You are not logged in.

Announcement

*** NOTICE: forum.openstreetmap.org is being retired. Please request a category for your community in the new ones as soon as possible using this process, which will allow you to propose your community moderators.
Please create new topics on the new site at community.openstreetmap.org. We expect the migration of data will take a few weeks, you can follow its progress here.***

#76 2013-09-17 19:47:48

kendzi
Member
Registered: 2008-08-20
Posts: 250

Re: Real time 3D map using WebGL

It was proposed to add nodes with tags roof:direction=begin|end or some relation with roles
roof:direction:begin and roof:direction:end

Offline

#77 2013-09-18 02:20:40

rayquaza
Member
From: DE-BW
Registered: 2012-11-18
Posts: 2,007

Re: Real time 3D map using WebGL

Wouldn't it be possible (and in most cases even easier) to calculate this from N/NE/E/SE/S/SW/W/NW-Values?

In case there's a communication problem: The question was, if a ball was placed on the center of a building with "roof:shape"=skillion direction=N and it's Way-Segments pointing to 10°/100°/190°/280°, would it roll to 10° or to 0°?

Offline

#78 2013-09-18 05:44:38

things-change
Member
Registered: 2011-10-27
Posts: 351

Re: Real time 3D map using WebGL

Right, that was my question. I think it would roll to 0°. Because this is set with the direction tag.

But the point is, maybe there are some roofs, where the roof direction goes not to one wall. But this are very few roofs, for sure.
But very often a building is not exactly to 0°,90°...., and the roof has its lowest side at one wall. So there has to be a tag that says 'This roof goes down to this wall'.

Offline

#79 2013-09-18 09:04:08

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

rayquaza wrote:

Wouldn't it be possible (and in most cases even easier) to calculate this from N/NE/E/SE/S/SW/W/NW-Values?

In case there's a communication problem: The question was, if a ball was placed on the center of a building with "roof:shape"=skillion direction=N and it's Way-Segments pointing to 10°/100°/190°/280°, would it roll to 10° or to 0°?

It will roll to 0° in F4maps, if you want it to roll to 10° you can specify the roof:slope:direction with clockwise angle in degree from north. My algorithm doesn't try to snap on edge direction (i'm not sure but i think kendzi3D use a snap method).

Offline

#80 2013-09-18 14:44:09

rayquaza
Member
From: DE-BW
Registered: 2012-11-18
Posts: 2,007

Re: Real time 3D map using WebGL

I suggest to use N/NE/…-Values for an approximate direction, which should be used to calculate something to fit to the rest of the object and numerical values (in degree clockwise relative to north) for an exact direction which should be used without additional calculations.

Offline

#81 2013-09-19 18:02:33

Dinamik
Member
Registered: 2010-08-12
Posts: 1,096

Re: Real time 3D map using WebGL

1)
wiki-F4:

Walls will use building:facade:material, building:material or material tags.
Roofs will use roof:material, building:roof:material, building:material or material tags

I think, that getting information about material (and, consequently, about colour) of roof from tags building:material or material is incorrect:
wiki:

building:colour=* Colour of the building facade. See colour=* for possible values.
roof:colour=* Colour of the building roof. See colour=* for possible values.
building:material=* Outer material for the building facade.
roof:material=* Outer material for the building roof.

I see, that, for example, building:colour=floralwhite + building:material=brick is rendered by F4 as building with floralwhite facade and firebrick roof. It is a mistake! Information about material and colour of roof can be got only from tags roof:material and roof:colour, because building:material and building:colour show material and colour only of facade.

2) Do you support tag roof:direction (as a tag roof:slope:direction)? They said me, that now it is better to use roof:direction instead of roof:slope:direction.

3) Could you explain, why this building is not rendered?
1a11ca350acd.png
16c242554395.png
building=apartments + building:parts=horizontal + building:levels=9: building:part=yes + building:levels=1, building:part=yes + building:levels=3 + building:min_level=1, building:part=yes + building:levels=5 + building:min_level=3, building:part=yes + building:levels=7 + building:min_level=5, building:part=yes + building:levels=8 + building:min_level=7, building:part=yes + building:levels=9 + building:min_level=7.

4) I noticed, that map shows name:en instead of name (if name:en exists). Could you, please, transfer this feature to separate option, which can be switched on and off by user? In many cases people want to see name, but not name:en.

Last edited by Dinamik (2013-09-20 05:16:55)

Offline

#82 2013-09-19 18:36:41

chris66
Member
From: Germany
Registered: 2009-05-24
Posts: 10,106

Re: Real time 3D map using WebGL

cmif4 wrote:
chris66 wrote:

any chance to get roof:ridge=yes evaluated?

The ridge proposal has not been approved yet and according to taginfo there are only 294 use of this key around the world so we don't plan to handle it for now.

roof:ridge=yes is used 5.000 times smile


Mapper aus dem Münsterland.

