Re[2]: SVG1.1 Geographic Coordinate System

On Thursday, January 10, 2002, 5:28:16 PM, Randy wrote:

RG>    Referencing the GML for a georeferenced coordinate system provides a
RG> standard way for identifying the coordinate system used for SVG coordinates.
RG> However, the SVG cartesian plane is inverted and there is more than one way
RG> to correct the view. Using a viewport mapping seemed to introduce problems
RG> with text orientation so I believe most users of SVG for mapping are simply
RG> inverting the y axis. Does the GML CRS have a standard method of identifying
RG> an inverted y axis on a well known coordinate system such as "UTM-13 NAD83
RG> feet" ?

Thanks for the comments Randy.

Yes, the WG is well aware of this problem. We tried to solve it by
also allowing a choice of a Y-up coordinate system on the SVG element
in SVG 1.1 but it was felt that this would cause implementation
difficulties if, say, a use in document A references a symbol in
document B and A and B have different settings for y-direction.

My feeling is that these are surmountable but there we are.

So the discussion moved to whether we should declare a new coordinate
system (possible but messy, the RDF to do that is really long - I can
post an example if people would like to see it bit it was overlong to
put in the 1.1 spec posted the other day).

Or alternatively, reference a well-known coordsystem that is Y-down
(but its not clear that these exist or are useful).

Or alternatively, say that when we define a geographic coord system
using the pair of well-known coordsystem plus well-known-projection,
that somehow has to take into accout the fact that in SVG Y is down.

As you say there are multiple ways to solve that, so this also has

In my own maps - not being a cartographer -  I do a translate and a y
scale by -1 to flip it - and then have to mess around with scripting
because, as you say, it turns all the text upside down

There is a truly hideous workaround for that but I would rather not
reveal it ;-)

I would be most happy to get feedback on the natural, correct and
useful way to solve this.

The primary use case is that of several SVG layers - for example one
representing all banks in a city, one representing all restaurants in
one area of a city and one representing the streets in a certain
district of the city - and to give each of these metadata which

a) asserts they are maps
b) describes the area mapped

such that they can be composited together in the client to show a
streetmap with nearby banks and restaurants, and courtesy of s tiny
SVG overlay constructed and updated on the fly from a GPS or whatever,
a 'you are here' symbol. This is mainly for the mobile market.

Clearly there are other use cases where it is handy to assert that an
SVG file is on fact a map, and what it is a map of.


Received on Thursday, 10 January 2002 12:53:31 UTC