osm2city.py development

[legoboyvdlp]

Code: Select all

SpawnPoolWorker-1 root INFO     writing cluster w010n50\w008n55\2827339city00603
 with 25 buildings
25 25 100.0%
SpawnPoolWorker-1 root ERROR    Exception occurred while processing tile 2827339
.
Traceback (most recent call last):
  File "C:\Users\Redpath\Documents\GitHub\osm2city\build_tiles.py", line 136, in
 process_scenery_tile
    buildings.process_buildings(the_coords_transform, my_fg_elev, my_blocked_are
as, my_stg_entries, file_lock)
  File "C:\Users\Redpath\Documents\GitHub\osm2city\buildings.py", line 654, in p
rocess_buildings
    cluster_elev, cluster_offset, prepare_textures.roofs, stats)
  File "C:\Users\Redpath\Documents\GitHub\osm2city\building_lib.py", line 1231,
in write
    ac.write(ac_file_name)
  File "C:\Users\Redpath\Documents\GitHub\osm2city\utils\ac3d.py", line 291, in
write
    f = open(file_name, 'w')
FileNotFoundError: [Errno 2] No such file or directory: 'C:/Users/Redpath/Docume
nts/FlightGear/Custom Scenery/projects/NI\\Buildings\\w010n50\\w008n55\\w010n50\
\w008n55\\2827339city00603.ac'
SpawnPoolWorker-1 root ERROR    ******* Exception in tile 2827339 - to reprocess
 use boundaries: -7.25_55.125_-7.0_55.25 *******
Traceback (most recent call last):
  File "C:\Users\Redpath\Documents\GitHub\osm2city\build_tiles.py", line 136, in
 process_scenery_tile
    buildings.process_buildings(the_coords_transform, my_fg_elev, my_blocked_are
as, my_stg_entries, file_lock)
  File "C:\Users\Redpath\Documents\GitHub\osm2city\buildings.py", line 654, in p
rocess_buildings
    cluster_elev, cluster_offset, prepare_textures.roofs, stats)
  File "C:\Users\Redpath\Documents\GitHub\osm2city\building_lib.py", line 1231,
in write
    ac.write(ac_file_name)
  File "C:\Users\Redpath\Documents\GitHub\osm2city\utils\ac3d.py", line 291, in
write
    f = open(file_name, 'w')
FileNotFoundError: [Errno 2] No such file or directory: 'C:/Users/Redpath/Docume
nts/FlightGear/Custom Scenery/projects/NI\\Buildings\\w010n50\\w008n55\\w010n50\
\w008n55\\2827339city00603.ac'
SpawnPoolWorker-1 root INFO     ******* Finished tile 2827339 *******


Full exception log once it finishes. It actually created the directory structure this time but all folders are empty.

Log:
https://pastebin.com/v1B6JKrM

[vanosten]

It must be a problem with how you specify paths I guess. I have never had problems using something like C:\\FlightGear\\fg_customscenery\\TEST, but then again I used something outside of C:\User and without blanks. Unfortunately I do not have a Windows computer at hand right now, maybe you have to write C:\\\User\\bla to escape unicode.

[legoboyvdlp]

Ok. Will try it then.

[vanosten]

I have closed a long time issue. Buildings with inner rings are now displayed correctly again with holes. The following pictures shows parts of Stockholm, where there are lots of inner rings due to the way OSM mapping is done.

[paju1986]

woww! I was wating a long time for this issue to be resolved great job vanosten!

Regards!

[Thorsten]

Hi Rick,

I've finally tried to the bottom of the road lighting issues you are seeing, and I believe now the fact that a normal map was requested by the effect but not actually supplied caused some oddities.

I've now disabled the normal map request in the effect and the result looks more reasonable to me - please give it a check,

There's some oddities in the test area (Prague) you sent as such, for instance there's small patches of different road type inserted into a highway and in the classic renderer some small patches are shown highly emissive (just a segment really...) - can't do anything about these, they're embedded into the incoming model already.

