Skip to content
Mavrick Blackburn
All projects

Case study

NatureEcho

Find where your nature photos live, for real.

Role
Design and build
Status
Unreleased
Visit natureecho.pages.dev (opens in a new tab)

01/06

  • The landing page — every wild place has an echo.
    The landing page — every wild place has an echo.
  • Three taps from a photo to a place.
    Three taps from a photo to a place.
  • The argument the whole project rests on: an AI asked to name real places will invent plausible ones with plausible coordinates, so every place is checked against iNaturalist's gazetteer. Confirmed pins are green and carry attributed photos; unconfirmed ones stay amber with no photo at all.
    The argument the whole project rests on: an AI asked to name real places will invent plausible ones with plausible coordinates, so every place is checked against iNaturalist's gazetteer. Confirmed pins are green and carry attributed photos; unconfirmed ones stay amber with no photo at all.
  • Matches are spread across countries rather than clustered in one — a mossy gully in one hemisphere rhyming with a cedar slope in another.
    Matches are spread across countries rather than clustered in one — a mossy gully in one hemisphere rhyming with a cedar slope in another.
  • What to expect back: a fern at eye level, not a drone shot.
    What to expect back: a fern at eye level, not a drone shot.
  • Questions, answered.
    Questions, answered.

Overview

Upload a photo of somewhere wild and a vision model proposes real places on Earth with the same biome. Every proposal is then checked against iNaturalist's gazetteer before it is shown as fact: confirmed places get a green pin, the gazetteer's own coordinate, and real attributed observation photos. Unconfirmed ones are still shown, but badged amber with no photo — a real photo beside an unverified name would be its own small lie. No account, and the photo stays on your device until you tap Analyze.

Technology

  • JavaScript
  • Cloudflare Pages
  • Cloudflare Functions
  • Wrangler
  • Vision models
  • Third-party APIs
  • Git & GitHub