Hi! I am working on sea polygons once again. After reading through the many generate-sea-related topics I am still not sure what the current state of the art is on the matter. - What is the current recommended command line option/combination that works for you? What other techniques do you need? I have e.g. read about manual adjustions to the data in some posts. Or did it not work for your area at all? - What level of confidence do you achieve with generate sea? Many posts still mention minor failures, disappearing islands etc. - Is it possible to save the resulting sea/land polygons for later re-use? Apparently the OSM data changes more frequently than the actual coastline and re-generating it from scratch every time you update is not necessary. It would be helpful to keep the sea data around, especially if you had to do manual adjustments to create it. But I have not read about such a feature. Did I miss something? thanks Nop -- View this message in context: http://n2.nabble.com/State-of-the-coastline-art-tp4829364p4829364.html Sent from the Mkgmap Development mailing list archive at Nabble.com.
On Tue, Mar 30, 2010 at 11:08:40PM -0800, NopMap wrote:
- What level of confidence do you achieve with generate sea? Many posts still mention minor failures, disappearing islands etc.
As far as I am concerned, it just works. Any errors will be logged and will come through my error message filter. You can fetch my scripts and files at http://www.polkupyoraily.net/osm/ and adapt to your area. Also, when debugging coastline errors (gap in coastline, coastline not extending to map border), you may want to use a command like this: osmosis --rx finland.osm.bz2 --tf accept-ways natural=coastline --used-node \ --wx finland-coastline.osm The resulting file should be so small that you can load it in JOSM to see the way and node ids of the broken coastlines.
- Is it possible to save the resulting sea/land polygons for later re-use? Apparently the OSM data changes more frequently than the actual coastline and re-generating it from scratch every time you update is not necessary.
Generating the coastline is not that slow any more. If you absolutely wanted, I guess you could generate a separate map layer with just the sea, and compile your regular map without --generate-sea. Marko
On Wed, Mar 31, 2010 at 12:30 AM, Marko Mäkelä <marko.makela@iki.fi> wrote:
On Tue, Mar 30, 2010 at 11:08:40PM -0800, NopMap wrote:
- What level of confidence do you achieve with generate sea? Many posts still mention minor failures, disappearing islands etc.
As far as I am concerned, it just works. Any errors will be logged and will come through my error message filter. You can fetch my scripts and files at http://www.polkupyoraily.net/osm/ and adapt to your area.
can confirm that. now also --generate-sea=multipolygon works and didn't have any failures on whole california which has some real complex coastlines. don't remember the exact mkgmap version. any newer one will do. I like it more this way because I don't need to have a typ file
Also, when debugging coastline errors (gap in coastline, coastline not extending to map border), you may want to use a command like this:
osmosis --rx finland.osm.bz2 --tf accept-ways natural=coastline --used-node \ --wx finland-coastline.osm
xapi is your friend and can achieve the same without downloading the planet or excerpts wget http://www.informationfreeway.org/api/0.6/way\[natural=coastline\]\[bbox=-124.5,32,-114,42\]-O coastline.osm
The resulting file should be so small that you can load it in JOSM to see the way and node ids of the broken coastlines.
- Is it possible to save the resulting sea/land polygons for later re-use? Apparently the OSM data changes more frequently than the actual coastline and re-generating it from scratch every time you update is not necessary.
Generating the coastline is not that slow any more. If you absolutely wanted, I guess you could generate a separate map layer with just the sea, and compile your regular map without --generate-sea.
Marko _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
On Wed, Mar 31, 2010 at 09:27:19AM -0700, Apollinaris Schoell wrote:
Also, when debugging coastline errors (gap in coastline, coastline not extending to map border), you may want to use a command like this:
osmosis --rx finland.osm.bz2 --tf accept-ways natural=coastline --used-node \ --wx finland-coastline.osm
xapi is your friend and can achieve the same without downloading the planet or excerpts wget http://www.informationfreeway.org/api/0.6/way\[natural=coastline\]\[bbox=-124.5,32,-114,42\]-O coastline.osm
Yes, xapi can be useful if you are interested in particular features, but it won't help you in debugging a problem with a particular planet extract. The data may have changed since the extract was taken, or the extract may not completely contain some features. I prefer to work with extracts, because I don't have access to a fast mirror of the OSM database that would quickly produce country-size map extracts on demand. Someone pointed out to me that the Geofabrik cutting polygons aren't set in stone. Indeed, I emailed Frederik Ramm today, and he promptly sent me the boundary polygon in JOSM format. I edited it slightly in order to exclude or include a few islands, and sent it back. It will be interesting to see if tomorrow's extract works "out of the box" (with the exception of tweaking the endpoints of the mainland coastline). If not, in a few iterations it will. (Now that I think of it, I should also fix the broken lake multipolygons near the Finnish-Russian border by editing the Geofabrik cutting polygon.) Marko
Hi! Just to let you know. Encouraged by your statements, I tried to do the German coast along the baltic sea with --generate-sea=multipolygon. The map was made up of 3 tiles. The result was rather disillusioning. In about 1/4 of the coast it looked good, on half of the coast the water was missing completely and about 1/4 was inverted with the land flooded and the sea clear. My conclusion is that I'll stick to the old technique for the time being (manually edited sea polygons derived from the shapefiles used by mapnik). bye Nop -- View this message in context: http://n2.nabble.com/State-of-the-coastline-art-tp4829364p4839472.html Sent from the Mkgmap Development mailing list archive at Nabble.com.
Am 01.04.2010 22:24, schrieb NopMap:
Just to let you know. Encouraged by your statements, I tried to do the German coast along the baltic sea with --generate-sea=multipolygon. The map was made up of 3 tiles.
The result was rather disillusioning. In about 1/4 of the coast it looked good, on half of the coast the water was missing completely and about 1/4 was inverted with the land flooded and the sea clear.
I use --generate-sea=extend-sea-sectors for the germany extract with some good results. Chris
What was the input? a geofabrik download or a osmosis extract with complete coastline ways? then you should try --generate-sea=extend-sea-sectors as Chris recommends. I create all osm files with osmosis and coastline will be complete and touch tile boundaries. another idea is to take the area.list from processing a germany extract and feed it together with a planet or europe extract into splitter to get complete coastlines. On Thu, Apr 1, 2010 at 1:24 PM, NopMap <ekkehart@gmx.de> wrote:
Hi!
Just to let you know. Encouraged by your statements, I tried to do the German coast along the baltic sea with --generate-sea=multipolygon. The map was made up of 3 tiles.
The result was rather disillusioning. In about 1/4 of the coast it looked good, on half of the coast the water was missing completely and about 1/4 was inverted with the land flooded and the sea clear.
My conclusion is that I'll stick to the old technique for the time being (manually edited sea polygons derived from the shapefiles used by mapnik).
bye Nop
-- View this message in context: http://n2.nabble.com/State-of-the-coastline-art-tp4829364p4839472.html Sent from the Mkgmap Development mailing list archive at Nabble.com. _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
participants (4)
-
Apollinaris Schoell -
Chris-Hein Lunkhusen -
Marko Mäkelä -
NopMap