Offline

#83 2013-09-20 09:31:44

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

Dinamik
1) the osm spec has changed, when i did the material/colour code the spec was :
"building:facace:colour/material" specify wall colour/material.
"roof:colour/material" specify roof colour/material.
"building:material" cover the entire building (it was explained as a shortcut for wall+roof tags).

[Edit: building:material & material tags won't be used anymore for roof on next release]

2) no, we don't support tag roof:direction but i could do it in the futur

3) i need my database request expert to understand what happened to this building, it looks like something filters it from my buildings request.

[Edit: during import, multipolygon lines order makes a self-intersecting polygon so that it is not rendered, we're fixing it and it should be ok on the map on next release]

4) this could be discussed, we already talk a lot about that within the team, we plan to use the browser language to choose the best matching name to display.

chris66

Our database relation import does not allow us to handle these tag for now.

Last edited by cmif4 (2013-09-20 10:46:08)

Offline

#84 2013-09-22 18:36:50

things-change
Member
Registered: 2011-10-27
Posts: 351

Re: Real time 3D map using WebGL

rayquaza wrote:

I suggest to use N/NE/…-Values for an approximate direction, which should be used to calculate something to fit to the rest of the object and numerical values (in degree clockwise relative to north) for an exact direction which should be used without additional calculations.

+1

N/NE/...-Values for snap on a line near to this direction, and 0-360° for exact directions is a tagging that I would prefer.

Offline

#85 2013-09-23 15:12:29

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

We just released a beta version of our ground elevation processing (use the big green button on the top right to enable it).

We're already aware that many glitch are still present, we're working hard to try to fix it.

Offline

#86 2013-09-23 16:08:13

Dinamik
Member
Registered: 2010-08-12
Posts: 1,096

Re: Real time 3D map using WebGL

Could you, please, give a link to place, where it is possible to see the result of this ground elevation processing?

Last edited by Dinamik (2013-09-23 16:16:33)

Offline

#87 2013-09-23 16:19:15

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

We get elevation data for the entire world, you can see it everywhere check on some hilly spots like The French Alps.

Last edited by cmif4 (2013-09-23 16:19:43)

Offline

#88 2013-09-23 17:00:13

Zkir
Member
From: Хрустальная Москва
Registered: 2009-02-21
Posts: 6,110

Re: Real time 3D map using WebGL

Interesting, but height for the buildings on slopes has become ambigous:
http://map.f4-group.com/#elevations.ena … i=-168.632

Looks that some change in tagging scheme will be necessary.


Истинные слова не не приятны, приятные слова не истинны.
True words are unpleasant; pleasant words are untrue.

Offline

#89 2013-09-23 17:04:36

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

We know that buildings are not so close to reality anymore when built on a sloppy spot.

For now buildings:parts are not aligned anymore with their "brothers", we're working on a fix to get every parts of a building inheriting from their outline building position.

Offline

#90 2013-09-23 17:07:36

Felis Pimeja
Member
Registered: 2010-10-27
Posts: 1,035

Re: Real time 3D map using WebGL

A little buggy, but looking awesome!

Offline

#91 2013-09-23 17:14:29

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

Felis Pimeja wrote:

A little buggy, but looking awesome!

That's still some kind of "beta version" we got lot of work in progress stuff waiting to be released...

Offline

#92 2013-09-23 20:03:51

!i!
Member
Registered: 2009-11-28
Posts: 3,313
Website

Re: Real time 3D map using WebGL

Looks interesting smile
Sadly the SRTM is in to low resolution at least for my part of the world, so it looks more odd and a lot of details are missed hmm
So there is still a need of getting a crowdsourced alternative, like http://www.opendem.info

As you started this thread with the intention to get feedback/interaction with the rest of the (3D) OSM community:
Personally, I would really like to please you, to work first on a maximum compatibility with other viewers esp. on the S3DB schema!
Even your visual progress is impressive (and as coder I understand, thats more interesting to work on bleeding edge features), it's getting really worse for the whole community, if we get different 'flavours' of micromapping and everybody has a slightly different interpretation of our tags. If we (the 3D community) want to motivate other mappers to enrich their regions, we have to offer unique tagging guides, otherwise we will get very cluttered results, that are spread all over the world roll


privater Account von KVLA-HRO-Mei

Offline

#93 2013-09-23 22:25:33

things-change
Member
Registered: 2011-10-27
Posts: 351

Re: Real time 3D map using WebGL

I agree, but there is the problem that S3DB is a very basic minimum consensus. The world, (and the tagging, of course) is much more complex.
So you are right: We're at the beginning of 3D-Mapping in OSM. When I fly true the F4 map all around germany, or around the world, there are very few buildings which have 3D tags.
I think, this is good: We see, what we tag, although we are just at the beginning. So we now have the chance to see the problems, to find solutions, and create a 'complete' tagging scheme.
So 'we' have to do our work. We have to create a Wiki-Page, that is standard, that is the base for all 3D tagging and that include all 3D tags. One starting page, an this has to be the base for all renderers.
So this is the 2nd step. But when I understand the post #11 by cmif4, the base for their rendering is the S3DB scheme.

