Describe the bug
Location error between layers
To Reproduce
Steps to reproduce the behavior:
The jar used to create these tiles was obtained from:
on 2nd July 2026
https://github.com/onthegomap/planetiler/releases/latest/download/planetiler.jar
The data from which the tiles were cut was sourced from:
https://download.geofabrik.de/australia-oceania/australia-latest.osm.pbf
And clipped by:
osmium extract --polygon lga.geojson --strategy smart --overwrite --output lga.osm.pbf australia.osm.pbf
java -Xmx4g -jar planetiler.jar --osm-path=lga.osm.pbf --output=lga.mbtiles --minzoom=5 --maxzoom=16 --download --download-dir=planetiler_sources --force
Expected behavior
Tiles produced at zoom level 15 should have similar location errors as zoom level 16 for same location.
Adjacent tiles should have the same location errors for on shared edges.
Tile 15/30307/19001.pbf
Centre slightly above middle has two circles with significant location discrepancy.
see 19001.png
Tile 16/60615/38002
Shows same two circles with significantly less location discrepancy.
see 38002.png
Tile 16/60614/38002
Shows nearly zero location discrepancy.
The discrepancy between 60614 and 60615 is also apparent as the common geographic edge between the two tiles have different displacement errors for the same features.
These 3 tiles should show the same level of location discrepancy if the discrepancy came from the source data.
Screenshots
If applicable, add screenshots to help explain your problem (include tile IDs or latitude/longitude for visual issues
with generated maps)
Environment:
- Hardware: Oracle VM
- OS: Lime
- Java version and distribution: 21
- Maven version: 3.16.0
Additional context
Looks as if map projections are using inconstant math between features
tiles.zip
lga.zip
.
Describe the bug
Location error between layers
To Reproduce
Steps to reproduce the behavior:
The jar used to create these tiles was obtained from:
on 2nd July 2026
https://github.com/onthegomap/planetiler/releases/latest/download/planetiler.jar
The data from which the tiles were cut was sourced from:
https://download.geofabrik.de/australia-oceania/australia-latest.osm.pbf
And clipped by:
osmium extract --polygon lga.geojson --strategy smart --overwrite --output lga.osm.pbf australia.osm.pbf
java -Xmx4g -jar planetiler.jar --osm-path=lga.osm.pbf --output=lga.mbtiles --minzoom=5 --maxzoom=16 --download --download-dir=planetiler_sources --force
Expected behavior
Tiles produced at zoom level 15 should have similar location errors as zoom level 16 for same location.
Adjacent tiles should have the same location errors for on shared edges.
Tile 15/30307/19001.pbf
Centre slightly above middle has two circles with significant location discrepancy.
see 19001.png
Tile 16/60615/38002
Shows same two circles with significantly less location discrepancy.
see 38002.png
Tile 16/60614/38002
Shows nearly zero location discrepancy.
The discrepancy between 60614 and 60615 is also apparent as the common geographic edge between the two tiles have different displacement errors for the same features.
These 3 tiles should show the same level of location discrepancy if the discrepancy came from the source data.
Screenshots
If applicable, add screenshots to help explain your problem (include tile IDs or latitude/longitude for visual issues
with generated maps)
Environment:
Additional context
Looks as if map projections are using inconstant math between features
tiles.zip
lga.zip
.