[OSM-talk] Tagging, map-features and rendering
Martyn Welch
martyn at welchs.me.uk
Wed May 31 12:11:57 BST 2006
Hi All,
I'd just like to share my experiences as a relative newcomer.
On Wednesday 31 May 2006 11:13, Steve Chilton wrote:
> I naively started using the import GPS traces through OSM interface
> route, but soon got frustrated by the wait for file to be validated and
> appear for editing (pure impatience to get on with the task, having done
> some surveying I wanted to get on with the next phase). Then I moved to
> uploading via the excellent JOSM interface.
I've found this as well! However I've also realised that JOSM doesn't upload
the raw data, so have found myself importing my traces into JOSM, so that I
can add the nodes and segments and uploading to OSM as I feel that it is
important that there is some data to back up the placement of my nodes and
segments.
This may become important if we need to:
1) Prove that the data hasn't been copied from a copyrighted source.
2) determine if someone has gone in and randomly vandalised the path of roads.
> My first surveys were
> circular bike rides that didn't cross over, re-trace, or doing anything
> else complicated. So, using the JOSM option to NOT open as raw data
> seemed the way forward - as it joined up the data to segments
> automatically.
Hm, didn't know it could do that!
> Incidentally, when you do import as raw data I find the
> data points very small in the editor. Have I missed something or can you
> re-size these to suit (as you can with nodes in the applet)? Then I did
> a small test survey of local roads near home, producing more complicated
> data points. Subsequent investigation shows that the autojoin now
> produces multiple paths, spurs, segments not in order that they were
> mapped in (important for rendering textpaths with Etienne's excellent
> Osmarender), etc. So, back to entering raw data and compiling the line
> segments in the editor (tagging as I go).
I tend to be logging as I am going about my daily business (though recently I
have found myself taking detours to map the path of roads ;-) ). As a result,
sometimes the data is less than perfect or is very sparse (when I'm driving
for example). As a result I tend to need to utilise the data from many traces
to accurately plot the path of roads - something that can be aided by others
uploading their raw data.
Unfortunately, Lancaster has a lot of places where roads don't start and
finish in logical places, for example, roads will bend around what are now
T-junctions! I also find driving at 30mph not very conducive to recording
road names. But I can get the segments, which I am tagging as and when I can
get concrete information about the extents of roads and their names...
> I started tagging using class=xxxx as that was what the applet's
> drop-down menus seemed to be leading me towards. However, when I
> discovered the Map Features page in the wiki and saw the first output
> from Nick's development renderer and Osmarender I realised this might
> need a re-think. Having scanned the osm-map-features and osmarender
> files I have now fixed on a subset of features that are recognised by
> this system of outputting. Seeing the ouput in SVG format gives you
> clues about tagging and compilation inconsistencies. I have gone back to
> the local data that I have collected and re-ordered segments [usually by
> deleting and replacing the segments as the applet's re-order segments
> tool is not intuitive to use (to me)].
>
I only started tagging after starting to use JOSM, so never knew about the
class=xxx tags! I also assumed that I should create and tag ways, rather than
the underlying segments. It was Osmarender and the Map Features page that
lead me down this path, though I am pleased with the results!
(http://wiki.openstreetmap.org/index.php/Lancaster)
> Enough of these ramblings and let's get positive. I today obtained my
> Garmin bike handlebar holder and am raring to get out some
> evenings/weekends and map more of the local area (Enfield in North
> London). Howvere, can I suggest that some information is put on wiki
> pages that highlights "tagging" matters? Once any responses to this note
> have been assessed I am happy to have a bash at writing that page. And
> also that an "Outputs" page be compiled - preferably with major input
> from the various renderer developers - explaining what they pickup on
> (ways, segments, class, highway, etc), what needs to be considered when
> compiling data, and what external routes onwards there are (eg SVG to
> PNG, SVG to Inkscape).
From my perspective that would be very useful. Also, if they pick up tags from
both the segments and ways, which take priority.
I'm half tempted to tag all the segments as "highway=unclassified" so that
they show up with Osmarender. Being able to print a nice A4 map of the basic
paths would greatly aid my ability to walk around town working out what I've
missed and where ways should start and end!
A step-by-step guide to getting a png would be very useful!
> I don't feel particularly competent to do that
> task. There is already an Osmarender page, but it could use summary of
> what the tool picks up on so folks are aware (forewarned is forearmed).
> Incidentally, reading Jo's mail I would vote for option 3 to auto change
> class= tags to highway= tags (sorry maybe that should read ADD rather
> than replace), and suggest that the idea of "core attributes" be
> propagated strongly amongst contributors to make data as usable as
> possible. Anyone who objects to this needs to shout out as to why they
> have reason to keep their class data (Nick's renderer?).
>
I agree. It would also be nice for these core attributes to be available in
the drop-down selection boxes in JOSM. I wish I had the ability to help and
add that myself...
> Whilst discussions are rightly going on about Foundation, copyright,
> data sources, appropriate development paths, etc it is important that
> the foundation (the data itself) is solid. By that I mean accurately
> compiled, relevantly tagged and accessible data.
>
> [Osmarender hint: if you use the highway=residential tag the 40%
> startOffset placement for the road name will invariably throw the name
> off or truncate it because these roads are often very short. Editing the
> file (in 2 places) to a more suitable number like 20% helps a lot]
>
Would it not be better to center the name in the middle of the road?
Martyn
--
Martyn Welch (welchm at comp.lancs.ac.uk)
PGP Key : http://ubicomp.lancs.ac.uk/~martyn/pgpkey/
More information about the talk
mailing list