[OSM-talk] Road cores and casings on standard Mapnik rendering

Phil! Gold phil_g at pobox.com
Fri Jan 20 16:41:49 GMT 2012

* Frederik Ramm <frederik at remote.org> [2012-01-20 09:16 +0100]:
> So then I stuck them into this table: http://wiki.openstreetmap.org/w/images/0/02/Difference_in_Core-Casing.PNG

That is an excellent reference.  Thanks!

> The "motorway" and "motorway_link_" are the strangest as they are
> smaller than all but a few roads of lower status than them.

I don't know for certain but I figured this was because motorways are
almost always in pairs and the thinness is a compensation for that

> Finally the order in which they render has links render very early,
> rather than just before there similarly named road (i.e. motorway_link
> to motorway).

I think the links should render before all roads residential and up
(possibly before service, too); IMHO, you get better-looking results that

> The main issue with this is in flyover junctions where roads don't
> flow nicely into one another; but there are other reasons also.

There are some issues with bridge rendering at the moment, in particular
that _link roads on bridges are *not* rendered as early as non-bridge
_link roads.  I've a mind to sort out the things about layered rendering
that bother me at the upcoming DC hack weekend and then submit a patch.

Can you give an example of a junction that doesn't look good to you?

> In conclusion: The answer to this is a bit opinionated so there
> isn't really one, but the suggestion I would make is that either all
> roads bar maybe residential and service should be the same size with
> the same casing

More prominent roads should be wider, I think, but the casings should be
consistent, yes.

> maybe oneday having lanes= taken into consideration.

My opinion is that lanes= would be better for a specialized rendering,
rather than the "main" map, because I think road thickness should be
correlated to road prominence, and using lanes= could lead to inversions
of that rule in places.

One of my favorite renderings is TopOSM: http://toposm.com/us/index.html
Its rules are very consistent and I like its progression of road widths.

...computer contrarian of the first order... / http://aperiodic.net/phil/
PGP: 026A27F2  print: D200 5BDB FC4B B24A 9248  9F7A 4322 2D22 026A 27F2
--- --
Sendmail may be safely run set-user-id to root.
                       -- Eric Allman, "Sendmail Installation Guide"
---- --- --

More information about the talk mailing list