osgEarth with airport lighting and structures

[Thorsten]

I sort of promised to show this in FG native scenery with Southwest regional definitions and procedural texturing, so here is how it comes out then:




I have to show this off - we now have Fresnel reflection on the runway and the possibility to make the airport keep sand...

[radi]

Fresnel reflection on the runway

Awesome!

What airport is that?

[Johan G]

What airport is that?

KNID, China lake Naval Weapons Test Center. One of my favorites. Well, but not too densely modeled and with little traffic. Very nice on my very low end computer (see my profile for the specs).

I sort of promised to show this in FG native scenery with Southwest regional definitions and procedural texturing, so here is how it comes out then:

That looks way better than global textures (i.e. nearly green as a golf course in comparison ). Also I really like how the mountains look in the second screenshot (i.e. pretty much photo realistic).

That said I think the surface at large (and close to the airfield) looks a slight bit to rough and not arid enough (i.e. with a slightly brighter "base" color and patches of bright and flat sand here and there). I kind of also miss the more-brown-than-green spotted pattern of round bushes that are common in US dessert areas like these (though that pattern "dissapears" at higher altitude). At lower altitude though a lot of small rocks tend to come out.

Three illustrative photos of the area from medium, low and very low altitude:


EA-18G VX-31 over Ridgecrest CA 2009 [Public domain], by U.S. Navy photo by Cmdr. Ian C. Anderson, USN, from Wikimedia Commons


US Navy 051215-N-9277A-003 An F-A-18F Super Hornet assigned to Air Test and Evaluation Squadron Nine (VX-9) returns to its^rsquo, home at Naval Air Weapons Station (NAWS) China Lake [Public domain], by U.S. Navy photo by Cmdr. Ian C. Anderson, from Wikimedia Commons


AH-1W VX-5 launching AIM-9L 1987 [Public domain], by USN, from Wikimedia Commons

[Thorsten]

That said I think the surface at large (and close to the airfield) looks a slight bit to rough and not arid enough (i.e. with a slightly brighter "base" color and patches of bright and flat sand here and there).

Well, there photo-texturing has an edge... The terrain is classified as shrubland, and it has to work pretty much everywhere in the Southwest, so we have to compromise. We could get much closer to the terrain around KNID, but we'd screw up elsewhere. You can use the dust shader effect to adjust aridness locally though.

I kind of also miss the more-brown-than-green spotted pattern of round bushes that are common in US dessert areas like these (though that pattern "dissapears" at higher altitude).

They're there (although perhaps a bit too much on the green side(, and they disappear at higher altitude to avoid flickering Moire patterns) I'll post a few close-up shots later.

[Thorsten]

Dot noise used for the bushes:



Pedestrian view of the scene - with Perlin noise for the rocks. Not perfect, but we don't have too many people walking through FG yet... Basically half the bushes are just painted onto the texture.

[castle]

Hi,

First go at building flightgear/osgearth. Downloaded clones of files called out in this form. osgearth build and install went well, but simgear is complaining.

Linking CXX executable test-simgear_structure-shared_ptr
CMakeFiles/test-simgear_structure-shared_ptr.dir/shared_ptr_test.cpp.o: In function `__static_initialization_and_destruction_0(int, int)':
shared_ptr_test.cpp:(.text+0x2d4): undefined reference to `boost::unit_test::ut_detail::auto_test_unit_registrar::auto_test_unit_registrar(boost::unit_test::test_case*, unsigned long)'
shared_ptr_test.cpp:(.text+0x3e0): undefined reference to `boost::unit_test::ut_detail::auto_test_unit_registrar::auto_test_unit_registrar(boost::unit_test::test_case*, unsigned long)'
CMakeFiles/test-simgear_structure-shared_ptr.dir/shared_ptr_test.cpp.o: In function `virtual_weak_ptr::test_method()':
shared_ptr_test.cpp:(.text+0x7a7): undefined reference to `boost::unit_test::unit_test_log_t::set_checkpoint(boost::unit_test::basic_cstring<char const>, unsigned long, boost::unit_test::basic_cstring<char const>)'
shared_ptr_test.cpp:(.text+0x9bd): undefined reference to `boost::unit_test::unit_test_log_t::set_checkpoint(boost::unit_test::basic_cstring<char const>, unsigned long, boost::unit_test::basic_cstring<char const>)'


boost version 1_43-0

Do I need an upgrade?

