontent.com/wri/global-power-plant-database/master/output_database/global_power_plant_database.csv 02:42 WRI Global Power Plant Database v1.3 (static, 2021). CC BY 4.0.

Not produced: military installations. The repo's provider queries OSM Overpass by viewport only and rejects global requests. A global Overpass pull would be slow and against that design. Datacenters: the repo has no keyless source.

CCTV (../cctv/)

cctv/cam_001.jpg … cam_117.jpg: 117 real frames, each fetched 2026-10-18 02:38:46–02:39:30 UTC and cover-cropped and resized to 640×360 (JPEG q88). cctv/index.json gives file, name, city, country, lat, lon, source, plus fetched_utc, orig_size and the frame url.

About 235 frames were downloaded from 20 sources, and every one was reviewed on contact sheets. Placeholders were rejected: Caltrans "Temporarily Unavailable", Vegvesen "Kamera er midlertidig ute av drift", Tallinn "NB! ÜHENDUS KAAMERAGA PUUDUB", Barcelona "Interrupció temporal de la imatge", and one all-black King County frame. Frames whose average hash repeats within the set were also dropped as placeholders, as were frames under 320 px wide or with luminance mean ≤25 or std ≤15.

Local time at capture: daylight in Sydney, NZ, Hong Kong and Singapore; evening or night in the Americas; pre-dawn night in Europe. The Hong Kong TD frames carry the department's own privacy-mask polygons.

Source Frames Country Catalog endpoint (frames from)
Live Traffic NSW 12 Australia https://data.livetraffic.com/cameras/traffic-cam.json (webcams.transport.nsw.gov.au)
NZTA Waka Kotahi 12 New Zealand https://trafficnz.info/service/traffic/rest/4/cameras/all (trafficnz.info/camera/*.jpg)
HK Transport Department 10 Hong Kong https://static.data.gov.hk/td/traffic-snapshot-images/code/Traffic_Camera_Locations_En.xml (tdcctv.data.one.gov.hk)
LTA Singapore 6 Singapore https://api.data.gov.sg/v1/transport/traffic-images (the feed listed only 8 cameras)
NYC DOT 9 USA https://webcams.nyctmc.org/api/cameras (…/{id}/image)
Caltrans CCTV 6 USA https://cwwp2.dot.ca.gov/data/d{4,7,11,3}/cctv/cctvStatusD0X.json
WSDOT 6 USA https://data.wsdot.wa.gov/travelcenter/Cameras.json (images.wsdot.wa.gov only)
King County DOT 3 USA ArcGIS TRAFFICCAMERA_POINT_2029 FeatureServer (info.kingcounty.gov)
TxDOT ITS 4 USA https://its.txdot.gov/its/DistrictIts/GetCctvStatusListByDistrict?districtCode=… (base64 JSON snapshot; Dallas and San Antonio worked, Houston and El Paso returned null)
City of Austin 4 USA https://data.austintexas.gov/api/views/b4k4-adkb/rows.json (cctv.austinmobility.io)
DriveBC 6 Canada https://www.drivebc.ca/api/webcams/
City of Toronto RESCU 6 Canada https://opendata.toronto.ca/transportation/tmc/rescucameraimages/Data/tmcearthcameras.json
City of Calgary 4 Canada https://data.calgary.ca/resource/k7p9-kppz.json
TfL JamCams 8 UK https://api.tfl.gov.uk/Place/Type/JamCam (jamcams.tfl.gov.uk S3)
Fintraffic Digitraffic 4 Finland https://tie.digitraffic.fi/api/weathercam/v1/stations (weathercam.digitraffic.fi)
Statens vegvesen 4 Norway ogckart-sn1.atlas.vegvesen.no datex_3_1:CctvSimple (kamera.atlas.vegvesen.no)
City of Tallinn 3 Estonia repo config/cctv_sources.tallinn.json (ristmikud.tallinn.ee)
Transpordiamet Tarktee 3 Estonia tarktee.transpordiamet.ee DATEX2 roadCameraLocations + roadCameraImages
Via Lietuva eismoinfo 4 Lithuania https://eismoinfo.lt/eismoinfo-backend/camera-info-table + layer-static-features/VKR
Servei Català de Trànsit catalog 3 Spain https://www.gencat.cat/transit/opendata/cameres.xml (the 3 kept frames are Barcelona IMI cameras on www.bcn.cat; SCT's own mct.gencat.cat host would not answer over HTTPS)

Skipped: Ontario 511 (now answers "Invalid Key"), DelDOT (video only), Warendorf (http-only host, and the proxy allows HTTPS only).

The frames are public agency imagery, used here under each agency's open-data or terms-of-use conditions. Credit the agency named in source. TfL: "Powered by TfL Open Data". NSW: © Transport for NSW. HK: © HKSAR Transport Department (data.gov.hk). Singapore: LTA via data.gov.sg (Singapore Open Data Licence). NZ: © NZ Transport Agency Waka Kotahi (CC BY 4.0).