On Sun, Feb 15, 2009 at 06:05:13PM +0100, Martin Koppenhoefer wrote: > 2009/2/15 David Lynch <[email protected]>: > > On Sun, Feb 15, 2009 at 10:23, Martin Koppenhoefer > > <[email protected]> wrote: > >> 2009/2/15 Knut Arne Bjørndal <[email protected]>: > >>> > >>> > >>> I also seem to recall that if I drew the buildings above the road core > >>> roads could easily get swallowed completely in some cases, which is a > >>> bad idea. > >>> > >> > >> of course the road would get swallowed when the buildings are mapped > >> on them, but this is bad mapping, and not bad rendering. > > > > Is accurately mapping a building that was built above or below a road > > "bad mapping?" > > well, it depends how it is tagged. "Accurately mapping" implies > clearly that the mapping is correct, but I wonder whether different > people have different ideas about what is considered correct. > - For a building above a road I would tag the road tunnel (less > correct but better rendering result) or the building bridge (more > correct but rendering worse).
You want to tag things correctly, yet call every road going beneath a building built over a road a tunnel? That's a perfect example of tagging for the renderer if you ask me. > - For a building below the road we are AFAIK still missing tags and > rendering rules, but I'm also interested in mapping them. The generic > layer=-1 for the building should at least solve the situation a little > bit regarding the rendering. This is not the most usual occurence of > neighbouring buildings and streets though, so I wouldn't consider it > the standard case for determining the best default rendering order. > > > Or accurately mapping the pedestrian streets in > > Edinburgh, Scotland (and some other old European cities, I assume) > > that are only just wide enough for two broad-shouldered adults to > > pass? > > well those for example I would not consider pedestrian but footway > (and bicycle=yes if applies), but anyway they would display better in > mapnik-rendering-order (the one I'm suggesting for t...@h as well) than > in current t...@h. Given a very small street in between buildings, not an uncommon thing in old cities, the current rendering with casings that (for an apparently unsolvable technical reason) go outside the exact boundaries of the building will in some cases swallow the street so much that it ends up being hard to see. I prefer knowing where the roads are to knowing the exact shape of buildings, so imho the current style is better than what you propose. You go on and on about how things would be better if we just did it like mapnik: Prove it by tweaking the rules to be your way and show us an example rendering, or at the /very/ least provide us with links so we don't have to go searching for your perfect examples for ourself. > > It sounds like you're implying that anywhere that the rendering > > of a building happens to cover the rendering of a road is the mapper's > > fault, when there are plenty of places that pure geographic accuracy > > would result in an unusable map. > > No, I don't even consider covering of a street by a building a > problem. Mapnik proves that it is better, just check various places in > densely mapped areas. Find solutions (proposals above) to get usable > maps AND geographic accuracy together. Btw.: we are not drawing maps > but feeding a geodatabase. I strongly disagree that it's better to cover a street with a building, when I'm trying to find my way using a map roads are more important to me than building outlines. In case you haven't noticed: Maps are not geographically accurate. In this instance the fact that we want to make roads wider (which is needed for them to show up big enough) collide with our desire to draw buildings in their accurate positions. If I hand-drew the map I'd just make the buildings slightly smaller to compensate, and make everything _look_ right, but osmarender can't do that. And please keep tagging schemes and such out of this, osmarender is just about displaying the data in OSM. -- Knut Arne Bjørndal aka Bob Kåre [email protected] bobk...@irc
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ Tilesathome mailing list [email protected] http://lists.openstreetmap.org/listinfo/tilesathome
