[mkgmap-dev] Sea generation

Fri Nov 5 17:14:19 GMT 2010

Thanks for all feedback! That's great!

Looking at the discussion I think the idea of an extra coastline file 
has the highest priority.
I think the implementation will mostly follow the suggestions of 
Apollinaris Schoell:
- coastline input file(s) will have OSM format
- natural=coastline tag will be removed from all other tile data
- the coastline data is processed by the current sea generation 
algorithms using the multipolygon algorithm
- for each tile the resulting big multipolygon is clipped to the tile 
borders and added to the tile data

Using this algorithm it should be possible to have multiple coastline 
input files (e.g. one for the Mediterranean Sea, one for Atlantic Ocean 
etc.) The only condition is that the coastline files define complete 
polygons. So the Mediterranean Sea file would have to add a coastline 
tag crossing the Strait of Gibraltar. The polygons itself might consist 
of multiple ways as it is used in the original OSM data. The sea 
generation algorithm connects them to polygons.

Some add ons maybe implemented later:
- shp support for input (is that needed? better preprocessing?)
- mp algorithm with land polygon
- comparation between the coastline input files and the OSM tiles with 
automatic adaption of changes that doesn't flood the tile

Please let me know if you have some additional ideas for this and the 
sea generation in general.

Have fun!
WanMil

> Hi all,
>
> I have started to rework the sea generation code. There have been
> several reports of problems with flooded tiles in the last weeks so I
> think it is neccessary to improve some parts of it.
>
> The first thing I've started is to implement a flooded tile blocker. The
> algorithm has been discussed a while ago on this list. It uses a
> quadtree implementation which is able to return all points contained by
> a polygon. The quadtree is filled with all highway points. If a sea
> polygon contains too many points that belong to highways something is
> wrong and the sea polygon is discarded.
>
> That's the idea... Let's see how it works.
> The quadtree might also be used to correct wrong sea polygons. But
> that's one of the further steps.
>
> Do you have any other ideas how to improve the sea generation and how to
> make it more failsafe? Then please let me know.
>
> Have fun!
> WanMil
>
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>