Zarafshan City Map
Zarafshan, house by house — in 3D.
In plain words
- What it is
- A 3D map of the city of Zarafshan: microdistricts, houses with floors and entrances, streets — and a walk through the city “with your own eyes”.
- Who it’s for
- Made to order for a team that works with the city’s housing. Part of the map is open to everyone.
- What it gives
- Find a house by address, see what is nearby and plan a walking route. The client’s team edits the data in a panel, and internal information stays hidden from outsiders.
The Zarafshan City Map is a 3D map of the city with microdistrict borders, houses with floors and entrances, OpenStreetMap objects and your own points. Guests can search by address, find where they are, build a walking route and see what is nearby; a separate page renders the city in full 3D with shadows, facade textures and trees.
Business context
A client needed a working map of Zarafshan’s housing stock — microdistricts, houses, entrances and their own indicators — built on open data, with commercial fields hidden from casual visitors.
The problem
Data about the city — houses, entrances, mahallas, businesses — lives in paper passports, spreadsheets and people’s heads, and generic maps do not show what a local team needs.
The solution
A map built on MapLibre GL and React, a three.js 3D walk and a PHP editing panel over a MySQL database. The team edits houses, buildings, places and indicators themselves; changes appear on the map immediately and every edit is logged.
What I built
Built alone for a client — data processing, map, 3D city, editing panel and API.
Architecture
- A public guest map and a signed-in working mode on one page; private fields never leave the server for guests
- Two renderers on the same data: MapLibre GL for the map, a separate three.js scene for the 3D city
Backend
- PHP read-only API v1: meta, layers as GeoJSON, houses with paging, districts, mahallas, address search, “what’s nearby”
- Editing panel: houses, buildings, photos, places, indicators and a change log with before/after and undo
- Import scripts: OpenStreetMap via Overpass, the client’s houses database, entrances, mahallas, roads
Frontend & apps
- MapLibre GL map: layers, a choropleth by indicator, own points and notes, a first-person street walk
- three.js city: 1,133 buildings, 918 roads at real width, shadows, facade textures, day and evening light
Database
- MySQL; 487 of 562 house records matched to building outlines, 1,370 entrances placed along facades
- 37 houses split into several OSM polygons merged back into one building
Security
- Guests see no indicators, notes or service fields
- API: open endpoints without a key, city data only with a key issued in the panel
- The confidential cadastre extract is never joined into the public API — a test checks it
Delivery & operations
- Changes made in the panel appear on the next page load — no rebuild or redeploy
Technical decisions & trade-offs
- The client’s data wins over OSM — Where the client’s houses database disagrees with OpenStreetMap — floors, house numbers — the client’s data is used: it was collected on site. (Trade-off: Corrections stay in the client database; OSM itself only improves if they are contributed back.)
- A second renderer instead of fighting the first — MapLibre cannot cast shadows or map facade textures per metre, so the realistic city is a separate three.js page built from the same data. (Trade-off: Two renderers to maintain.)
- Privacy by a missing join — The public API has no connection to the cadastre extract with personal data at all, rather than a list of hidden fields; a test checks it. (Trade-off: Public users get only summaries per house.)
- Only legal data sources — Yandex Maps and 2GIS data were not used — their terms forbid extraction; geodata comes from OpenStreetMap under ODbL. (Trade-off: Fewer ready-made places; gaps are filled by the client in the panel.)
Core modules
- 3D map with microdistrict borders
- Houses: floors, entrances, apartments, mahalla
- OpenStreetMap objects
- Address search, “where am I”, walking routes, nearby places
- Choropleth by your own indicators
- Own points and notes
- First-person street walk
- Real 3D page: shadows, facade textures, trees
- Editing panel: houses, buildings, photos, places, indicators
- Change log
- RU · UZ
Results
- 1,133 buildings and 918 roads in the 3D city
- 487 house records matched to outlines, 1,370 entrances placed
- Geodata from the OpenStreetMap export of 26 September 2026
c1.orax1.vip
Boburbek