Also, the road mesh is fairly coarse and blocky, so even working diffuse shading generates sharp discontinuities in color where the road slope changes from segment to segment. You ought to be able to pass a flag to the 3d model to instruct the shader to smooth these - not sure how much it improves though, the smoothing sometimes does funky things as well.

There's yet more subtly going wrong - I can confirm the bug with cloud shadows, but I'm somewhat mystified by that one, because it definitely is working with random buildings and the technology is pretty much the same and I don't see obvious bugs. Will keep looking.

[vanosten]

Hi Rick,

I've finally tried to the bottom of the road lighting issues you are seeing, and I believe now the fact that a normal map was requested by the effect but not actually supplied caused some oddities.

I've now disabled the normal map request in the effect and the result looks more reasonable to me - please give it a check,

Thank you very much for the investigations and corrections. I will try in the weekend, when I have access to a capable computer.

There's some oddities in the test area (Prague) you sent as such, for instance there's small patches of different road type inserted into a highway and in the classic renderer some small patches are shown highly emissive (just a segment really...) - can't do anything about these, they're embedded into the incoming model already.

I did not think about continuity of road types - I definitively need to improve that part. I hope that some of the small patches will go away once I have a better land-use model behind the scenes. That is what I am currently working on - a bit optimistically for 2018.1 (will not require changes on the FG side).

[vanosten]

Hi Rick,

I've finally tried to the bottom of the road lighting issues you are seeing, and I believe now the fact that a normal map was requested by the effect but not actually supplied caused some oddities.

I've now disabled the normal map request in the effect and the result looks more reasonable to me - please give it a check,

I can confirm that it looks good now. Thanks a lot! The rest of the residuals are osm2city residuals.

[Thorsten]

Okay - glad it was so simple...

[vanosten]

I have introduced code to rectify the angles of buildings. See the related description in the manual. I was confident enough with the code to set parameter RECTIFY_ENABLED to True by default. If you see any odd behaviour then please let me know.

PS: there will be no changes in e.g. textures for the upcoming FG version 2018.1. So everything running for version 2017.2 is still fine. However as was discussed in some recent posts: I am planning to change considerably how textures are used. More about that when I have POC code to show.

Within the next 2 weeks or so I hope to be able to provide something like in the following image:


The yellow buildings not being in OSM, but instead set by some heuristics (mainly land-use information (grey) and street information (amber)). This will make it possible to populate areas, where there are no mapped buildings in OSM. At least a first step into the right direction.

[Thorsten]

This will make it possible to populate areas, where there are no mapped buildings in OSM.

Sweet

Now we 'just' need to solve the storage and distribution problem of the bazillion terabyte this will need for the whole planet

[vanosten]

osm2city generated buildings (intentionally without textures), where there are no buildings in OSM: https://www.openstreetmap.org/relation/ ... /-159.4581.

Next step is to improve land-use type recognition - which needs to be more fine grained than what OSM offers.

If someone with strong 3D coordinate system knowledge and C++/Python experience has a bit time to spare, then I would be glad if you would ping me. I have the basic code to read BTG files ready in order to get land-use info, where OSM does not have (e.g. https://www.openstreetmap.org/relation/ ... /-159.3896 is lacking "grey" background for residential zones). However my transformation code from the specific cartesian coordinate system to global lon/lat has bugs, which I cannot resolve.

[Catalanoic]

Really nice feature to be implemented, step forward for osm2city into FG. Don't care about online hdd space, temporal constraint.. dont stop!

[Thorsten]

However my transformation code from the specific cartesian coordinate system to global lon/lat has bugs, which I cannot resolve.

Feel free to send me a few details, due to a few year of spacefight I'm usually quite good with coordinate transformations...

[paju1986]

Hi vanosten.

Very nice work.
I want to ask how it will behave when you already have buildings on osm, will it respect the real buildings?. or you have to manually tell what zone needs to be fill with "fake" buildings?.

Maybe a way will be to consider the "building density" of a determined zone, if the landuse tells that is a urban area but it has a very low building density on osm, then the procedural buildings kicks in.

Regards!