playbooks / image-geolocation

Photo → where (and when) it was taken

The classic geolocation workflow. Read the metadata, reverse search, extract clues, narrow the search area, confirm with imagery, then estimate the time from shadows.

start: ▣ imageyields: location, file / doc, urllevel: intermediate

1. 01

   Check metadata first

   Original files can carry GPS coordinates, a timestamp and the camera model. Most social platforms strip EXIF data, but files from email, cloud shares or forums often keep it. Error-level analysis can also flag edited regions.

   • ExifToolfreeossclidesktopdetails
   • FotoForensicsfreedetails
   • Forensicallyfreedetails

   → can yield: ⊕Location ▧File / Doc

2. 02

   Reverse search, then crop and repeat

   Search the whole image first, then crop to distinctive parts (a building, a sign, a mountain profile) and search each crop. Each engine has a different index. Yandex excels at landscapes and buildings in Eastern Europe, Lens at landmarks and products.

   • Google Lensfreemobiledetails
   • Yandex Imagesfreedetails
   • Bing Visual Searchfreeapidetails
   • TinEyefreemiumacct optionalapiextensiondetails

   → can yield: ⇗URL ⊕Location

3. 03

   Extract clues

   List everything visible: script and language on signs, what side of the road traffic drives on, road markings, bollards, plate formats, power-pole designs, vegetation, architecture, terrain. Each clue cuts down the candidate area.

   AI geolocation estimators can suggest a starting region, but treat their output as a hypothesis, never as an answer.

   • GeoSpypaidacct reqapidetails
   • PlatesManiafreeacct optionaldetails

   → can yield: ⊕Location

4. 04

   Narrow with map queries

   Turn clues into a spatial query: "a pharmacy within 50 m of a roundabout and a church, next to a river". Overpass Turbo and Bellingcat's OSM Search find every match in OpenStreetMap data.

   • Bellingcat OpenStreetMap Searchfreeossdetails
   • Overpass Turbofreeossapidetails
   • Wikimapiafreeapidetails
   • PeakVisorfreemiumacct optionalmobiledetails

   → can yield: ⊕Location

5. 05

   Confirm with imagery

   Match the candidate site from multiple angles using street-level and satellite imagery. Historical imagery helps when the photo is old or the site has changed.

   • Google Earthfreedesktopmobiledetails
   • Google Maps & Street Viewresource
   • Mapillaryfreeapimobiledetails
   • KartaViewfreeossapimobiledetails
   • Esri World Imagery Waybackfreedetails

   → can yield: ⊕Location

6. 06

   Estimate the time of day

   With the location confirmed, shadow direction and length give the sun's azimuth and elevation, which narrows the time and sometimes the date.

   • SunCalcfreedetails
   • ShadowFinderfreeossclidetails