Hi everybody.

Yes, Wiktor get it right. I'll just add a few thoughts:

2. It might be of some value, if the site would present where the
> model is used (maybe some overpass query?)
>

Yep, it would be nice to have such statistics.
If models binded to osm objects via tags - it's rather easy to get
statistics,
using Overpass or some kind of diff listener bot.
But as for me it's not a number one priority.

Most difficult part is:

There is also the issue of putting the object on the map - what
> physical size object has, and what's the orientation. This could be
> provided with any coordinate system and some conversions might be
> needed (proj library might assist you in this task).
>

As for scale conversions i think that the main issue would be to find out
> the correct scale based on user's input, so perhaps it would be the best to
> provide a unit standard (for example 1 blender unit = 1 meter) and in this
> way we can defer the required scale to 3d modelers instead of trying to
> guess or approximate scale.
>

>From my point of view, scale isn't a big problem. As a rule, for
architectural models 1unit = 1meter.
Orientation and matching model coordinates origin with osm object
coordinates much more difficult.

I have a small propozal http://wiki.openstreetmap.org/wiki/Relation_3D:Model
but all interesting things about it were said here
http://forum.openstreetmap.org/viewtopic.php?id=28248

As I think after this conversation, there is no need to use relation, or at
least new relation.

Orientation could be specified on osm object via tag 3d:model:azimuth
and as fallback (if tag is missed) could be calculated as orientation of
longest edge of bounding box.

So the hardest thing is origin.
For 3d models 0 point of coordinates might be located at:
1. the center of model
2. weighted center of model
3. at the corner of a model (to have positive values along all axises)

So first what we need to do is to choose what we use as an origin of the
osm objects.
(I think analog of postgis ST_PointOnSurface should be good enough for this
purpose
http://gis.stackexchange.com/questions/76498/how-is-st-pointonsurface-calculated
).

Second, we need a tool, which allows user to move zero point of a model
using 3d lib application interface.
It might be semitransparent model footprint above osm background.

Such tool will allows user to check and edit azimuth and model zero-point
at once.

2015-03-25 3:36 GMT+05:00 Peter Barth <osm-p...@won2.de>:

> Hi Kristijan,
>
> sorry for the late reply. Wiktor already answered most of your questions
> and gave you some more details. So I'll limit my answer to the open
> question.
>
> Kristijan Trajkovski schrieb:
> > Also what should go into the 'managing models' part of the application?
> > Updating models with new versions, deleting models, something more?
>
> That's what we thought of, yes. There might be even different people
> modelling the same object. E.g. person A and B, each one has a nice
> model for the Eiffel Tower. Additionally it might be a good idea to have
> different level of details for a model. Starting from a rough model with
> just a view primitives/vertices/triangles up to high detail models,...
>
> Peda
>
>
> _______________________________________________
> dev mailing list
> dev@openstreetmap.org
> https://lists.openstreetmap.org/listinfo/dev
>



-- 
Thank you for your time. Best regards.
Dmitry.
_______________________________________________
dev mailing list
dev@openstreetmap.org
https://lists.openstreetmap.org/listinfo/dev

Reply via email to