Quick movie facts from the command line. Point it at a movie or a person and get the one answer you wanted, without opening IMDB or asking a voice assistant.
It reads as "cinema atlas": a reference and a literal map. Filming locations resolve to coordinates and link straight to Google Maps and Google Earth.
- Where a movie was filmed, on a map.
- Who is in a movie, and who directed it.
- What else a person was in or directed.
- Who someone is and their notable roles.
One click stands the site up on Render's free tier from the blueprint in this
repo. Set CINATLAS_TMDB_KEY when prompted; add a custom domain in the service
settings afterward.
go install github.com/dcadolph/cinatlas@latest
Or build from a clone:
go build -o cinatlas .
Data comes from TMDB, which needs a free key. Create one at themoviedb.org under Settings, then API. A v3 key or a v4 read access token both work.
export CINATLAS_TMDB_KEY=your_key_here
cinatlas where "No Country for Old Men"
cinatlas at "Monument Valley"
cinatlas cast Heat
cinatlas films "Denis Villeneuve"
cinatlas who "Stephen Tobolowsky"
cinatlas serve
at searches the other direction: name a place, get the films with recorded
locations there. Coverage mirrors Wikidata, so film hubs are rich and small
towns can be empty.
serve runs the cinatlas website locally on 127.0.0.1:8878 and opens it in your
browser: one search box, poster and cast cards, filming locations on an embedded
map. Same data, same cache, zero setup beyond the key.
Output is human-readable text by default. Add --json for machine output, and
--pretty to indent that JSON. Logs go to stderr, so data and diagnostics never mix.
Successful responses are cached on disk for 24 hours, so repeat questions answer
instantly and stay clear of API rate limits. The cache lives in the OS user cache
directory under cinatlas. Use --refresh to bypass it, CINATLAS_CACHE_DIR to
move it, and CINATLAS_CACHE_TTL (a Go duration like 1h or 72h) to change
freshness.
Locations merge from tiers, each labeled by source in the output: Wikidata filming locations (structured), Wikipedia mining (street-level places linked from the article's filming section), and production countries as the coarse fallback. Films with zero pins link the IMDB locations page instead. Where the story is set shows separately from where it filmed.
On the website every movie gets an interactive globe of its pins: the
"Open the globe" button, or /globe?id=<tmdb id>.
TMDB supplies cast, crew, filmography, images, and IMDB ids. Wikidata supplies filming locations, settings, and coordinates. Wikipedia supplies street-level filming places. IMDB is a link target only.
Early. The command surface and data path are in place. Response caching, richer location coverage, and the web interface are next. See the design plan for the roadmap.