Board index FlightGear Media

US Southwest

Screenshots, videos, sound recording etc. taken in/with FlightGear.

US Southwest

Postby Thorsten » Tue May 13, 2014 10:49 am

I'm working on a new regional texture scheme for the US Southwest, and I think it's one of those areas which come out exceptionally nice...

Here are a few postcards from Grand Canyon and surroundings.

Image
Image
Image
Image
Image

I have yet to look at other scenic spots (the Escalante, Zion,...) - Bryce and Arches probably won't come out well, too small-scale structures, and I'm still struggling with the transition across the Sierra Nevada, but I think a vast region can be substantially improved here. Well - needs lots of shader magic of course - there's no way to get this with textures alone.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: US Southwest

Postby sanhozay » Tue May 13, 2014 12:42 pm

Beautiful! 8)
sanhozay
 
Posts: 1207
Joined: Thu Dec 26, 2013 12:57 pm
Location: EGNM
Callsign: G-SHOZ
Version: Git
OS: Ubuntu 16.04

Re: US Southwest

Postby someguy » Tue May 13, 2014 3:00 pm

Very nice, Thorsten. Are you using the real-life geological regions, e.g., the Colorado Plateau, to assign textures?
User avatar
someguy
 
Posts: 1650
Joined: Tue Nov 25, 2008 6:54 am
Location: USA
Version: 2019.1.1
OS: Mac OS X 10.11.6

Re: US Southwest

Postby daveculp » Tue May 13, 2014 4:43 pm

Looks just like home! Excellent work, Thorsten.
User avatar
daveculp
 
Posts: 505
Joined: Sun Feb 24, 2013 2:50 am
Location: Las Vegas, USA
Callsign: DCulp
Version: 2017.3.1
OS: Ubuntu 17.10

Re: US Southwest

Postby Thorsten » Tue May 13, 2014 5:50 pm

Are you using the real-life geological regions, e.g., the Colorado Plateau, to assign textures?


Well, it'd be nice to teach FG geology... it'd simplify a lot of things. No, I go by aerial and ground based imagery as usual and try to minimize tensions between different trial regions. The whole things gets really messy where the geology changes, for instance look at Carson going up to Lake Tahoe here

Image

The red rock strata change quite quickly to something rather different, which is a challenge... That's what FG makes of it - not perfect, but okayish:

Image

Looking at more scenic spots in the region:

That's how Yosemite Valley comes out (yep, that's perfect):

Image
Image

Image

Zion National Park (close enough):

Image

Image

And Panamint Valley (the lakebed could be a bit more blurred, but otherwise it's pretty good):

Image

Image
Last edited by Thorsten on Tue May 13, 2014 5:59 pm, edited 1 time in total.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: US Southwest

Postby J Maverick 16 » Tue May 13, 2014 5:58 pm

Awesome scenery!! I love USA!!! :D
Breakin' the sound barrier every day!

Scenery designer, basic livery maker, aircraft developer (KC-137R, F-16, A330, C-32J, MH-60)
Flying with Thrustmaster FCS Flight Pack + MFD Cougar
Aviation media on IG at: @j_maverick16.aviation
User avatar
J Maverick 16
 
Posts: 910
Joined: Sat Feb 01, 2014 7:26 pm
Location: Northern Italy
Callsign: J-Mav16
Version: 2020.3.19
OS: macOS

Re: US Southwest

Postby galvedro » Tue May 13, 2014 6:28 pm

Thorsten wrote in Tue May 13, 2014 10:49 am:I think it's one of those areas which come out exceptionally nice...


Indeed.
galvedro
 
Posts: 145
Joined: Mon Oct 07, 2013 1:55 pm
Location: Sweden

Re: US Southwest

Postby hvengel » Tue May 13, 2014 7:51 pm

Thorsten wrote in Tue May 13, 2014 5:50 pm:...The whole things gets really messy where the geology changes, for instance look at Carson going up to Lake Tahoe here

...

The red rock strata change quite quickly to something rather different, which is a challenge... That's what FG makes of it - not perfect, but okayish:



I drive through Carson City up to the lake and back about 5 times a week (my office is a 5 minute walk from the beach) so I am very familiar with this area. The something else is decomposed granite which is close to gray but has a slightly tan/red tone. The photo you are using is over saturated and a little underexposed so the "red rock" is not nearly that red IRL. The FG version of the Carson City, Lake Tahoe area is slightly too red at lower altitudes and not red enough at higher altitudes as the color contrast between lower and higher elevations is not nearly as distinct as it appears in either the photo or in your FG rendering.

The lake is very deep (over 1600 feet at it's deepest point) and is very clear so most of the lake should be a very deep blue but not as much as in the photo. The Carson River (the stream in the valley) is fast running and tends to be more brown than blue IRL.

The valley floor is both too green and not green enough. The area to the south of Carson City (off to the left side of the photo) is a valley floor that is used for hay and alfalfa and also for grassing cattle. These fields are irrigated so there will be distinctive patterns related to contrasts between irrigated and not irrigated areas. If this was not irrigated it would be high desert with very little green even in the spring. The photo is clearly a spring time photo with lots of green and some snow in the mountains. Most of the year there is less green than in the photo.

