Giters

answerquest / maptiles_recipe_buildings

recipes to serve vector and raster map tiles of datasets by https://github.com/microsoft/GlobalMLBuildingFootprints

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

maptiles_recipe_buildings

recipes to serve vector and raster map tiles of datasets by https://github.com/microsoft/GlobalMLBuildingFootprints

Latest status:

  • as of: 2022-05-30
  • Vector AND Raster map tiles layers working, see: https://server.nikhilvj.co.in/tileserver/ (Tileserver GL deployment)
  • Vector tile url: https://server.nikhilvj.co.in/buildings1/data/india_buildings_z14/{z}/{x}/{y}.pbf
  • Raster webp url: https://server.nikhilvj.co.in/buildings1/styles/basic/{z}/{x}/{y}.webp (reccommended, is faster than png)
  • Raster png url: https://server.nikhilvj.co.in/buildings1/styles/basic/{z}/{x}/{y}.png

Converting to mbtiles using Tippecanoe

nohup tippecanoe -o india_buildings_z14.mbtiles \
-n india_buildings_z14 \
-N "from https://github.com/microsoft/GlobalMLBuildingFootprints India download, processed by Nikhil VJ, https://nikhilvj.co.in" \
-z14 --extend-zooms-if-still-dropping \
--drop-densest-as-needed --drop-smallest-as-needed \
--coalesce-smallest-as-needed \
--coalesce-densest-as-needed \
--read-parallel --exclude-all \
India.geojsonl &
  • Created a 2.5GB .mbtiles file in about 3 hrs. Did not consume too much RAM as it didn't have to load the entire data into memory.

Serving vector tiles

docker run --rm -it -v $(pwd):/data -p 7100:8080 -p 7101:80 maptiler/tileserver-gl --verbose -b 0.0.0.0 -u "https://server.nikhilvj.co.in/tileserver/"
  • output:
Starting tileserver-gl v3.1.1
No MBTiles specified, using india_buildings_z14.mbtiles
[INFO] Automatically creating config file for india_buildings_z14.mbtiles
[INFO] Only a basic preview style will be used.
[INFO] See documentation to learn how to create config.json file.
WARN: MBTiles not in "openmaptiles" format. Serving raw data only...
{
  "options": {
    "paths": {
      "root": "/app/node_modules/tileserver-gl-styles",
      "fonts": "fonts",
      "styles": "styles",
      "mbtiles": "/data"
    }
  },
  "styles": {},
  "data": {
    "india_buildings_z14": {
      "mbtiles": "india_buildings_z14.mbtiles"
    }
  }
}
Starting server
Listening at http://0.0.0.0:8080/
Startup complete
- live on: https://server.nikhilvj.co.in/tileserver/

Serving raster tiles

  • As per https://tileserver.readthedocs.io/en/latest/config.html ,
  • serve_rendered: true needs to be there in the style configuration; and it's not there when doing default
  • First deployed tileserver-gl without any config specified, it gave vector tiles and a tilejson
  • Used a locally run docker instance of https://maputnik.github.io/editor , loaded empty style, added data source as tilejson above
  • Added a layer, the source/layer names were auto-added by maputnik.
  • Was able to get the buildings appearing in preview in maputnik. Made a style and saved the style json.
  • Edited style json to change data source url to "mbtiles://india_buildings_z14.mbtiles" instead of the pbf url that maputnik had saved.
  • made config.json point to this style, and kept "serve_rendered": true
  • placed both config.json and the style json in the same folder as the .mbtiles file which is mounted as /data in docker command.
  • Re-deployed with config:
docker run -d -v$(pwd):/data -p 7100:8080 -p 7101:80 maptiler/tileserver-gl -c config.json --verbose -b 0.0.0.0 -u "https://server.nikhilvj.co.in/buildings1/"
  • and now in addition to vector tiles, TileServer GL deployment is giving rasters also: https://server.nikhilvj.co.in/buildings1/
  • the config.json and india_buildings_z14_style.json files are uploaded in this repo.

Publishing on a URL with SSL

About

recipes to serve vector and raster map tiles of datasets by https://github.com/microsoft/GlobalMLBuildingFootprints

License:MIT License