What are the problems actually, where does f4 map differ from S3DB?

Offline

#94 2013-09-24 11:48:31

Dinamik
Member
Registered: 2010-08-12
Posts: 1,096

Re: Real time 3D map using WebGL

I've found, that natural=stone at areas (outlines of big stones) are rendered as areas with many small stones. I think, that is is incorrect rendering: areas with many stones are tagged with natural=bare_rock. Natural=stone - one stone.
outline, link:
51af40f5c11f.png

Last edited by Dinamik (2013-09-24 11:54:47)

Offline

#95 2013-09-24 12:08:54

!i!
Member
Registered: 2009-11-28
Posts: 3,313
Website

Re: Real time 3D map using WebGL

things-change wrote:

I agree, but there is the problem that S3DB is a very basic minimum consensus. The world, (and the tagging, of course) is much more complex.

Yep, it's Simple 3d buildings and a first step to get away from the basic LOD1 extruded buildings that we had before.
As you might know, we collect our experiences/impressions from the practical tagging with the schema here:
http://wiki.openstreetmap.org/wiki/Talk … _Buildings
And some started to play with other approaches e.g. to describe roofs more detailed.
So there is no reason, why this shouldn't result in a complex 3d buildings schema wink

things-change wrote:

I think, this is good: We see, what we tag, although we are just at the beginning. So we now have the chance to see the problems, to find solutions, and create a 'complete' tagging scheme.

Not sure if ''complete" is a good target, as describing 3d this way will always be a compromise between comfort of mapping/tagging and the level of details that you can realise.

things-change wrote:

So 'we' have to do our work. We have to create a Wiki-Page, that is standard, that is the base for all 3D tagging and that include all 3D tags. One starting page, an this has to be the base for all renderers.
So this is the 2nd step. But when I understand the post #11 by cmif4, the base for their rendering is the S3DB scheme.

Here I have to disagree with you, we currently don't lack of wiki pages
http://wiki.openstreetmap.org/wiki/3D
http://wiki.openstreetmap.org/wiki/3D_Development
http://wiki.openstreetmap.org/wiki/Micromapping
IMHO currently it makes no sense to start a list  for 3D tags beside buildings, as there are currently to much differences at OSM-3D, O2W Maps, F4, Kendzi on what is how interpreted (e.g. nature, city furnitures, materials, ...)

things-change wrote:

What are the problems actually, where does f4 map differ from S3DB?

Please check the thread backwards to find some problems. Mostly it's the way how building:parts are mixed and support of all roof types.
A visual check can be done by having a look at the already tagged areas (even with support of our community)
http://wiki.openstreetmap.org/wiki/Simp … Demo_areas


privater Account von KVLA-HRO-Mei

Offline

#96 2013-09-24 13:39:05

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

Dinamik wrote:

I've found, that natural=stone at areas (outlines of big stones) are rendered as areas with many small stones. I think, that is is incorrect rendering: areas with many stones are tagged with natural=bare_rock. Natural=stone - one stone.

We had a bug that natural=stone lead to polygons being handled as monolith=alignment. It is already fixed internally and will be ok on further release.

Offline

#97 2013-09-24 14:09:05

Felis Pimeja
Member
Registered: 2010-10-27
Posts: 1,035

Re: Real time 3D map using WebGL

Offline

#98 2013-09-24 14:13:42

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

Felis Pimeja wrote:

height=235 (according to wikipedia it's 235cm tongue)

Last edited by cmif4 (2013-09-24 14:20:00)

Offline

#99 2013-09-27 16:03:18

things-change
Member
Registered: 2011-10-27
Posts: 351

Re: Real time 3D map using WebGL

Hy, here some points about the map:

-Building=yes + roof:shape=mansard is still roof without building on the map:

http://map.f4-group.com/#lat=52.2215825 … hi=149.026

-what about road markings on minor roads? I think they shouldn't have markings

-And tombstones on amenity=graveyard is good looking, but (for me) i think there too much of them.

http://map.f4-group.com/#lat=52.2186492 … hi=149.026

Offline

#100 2013-09-27 16:20:32

cmif4
Member
From: Paris
Registered: 2013-06-10
Posts: 121
Website

Re: Real time 3D map using WebGL

things-change wrote:

-Building=yes + roof:shape=mansard is still roof without building on the map:

This is now fixed (wait for next release to see it on your own), it was an issue about giving shapes but neither height nor levels.

things-change wrote:

-what about road markings on minor roads? I think they shouldn't have markings

This is already under discussion on our support

things-change wrote:

-And tombstones on amenity=graveyard is good looking, but (for me) i think there too much of them.

Graves are added randomly every 2 to 6 meters, i think this is quite close to real life, anyway i'll have a look on the randomizer configuration.

Offline

Board footer

Powered by FluxBB