How we research filming locations

TMDB — Canonical title metadata: original title, release date, cast, crew, plot summary, streaming availability, official trailer. We do not edit this; we mirror what the cast and crew themselves submit to TMDB.

Perplexity Sonar Deep Research — A search-grounded language model that surfaces filming-location reports from news, fan databases, and behind-the-scenes interviews, with source URLs attached to every claim.

Cross-verification with Claude — Every claim from the research stage is re-checked against its cited source. Claims whose sources don't actually support them are dropped before the page is built.

Google Places API — All practical information — exact address, coordinates, opening hours, photos, and current operational status — comes directly from Google Places. We do not let language models guess this kind of information.

A claim about a filming location only makes it onto a page if at least three independent reputable sources agree on it, or the location is directly matched to a verified Google Place ID. Production trivia (director quotes, behind-the-scenes notes) requires at least one source URL, which we link inline.

We deliberately do not publish:

Every month we re-verify a 5% sample of published pages. If a location's status changes (closure, renovation), if cited sources have gone offline, or if newer reporting contradicts what we wrote, the page is unpublished and re-researched.

We will. Every page has a "Report an inaccuracy" link at the bottom. Three independent reports on the same claim automatically unpublish the page for review. If you've reported something and want to know what happened, email hello@hizine.net.

Most of the writing on this site is drafted by language models from research material we fetch and verify. We use AI because it lets one operator maintain a globally accurate, multi-thousand-page film-tourism reference. We use cited sources, automated cross-checks, and a public correction channel because automation without verification produces unreliable content — and unreliable content is worse than no content for travelers planning real trips.

Posters are sourced via TMDB and re-served from our own CDN at editorial sizes (up to 1000×1500px) for performance and stability. Images are unmodified except for resizing and format conversion to WebP. Each poster is attributed to TMDB and, where known, the producing studio. We do not store or serve original-resolution artwork. Trailers and scene videos are embedded directly from official YouTube channels — we host no video content of our own.

Location photographs come primarily from Google Places, with attribution to the photographer shown beneath each image.

If you are a rights holder and would like content removed, contact hello@hizine.net and we will take it down within 24 hours.

Hizine Editorial is the in-house authorial voice of Hizine. We do not attribute individual writers because most editorial work is produced collaboratively between human curation and AI-assisted research, with all factual claims verified against cited sources before publication. For correspondence with the editorial team, contact hello@hizine.net.

We earn a commission when readers book tours, tickets, or transit through our Klook links. Editorial selection of locations, scenes, and travel guidance is independent of which products earn commission. We do not feature Klook's bestsellers preferentially; we feature Klook products that genuinely match the location covered on the page.