Mark Burton wrote:
Firstly, I would like to thank everyone who responded to my "Understanding the sea" post. From those posts and my own study, I finally worked out what it does. And yes, it works pretty well inasmuch that it generates some useful polygons. The "sea sector" heuristic doesn't work for me so I have been disabling it.
The problem is, the end result is "butt ugly". Our MP implementation really can't do a good job on zillions of little islands.
After frigging around for a couple of days trying to improve the visual quality, I tried the obvious alternative technique of not using a MP but, instead, just having a sea background and then put the land on top using plain polygons. Guess what? It works very nicely as the attached pics show. I now have a splendid map of the Baltic area (OK it's weird at the edges but that's not a problem for now).
Here's the (future) commit comment for this patch:
Provide an alternative to using a multipolygon when generating the sea.
The --generate-sea option can now take a comma separated list of values:
no-mp - don't generate the sea using a multipolygon - instead, generate a background sea polygon and then individual land polygons with tag natural=land. This requires a suitable land polygon type to be defined in the style file (suggested type is 0x010100) and the polygon must be defined in the TYP file as having a higher drawing level than the sea polygon type.
no-sea-sectors - disable the generation of "sea sectors" when the coastline fails to reach the tile's boundary.
With no values specified, the --generate-sea option should work as before.
Oh yes, this patch doesn't include it, but I am also disabling the DP code for polygons as it makes them weird.
Cheers and thanks again for the helpful postings.
Mark
Hurrah! Does this also mean that, with suitable definition of the polygon used for land, we can finally banish the "Garmin yellow" background to history? -- Charlie