water matcher #416
water matcher #416
Conversation
|
Current planet tiles up to z15 for the water layer: |
|
Since the water layer is closely related to the earth layer, I also touched the Earth.java file. Islands in the Mediterran for example appear now at the same zoom level as the corresponding hole appears in the water layer. At least mostly. I found that if we use FeatureMerge.mergeNearbyPolygons without any buffer, i.e., buffer=0, then there are sometimes artifacts like spikes in earth and water polygons around zoom levels 6, 7, 8. I chose therefore to buffer but if the buffer is too large, e.g., buffer=0.5, then islands don't get corresponding holes in the water layer. The smallest buffer I saw had an effect on removing the spikes artifacts was the tile resolution, i.e., 0.0625. Since I was touching the Earth.java file, I saw that the Antarctica fix is there but it should really be in the Landcover.java file. So I moved it over... |
|
There are also some minor tweaks to the Landcover.java polygon merging. I can also revert those. I think if we want to go deeper into landcover I would do it in a separate pull request. Google maps buffers landcover by the way, I guess to avoid the gaps that you always get due to feature dropping and simplification:
|
|
Uploaded a new demo file to protomaps.dev with the layers water, earth, and landcover: I just noticed that the Atlantic Ocean label and similar ones are missing below zoom 6. Need to fix this... |
|
The last commit introduces minzooms for some place=sea labels.
|
|
Closes #229 |
|
Some place=sea come at z3, all others at z6. Looks like this for the less important sea labels at z6 in the Mediterranean:
|
|
Updated pmtiles file can be found at: |
|
MinZoom for points should be 15, and not 12. Otherwise we get all these fountains etc on the map at z12 |
|
Also these bays from nodes in Croatia are a bit too much at z12. Better to show them only from z15 on... |
|
I found that it is important to use |
|
Copied filter rules from tilezen's water.yaml file. Each copied filter rule was marked with "done". Each skipped item was marked with "skip" in wipfli/vector-datasource#1
Skipped are the following
Rules in tilezen were evaluated top to bottom, we do pick the last. So I reversed the order of the filter rules, e.g., covered=yes appears last here, in tilezen it was first...