JW

[poweroftwo]

Yes, I recommend updating to boost_1_55.

jeff

[castle]

Yes, upgraded from 1_43_0 to 1_55_0 but the error persists. Hacked around it by disabling simgear test applications. OS is Linux.

FYI, running a full scale 737 sim with a 210 degree collimated display.

The posted pics and videos are terrific. When this new feature is up and running, it will be a show-stopper. Unfortunately, pics and videos don't capture the full impact of the virtual images and sense of depth, you have to see it up close and personal.

Nice work!

JW

[Hooray]

Yes, I recommend updating to boost_1_55.

I'd suggest to change the top-level CMakeLists.txt file accordingly for the "osgEarth" build mode then, so that people don't miss this

[Thorsten]

When this new feature is up and running, it will be a show-stopper.

Hm - that's a double-edged expression...

1) one that stops or could stop the progress, operation, or functioning of something
2) something or someone exceptionally arresting or attractive

Language can be funny sometimes - I wasn't aware of 2) at all and pondered the meaning of the sentence quite a bit.

[castle]

He, he

meant it as a positive reference, as when in the theater a particular performance or piece by an actor or singer so overwhelms the audience that the show is brought to a stop by a standing ovation. (Some say English is the hardest lanquage to learn for non-native speakers because of such double meanings and context specific meanings and idioms) but I digress

We're not there just yet, got it running today, but for some reason the scenery tiles from the network sources are not being loaded and/or processed, Guess we can't post images here as attachments so I sent it over to the fg email server.

Might it be a firewall issue? what port is being used to connect to the sources?

JW

[Johan G]

The terrain is classified as shrubland, and it has to work pretty much everywhere in the Southwest, so we have to compromise. We could get much closer to the terrain around KNID, but we'd screw up elsewhere.

Ah, that of course.

Dot noise used for the bushes
...
Pedestrian view of the scene - with Perlin noise for the rocks. Not perfect, but we don't have too many people walking through FG yet... Basically half the bushes are just painted onto the texture.

Looks rather close to me, on particular the dot noise bushes.

[poweroftwo]

A few more views of KNID from the osgEarth Flightgear development with fog now implemented:


KNID with Clouds (Bing imagery as source)


KNID scrub (Bing imagery as source); not as detailed as Thorsten's but not bad. I am sure he could augment the effect on the osgEarth imagery as well as the native Flightgear terrains.


KNID (Bing imagery as source)


KNID in bottom of view (ArcGis imagery as source) from around 130Kft MSL

[Thorsten]

Pretty

I am sure he could augment the effect on the osgEarth imagery as well as the native Flightgear terrains.

You could run dot noise or Perlin on the osgEarth just as well - I guess internally it's just a texture, and we know what to do with textures in a shader. The issue is that I think in photoscenery we've lost the information what terrain is - I can't tag the road in the shader as something not to be covered by dot or Perlin any more, which works in vector terrain because we always know what the landclass is (or do you retain the vector terrain underneath?).

KNID in bottom of view (ArcGis imagery as source) from around 130Kft MSL

What's the problem with the horizon - your fog doesn't seem to blend properly there? I know it's tricky - anything you need a hand with?

[poweroftwo]

Technically, when TerraSync is enabled, the native FG scene data is still present in memory, so yes I would suspect that the class data could be accessed. I would prefer to access some osgEarth layer produced from some global landclass data though.

Here are some examples from the Pelican Mapping team (osgEarth developer) that show detail texture applied to osgEarth.

https://plus.google.com/104014917856468748129/posts/PJ7Jj6sS5iV


https://plus.google.com/104014917856468748129/posts/2MgR9esMq14


-- Fog ---

Yes thanks, I could definitely use the help with fog visibility. The difference is particularly noticeable at night when the light points are enabled. I can produce some screenshots later.

osgEarth utilizes a shader composition framework in order to work well with other shader systems. Description here: http://docs.osgearth.org/en/latest/developer/shader_composition.html

I am using the default fog shader from here:
https://github.com/poweroftwo/osgearth/blob/OsgEarthNext/src/osgEarthUtil/Fog.cpp


thanks,
jeff