Over all this rendering is a big improvement and with a few tweaks could be very close to IRL.
hvengel
Retired
 
Posts: 1127
Joined: Sun Dec 24, 2006 5:35 am
Location: Minden Nevada

Re: US Southwest

Postby Thorsten » Wed May 14, 2014 6:51 am

Over all this rendering is a big improvement and with a few tweaks could be very close to IRL.


The problem is that this isn't 3d modeling, but trying to give the right hints to an essentially automated process. The regional definitions hold for an area of ~200 square degrees, that's about 2 million square kilometers. I am stuck with whatever the landcover data is (it sure isn't CORINE resolution) and all I can do is to try to lessen the problems of the landcover classification - I can't re-classify terrain.

For instance, the upper rock strata are that grey, because they have to have that color at Yosemite, and Yosemite is too close that one could make a different regional definition.

The valley floor is too green because patches of the same landclass appear further in the Sierra Nevada (even close to Yosemite) where the red appears totally out of place in an otherwise green/grey scene - so in order to minimize the tension, I have to make it too green for Carson and too red for Yosemite.

I can't use any proper irrigated fields texture in Utah, because the landclass appears in the Uinta mountains, and a summit covered by irrigated fields looks really silly.

It's these kinds of problems which limit the accuracy of what I'm trying to do - I have to compensate by using landclasses as proxies for geology, regionalize 'islands' (the Uinta mountains are for instace an isolated region of forest patches surrounded by a 'sea' of desert) and so on.

Sometimes I think it'd be cool to re-generate the scenery putting geological info in to get at least the rock strata right... The problem is that this ain't even encoded in the landclasses right now - terrain at the Tahoe shore is EvergreenForest, and all the granite you see is drawn by the shader which just got told by hand that it should make the rock grey. Needless to say, the procedure is very coarse.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: US Southwest

Postby f-ojac » Wed May 14, 2014 12:15 pm

We'll have to wait for NLCD to arrive.
f-ojac
 
Posts: 1304
Joined: Fri Mar 07, 2008 10:50 am
Version: GIT
OS: GNU/Linux

Re: US Southwest

Postby Thorsten » Wed May 14, 2014 12:18 pm

Probably also doesn't come with a geology layer though...
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am

Re: US Southwest

Postby Hooray » Wed May 14, 2014 2:48 pm

Thorsten wrote in Wed May 14, 2014 12:18 pm:all I can do is to try to lessen the problems of the landcover classification - I can't re-classify terrain [...]
Probably also doesn't come with a geology layer though...


Sightly offtopic, but we've been considering to expose so scalled ESRI Shapefiles to Nasal/Canvas for charting/mapping purposes, i.e. all the stuff that's no longer accessible after the scenery is "compiled", including things like borders, coast lines and so on - after talking with psadro_gm, that's apparently much easier than telling TG to encode such info somewhere for /potential/ late ruse.

Such shapefiles could live somewhere in $FG_ROOT, or even fetched on line.

So far, we've been contemplating to add a new Canvas::Element, but if this is something that could be generally useful outside Canvas, we could certainly use cppbind to expose shapelib/OGR GDAL to Nasal, so that such info would become available. I understand that osgEarth already links in those two libs, so dependency-wise this may not be a huge problem - otherwise, we'll have to use shapelib itself.

If this is not exposed to Nasal but just to Canvas, it would still be possible to render such information to a texture and access that in your effects/shaders.

I haven't touched the canvas/shader stuff in a while (and it needed to support effects, not just shaders), but back then it worked nicely - so basically, there's a workaround possible even if we'll just add ESRI support to Canvas. The main thing is keeping it a generic pipeline where a canvas can use shaders and effects, but also become accessible to the effects shader framework - anything else could then be built on top.

PS: I was really surprised seeing a frame rate counter in some of those "photos" :D
Please don't send support requests by PM, instead post your questions on the forum so that all users can contribute and benefit
Thanks & all the best,
Hooray
Help write next month's newsletter !
pui2canvas | MapStructure | Canvas Development | Programming resources
Hooray
 
Posts: 12707
Joined: Tue Mar 25, 2008 9:40 am
Pronouns: THOU

Re: US Southwest

Postby Thorsten » Thu May 15, 2014 6:46 am

If this is not exposed to Nasal but just to Canvas, it would still be possible to render such information to a texture and access that in your effects/shaders.


I think this'd be cool, but we should spend a few thoughts beforehand on how to best pass the information and how to represent it to the shader.

For instance, the global water depth map is cool, but I'm not sure we'd want to pull the same gig with the terrain. First, right now it is unconditionally loaded, adding a large chunk of memory consumption regardless of whether you use it or not. That'd unfortunately also be true for other maps we pass - although assembling the texture dynamically might ease this.

Then the lookup function for the map is actually rather complex and runs per vertex. This isn't a problem for water, as the ocean mesh is very wide and has few vertices, so we can do a huge number of computations in the water vertex shader. But it is a problem for terrain without a LOD system.

On the other hand, the actual info we need is /probably/ rather small - given the possible range of rock colors, I think even a single uniform for the whole scene might encode sufficient information to do much better than we're doing now (perhaps like my water color interpolation code).

So there's always a tradeoff - having such info available would be nice per se, but the gain in visual quality needs to be in a reasonable relation with the computational effort.
Thorsten
 
Posts: 12490
Joined: Mon Nov 02, 2009 9:33 am


Return to Media

Who is online

Users browsing this forum: No registered users and 2 guests