[OSM-talk] Actually using OpenStreetMap and the usabilityof the current maps

Juan Lucas Dominguez Rubio jldominguez at prodevelop.es
Mon Jul 28 17:28:35 BST 2008

Hello, Inge. The maps shown at http://www.openstreetmap.org <http://www.openstreetmap.org/>  cannot be taken seriously. They have mistakes (religious symbols out of place), absurd rendering criteria (large pint glasses) and traces of mathematical mediocrity (watch the scale bar in any Scandinavian city : it's simply wrong). It's sad to see those things are not corrected, because it would take a little effort. But somehow, nobody cares.
So those guys are right : if you want a sensible map right now, you need to setup your own renderer.
This is a mail that I have been wanting to send for some time, but wanted to 
think a little more about the subject before I actually did.
The topic is how the maps of OpenStreetMap are actually used by ordinary 
users. I know that the data of OSM is supposed to be used in new exciting 
ways like the cycle maps, but the majority of the users are just going to use 
what the programmers have made available to them.
So the question then becomes, is the current renderings good?  For which 
Before we can discuss how good the maps are, we have to describe the intended 
use cases. I will start with my own here, and hope that you will fill in your 
own ways of using maps in general and OSM in particular.
I recently bought a cheap navigator, but before that I often used a commercial 
Swedish map services to navigate to places when I went there for my work. I'd 
print out the map on paper on a low zoom level, showing where I would go on 
large roads. Then I'd print out maps using higher and higher zoom levels 
closer and closer to my goal so that I can see which intermediate and smaller 
roads that I'd have to take to reach my goal.
So, would OSM work for that usecase? No, I don't think so.  Here is why:
 * Names!  There are far too few names on the map, especially on low zoom 
levels. It's difficult to get a feeling for where you are and orient yourself 
on the map if you cannot find names on the map. The commercial maps show lots 
and lots of names, and that is a good thing. We should make names appear on 
the maps earlier.
* Distinctions between roads. In opposition to the case for names, there are 
too many roads on the large scale maps.  Here is what the current map looks 
like around my home city: 
There is too little distinction between the motorway, the few primary 
highways and the secondary.  I don't think the tertiary highways should even 
be on that map. Once they are all mapped they will provide a messy background 
making the important roads even more difficult to see.
* Marking important roads. In the map above, you can also see that there is no 
marking of even the motorway (E4) or primary roads (in this case national 
roads 34 and 50). This is like names for cities, towns and villages: it makes 
it more difficult to follow where you are on the map.
So, what are other use cases for OSM? Are the current OSM renderings good for 
those use cases? Do we need more different renderings for different use 
I think that OSM has reached a state of maturity where we need to start 
discussing how the default renderings are used in real life.
Juan Lucas Domínguez Rubio
Prodevelop SL, Valencia (España)
Tlf.: 96.351.06.12 -- Fax: 96.351.09.68
http://www.prodevelop.es <http://www.prodevelop.es/> 


De: talk-bounces at openstreetmap.org en nombre de Inge Wallin
Enviado el: lun 28/07/2008 16:59
Para: talk at openstreetmap.org
Asunto: Re: [OSM-talk]Actually using OpenStreetMap and the usabilityof the current maps

On Monday 28 July 2008 16:24:01 Jukka Rahkonen wrote:

> Tilesets are for sure great for serving maps effectively for big audience.
> However, we have already seen that predefined layouts will never make
> everybody happy.  Perhaps one day old-fashioned map servers which are
> rendering maps on demand will have more place again. A public WMS server
> delivering OSM data through  together with a user contributed library of
> SLD files for styling would be a nice thing to see. Unfortunately it would
> most probably be overloaded and slow :(

For what it's worth...  In the next version of Marble[1], we plan on
supporting what we call 'vector tiles'. This means giving all the points,
vectors and polygons of a certain square as one dataset. This dataset will
then be rendered on the fly by Marble.

We hope to achieve at least two advantages:
  * The vector tiles will hopefully represent less data than the pixel tiles.
  * We will be able to create dramatically different renderings from the same
data, thus removing the limitation of the pre-rendered tiles.

Watch out for more about this in 6 months or so.


[1] http://edu.kde.org/marble/

talk mailing list
talk at openstreetmap.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20080728/59e9de65/attachment.html>

More information about the talk mailing list