XP Scenery Generator [WIP]

[Ironmaiden]

Blacky where is it ?

[blacky75]

where is what?

[Ironmaiden]

where is what?

 

The compiled germany data.

[blacky75]

the german data is under the german folder.... all under http://simheaven.com/experimental/Europe/Germany/

A full version (one big file download) might be available tomorrow. Pilotbalu will do it.

For france I'm waiting the Beta 1.2, maybe tomorrow too.... ;-)

[Ironmaiden]

the german data is under the german folder.... all under http://simheaven.com/experimental/Europe/Germany/

A full version (one big file download) might be available tomorrow. Pilotbalu will do it.

For france I'm waiting the Beta 1.2, maybe tomorrow too.... ;-)

 

So this German data is compiled using W2XP 1.1 Beta ?

[blacky75]

beta1.... but I think the only difference with beta1.1 was about something else, not noticable here. You can get DE and fly around ;-)

[Ironmaiden]

I have Nuremberg using W2XP 1.1 Beta

[blacky75]

personaly I prefer the Nürnberg scenery with surrounding fürth... Not that the OSM version is ugly, but the scenery is more what it looks like, as it was dedicated specially to that region ^^

Edited March 15, 2014 by blacky75

[blacky75]

put some nice pics online of RUSSIA here -> http://world2xplane.com/scenery/ .... they've got load of OSM data too.

 

here just one of 'em...

 

[Ironmaiden]

Blacky,

 

 Nice , but why  some of the buildings got weird textures ? 

[blacky75]

maybe they had color info in OSM? Tony can awnser this maybe more precisely.

[tonywob]

Nice , but why  some of the buildings got weird textures ? 

 

Yeah, because someone in OSM chose those exact colours for the building. There are also some really funny bright red buildings in Berlin which look like lego bricks.

 

At the moment, I'm not sure what to do about it, except limit the range of colours, or change the colours to more pastel colours. I'm open for ideas

[AngeloM]

I'd not change colours programmatically, in this way it's simpler to check which objects have strange colours and, possibly, correct them.

If you change colours during the conversion the overall result is obiously better, but this only hide the problem instead of solve it at the root.

[blacky75]

add an option in your soft where the user you can choose

 

-not applying any colors,

-applying as they are,

-or using a pastel palette :-)

 

that would be perfect

Edited March 16, 2014 by blacky75

[tonywob]

It's very difficult, because not only have people used lots of different conventions for specifying colours, e.g. RGB, or things like "brown chocolate", they've also done it sometimes in their own language ;-)

[blacky75]

adding a translator might help ^^ I'd say, for the moment, it's just minor

[tonywob]

Agreed, it's not too bad. I think a quick solution is to use preset colours for red,green,yellow and blue as they seem the most common. The rest will probably just need fixing inside OSM

 

I'll add this into the beta

[mroe]

At the moment, I'm not sure what to do about it, except limit the range of colours, or change the colours to more pastel colours. I'm open for ideas

 

IMO , at this point , a fac/obj-set  with basic (or most common)  wall/roof colors to choose from should be enough.

This makes new columns  for roof / wall color in the csv necessary.

Something which I would recommend anyway

For automatic constructed object , a common palette for different wall- and roof-types could be a solution . (And then choosing the best match)

This gives the possibility to fine-tune the palette  without rebuild the whole world.

[AngeloM]

Is there a way to know which Countries have the most number of contributions?

We know (how?) that Germany and France are very detailed, but what about Countries outside Europe?

 

[tonywob]

@mroe

 

Agreed, the program currently generates unique models for these objects along with the colours, so I think perhaps a shared texture for some of the colours would work, e.g. Red Brick, Brown Roof, etc.. I've tried various approaches in the past, and the current one I'm using seems to work best. I'm experimenting at the moment with some colour options to see what works.

 

Regarding the CSV file, that's a good idea, and I can add it into the best-fit rule to prefer buildings which have matching roofs.

 

I also think we should separate the colours out into a configurable file, so people can change them. This way, the colours aren't hard-coded and people can change them to their liking (and also easily add new colours).

 

Is there a way to know which Countries have the most number of contributions?
We know (how?) that Germany and France are very detailed, but what about Countries outside Europe?

 

Just go of the filesize (compared to real size) I guess. e.g. Look at the difference between the file sizes for Spain vs Germany. Generally, you can get a good guess of how detailed a country is by simply opening up the map on openstreetmap.org and seeing how much white there is. The more white, the worse it will look

[tonywob]

Just a quick tip for anyone converting scenery, have a look at osmfilter

 

Using this filter command decreases the OSM filesize by half. I've just tried it on the UK using this command which strips out all the stuff we don't use.

 

osmfilter england-latest.o5m --drop-tags="highway=turning_circle highway=crossing highway=give_way highway=mini_roundabout highway=junction traffic_sign=* railway=* operator=* ref=* adr_les=* barrier=* bridge=* tunnel=* access=* bicycle=* foot=* horse=* surface=* designation=* highway=primary highway=secondary highway=tertiary highway=motorway highway=service highway=footway highway=road highway=track addr:city= source= source_ref= created_by= addr:housenumber= addr:postcode= addr:post_code= addr:street= phone= amenity=school name= building:street= building:postcode= building:post_code= building:city= building:country= building:county= building:house_number= building:house_name=" --keep="highway=residential building=* amenity=place_of_worship man_made=lighthouse man_made=water_tower man_made=crane man_made=windmill landmark=windmill man_made=tower man_made=mast man_made=chimney landmark=chimney generator:source=wind natural=wood natural=scrub landuse=forest landuse=allotments natural=forest leisure=garden leisure=park landuse=cemetery" --drop="highway=primary highway=secondary highway=tertiary highway=motorway highway=service highway=footway highway=road highway=track amenity=school landuse=commercial landuse=residential amenity=parking landuse=retail landuse=farmland landuse=industrial natural=coast waterway=* natural=water waterway=*" --out-o5m > england.o5m

 

It basically removes all tags from the OSM file which we don't use, and speeds up processing a lot. You then convert it back from o5m to pbf using osmconvert. It shrinks Germany down do 1.2GB, and it processes twice as quick :-)

 

I'll play around with it a bit more and add instructions on the website

 

Edit: I'll get the program to generate this command automatically using the rules file

Edited March 16, 2014 by tonywob

[alpilotx]

If you need distribution statistics about OSM tags, try this one:

http://taginfo.openstreetmap.org/

 

Here for example you can get stats about all entities tagged with "building":

http://taginfo.openstreetmap.org/keys/building#map

[tonywob]

Thanks for the info Andras. Interestingly it also shows how many people have used building=yes, instead of building=residential,house etc. Over 90% of the tags are just generic

[alpilotx]

Or some great data visualization here:

http://www.itoworld.com/

 

For example all about buildings:

http://www.itoworld.com/map/122

[tonywob]

Super, thanks Andras. Will add these onto my blog for people to use.