ZeroTrace OSINT
Visual & Geolocation
Eight tools for image metadata, reverse search, geolocation clue extraction, GPS trails, sun-shadow analysis, aerial comparison, and photo clustering.
The Visual & Geolocation discipline combines image-level investigation (EXIF, hashes, reverse search) with place-level investigation (sun position, aerial imagery, geo clues). It is the area of OSINT most commonly asked about and the area where investigators most often hit a "what tool do I even use for this" wall.
The toolkit's design principle here: prefer local computation, prefer free public sources, never use face recognition. Every tool runs on your machine; the few that hit external services name those services on the result.
What's in this section
| Tool | What it does | Best when |
|---|---|---|
| Image metadata | Full EXIF / IPTC / XMP parsing, GPS extraction, perceptual hashes, thumbnail extraction | You have a photo. Start here. |
| Reverse image | Composes deep-link reverse-image-search queries across Google Lens, Yandex, Bing, TinEye, SauceNAO, Karma Decay | "Where else has this image appeared?" |
| Geolocation clues | OCR, language and script detection, license-plate format matching, signage colour analysis | "Where was this photo taken?" image-geolocation challenges |
| EXIF GPS trail | Bulk EXIF extraction with map polyline, distance, speed, anomaly detection | A folder of photos with GPS data |
| Sun & shadow solver | Solar-position math for latitude estimation from photo timestamp | "What latitude was this taken at?" |
| Aerial comparator | Side-by-side OSM, Esri, Bing, Sentinel-2 imagery for a coordinate | Verifying a candidate location against satellite imagery |
| Photo clustering | Perceptual-hash clustering across many photos for cross-source matching | "Are these the same photo?" — without face recognition |
What this section deliberately does not include. No face recognition, no biometric matching, no search-by-face. The legal and ethical risks of those capabilities outweigh the OSINT value, and the available implementations are paid services we do not subsidise. The closest thing the toolkit offers is photo-content hashing — matching identical or near-identical images, not identical people.
Common starting points
| You have... | Best first tool |
|---|---|
| A single photo | Image metadata (then reverse image, then geo clues if no GPS) |
| A photo and a candidate location | Aerial comparator |
| A photo with no obvious clues | Geo clues + reverse image |
| A folder of photos | EXIF GPS trail |
| A set of profile photos to cross-reference | Photo clustering |
| Just a date / time and a sun-position guess | Sun & shadow solver |
A typical image-geolocation workflow
For an unknown photo where you need to figure out where it was taken:
- Image metadata — start with the cheap wins. EXIF GPS, if present, is the answer. EXIF camera details narrow the device family.
- Reverse image — does the photo appear elsewhere on the internet, with location context?
- Geo clues — let the OCR find readable text (street names, business names, license plates). Let the language detector narrow the country.
- Sun & shadow solver — if the timestamp is reliable and you can guess sun bearing, narrow the latitude band.
- Aerial comparator — once you have a candidate area, verify against satellite imagery.
This is the "image geolocation challenge" workflow that journalists, investigators, and the OSINT community use against unknown photos. The toolkit packages each step as a dedicated tool.
Privacy and consent
Image investigation can rapidly cross into territory where the subject of a photo has not consented to being investigated. The toolkit does not collect or transmit any uploaded image — every photo you load stays on your machine. But the outputs of image investigation can be sensitive. Apply judgment about what to publish, share, or store, particularly where private individuals are concerned.
What the toolkit does not store
Photos you load into the application are processed in memory and on local disk. They are not transmitted to ZeroTrace. They are not logged centrally. The only network calls the visual tools make are explicit reverse-image-search deep links you click and the optional aerial-imagery tile fetches in the comparator.