[OSM-talk] 3D somehow not compatible with our map and editing concepts / capabilities?

[Marco Boeringa]

I agree there are currently issues. It is an unstoppable trend, but it 
would be good if there was better support and enforcing of "best 
practices" so as to avoid loosing the ability to create proper 2D maps 
in a quest to map every detail of 3D (or for that matter Indoor) buildings.

One thing I have posted about before on the OpenStreetMap Forum is that 
it is paramount that users add proper building relations 
(type=building), and group all the individual building:part elements in 
there. Unfortunately, many people don't do this, even thought it is 
written here: http://wiki.openstreetmap.org/wiki/Simple_3D_buildings.  
But admittedly, the Wiki page could be a bit more clear and specific 
about this.

More importantly though, the building relations should ALWAYS include a 
building footprint, with role=outline added to it. The footprint can be 
either a simple closed way, or if inner courts are present, preferably 
be a multipolygon itself. The tags for the Simple 3D building as a whole 
should preferably be on this outline, as the outline can then be used as 
the source data for. If the tags are put on the type=building relation 
(which may seem logical), it will become harder for (automated) data 
users to get to the building tags.

If all Simple 3D buildings had proper role=outline features and building 
relations to group the parts together, navigating the data on the 
OpenStreetMap website or in OSM editors, would be fare less painful. 
Finding the tags related to the building from a selected building:part 
would just mean first moving up to the type=building relation, then down 
to the role=outline feature. The links on the website would properly 
guide you.

Also, the role=outline feature would serve as the source for creating 
proper 2D building maps, allowing filtering out of the outlines, and 
ignoring building:part (if desired).

Marco



    With this mail I would like to open a general discussion, whether it
    makes sense to add detailed 3D data into the current OSM db.

    Living in a historic city with lots of tourists (many of them
    mappers apparently), and lots of famous monuments, I am observing
    for years now, that more and more detailed 3D objects get mapped.

    While at first this seemed to be an interesting (and maybe logical)
    development of some advanced mappers, to further push the limits of
    mapping, more and more doubts have grown in the meantime whether
    this kind of data is sustainable. Particularly because the raised
    complexity leads to many errors, where people recreate already
    existing objects or add localized name tags (or other tags) to
    (building:)parts that are mainly there for geometric representation
    in 3d, but are not the objects that actually represent the feature
    (i.e. those that have most of the tags). Subsequently other mappers
    find these objects (with some tags) and add more, and after a while
    it can become plain chaos, until someone with a lot of time
    dedicates herself to clean the mess up.

    And honestly, I can understand this happening, these objects are
    really complex and after something has been "3D-fied" it becomes at
    least time consuming, if not completely confusing to make any simple
    edit (like adding a new tag), because you have to search the "main
    object" and understand where to put the tag.

    I believe there is something conceptually wrong with adding those
    3D-monsters into the common db and require from everybody to
    understand them, without proper support or hierarchy on an API- or
    editor-level.

    (a side-issue is that many monuments like columns, obelisks and
    similar are modelled as "building:parts", where there clearly is
    nothing that is a building, but rather a massive stone)

    Some examples (load them in your editor to understand what I am
    talking about):

    https://www.openstreetmap.org/#map=19/41.90224/12.45784
    https://www.openstreetmap.org/#map=19/41.90297/12.46658
    https://www.openstreetmap.org/#map=19/41.89591/12.48466 (the
    Trajan's column, a simple column consisting in osm of 9 concentric
    objects! Find the right one, if all of them get their name rendered
    at the same spot in the editor)
    https://www.openstreetmap.org/#map=19/41.89854/12.47695 (the
    Pantheon, countless times there pop up duplicates as nodes)

    What are your experiences?


    Cheers,
    Martin

    _______________________________________________
    talk mailing list
    [hidden email]
    <http://gis.19327.n5.nabble.com/user/SendEmail.jtp?type=node&node=5875747&i=0>

    https://lists.openstreetmap.org/listinfo/talk
    Remove Ads
    <http://gis.19327.n5.nabble.com/template/NamlServlet.jtp?macro=site_payment>



---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20160617/4b6eab25/attachment-0001.html>