Board index FlightGear Development Scenery

Next World Scenery build

Questions and discussion about enhancing and populating the FlightGear world.

Next World Scenery build

Postby statto » Tue Jun 13, 2017 12:18 am

Is there a plan for the next world scenery build? If so, what's the plan? If not, what needs to happen to make it work?
Custom Scenery available from http://www.stattosoftware.com/flightgear
statto
 
Posts: 2076
Joined: Fri Jan 25, 2008 9:57 pm

Re: Next World Scenery build

Postby erik » Tue Jun 13, 2017 8:58 am

I've also wondered if it isn't about time to do a new scenery build just with updated airports.
Too much has changed for my local airport for example.

I'm not advocating to rush out the new scenery techniques, just a rerun of the old scenery with updated airports.

Erik
erik
 
Posts: 1481
Joined: Thu Nov 01, 2007 1:41 pm

Re: Next World Scenery build

Postby D-ECHO » Tue Jun 13, 2017 10:21 am

I can help with some computing power (2 i3 PCs but maybe for preprocessing..) and data collection/adjusting
User avatar
D-ECHO
 
Posts: 1327
Joined: Sat May 09, 2015 12:31 pm

Re: Next World Scenery build

Postby erik » Tue Jun 13, 2017 11:08 am

I would also be willing to throw some computing power at it if necessary.

Erik
erik
 
Posts: 1481
Joined: Thu Nov 01, 2007 1:41 pm

Re: Next World Scenery build

Postby Torsten » Tue Jun 13, 2017 12:18 pm

statto wrote in Tue Jun 13, 2017 12:18 am:Is there a plan for the next world scenery build? If so, what's the plan? If not, what needs to happen to make it work?

I don't know of any plan for a WS build.
From what I observed, it mostly needs somebody taking lead on this. IIRC, a complete world scenery built from the terragear toolchain was an extremly long running task which failed at random points.
If we could somehow improve the process to handle the rebuild of individual tiles and remove the requirement to build all in one huge process, that would help to continuously rebuild the world.

Anyway - I don't think anybody is actively working on the global scope.

Torsten
flightgear.org - where development happens.
User avatar
Torsten
 
Posts: 604
Joined: Fri Feb 01, 2008 9:22 pm
Location: near Hamburg, Germany
Callsign: offline
Version: next
OS: Linux

Re: Next World Scenery build

Postby D-ECHO » Tue Jun 13, 2017 12:33 pm

@Torsten: is there any possibility that a Change in terragear Makes it possible to regenerate single tiles without having ugly seams?
User avatar
D-ECHO
 
Posts: 1327
Joined: Sat May 09, 2015 12:31 pm

Re: Next World Scenery build

Postby curt » Tue Jun 13, 2017 1:14 pm

We know that even the slightest seam imperfections can turn into ugly visual artifacts depending on the view angle. In order for a newly generated tile to match up against a neighbor, each tile needs to know what the edge of the neighbor looks like (if it has been built yet.) Each edge joins two tiles. Each corner joins 4 tiles. Terragear has a 'first come first serve' policy for defining an edge or a corner. Whenever a new tile is generated, if an edge or corner doesn't exist, then that tile gets to define those. Terragear develops a giant structure of all the shared elements between tiles as it builds up the world.

When a fresh world build is launched, all the shared edge information should be wiped out and recreated as the build progresses. Edges are somewhat complex and subtle ... if a polygon shape spans a tile, there needs to be an edge point at the precise intersection between the polygon edge and the tile edge. Then if you dig in deep to what can happen when you are forced to arbitrarily slice polygons, you have to start thinking about numerical degeneracy, consistency of shapes, etc. Much of the complexity of terragear is there to deal with special degenerate cases that arise when we throw the complexity and randomness of the entire world at our code.

To address the thoughts about sharing the load of a new world build, it has been my experience that the overwhelming bottleneck is data, not computations. An effective world building system needs to access and transfer a *lot* of data, even on a local network, transferring data back and forth, reading, saving, etc. by far outweighs the computational time. Disk bandwidth and network bandwidth are the limiting factors.

Best regards,

Curt.
Aerospace Engineering and Mechanics
University of Minnesota
curt
Administrator
 
Posts: 1086
Joined: Thu Jan 01, 1970 12:00 am
Location: Minneapolis, MN

Re: Next World Scenery build

Postby Lydiot » Tue Jun 13, 2017 2:39 pm

curt wrote in Tue Jun 13, 2017 1:14 pm:To address the thoughts about sharing the load of a new world build, it has been my experience that the overwhelming bottleneck is data, not computations. An effective world building system needs to access and transfer a *lot* of data, even on a local network, transferring data back and forth, reading, saving, etc. by far outweighs the computational time. Disk bandwidth and network bandwidth are the limiting factors.

Best regards,

Curt.


So out of curiosity then;

Is the problem that no one person will rebuild the entire scenery in 'one go', and that there are multiple rebuilds that take place which then necessitates having access to all that data while the data is remotely located?

Or if I put it another way: If we can download all of the current scenery, doesn't that mean that this data transfer could be all local, on one machine? I.e. download it all to a local machine, rebuild, then upload back to the shared server?

I'm just curious because I like others might be able to lend computational power in the foreseeable future.
Lydiot
 
Posts: 921
Joined: Tue Oct 22, 2013 10:50 pm

Re: Next World Scenery build

Postby wlbragg » Tue Jun 13, 2017 5:17 pm

Postby psadro_gm » Tue Jul 10, 2012 9:15 pm
Yes, it is also a goal of mine to document the design, and what exactly each tool does in terragear. This is for the sole purpose of attracting new developers. A while back, Ralf Gerlich did an outstanding job documenting the .btg file format. http://wiki.flightgear.org/BTG_file_format This helps tremendously with brainstorming ideas how to add things like multiple texture coordinates per vertex, etc.

If the terragear toolchain was documented somewhere (not just how to make scenery, but what it is doing to actually generate textured triangles), perhaps more people would give it a longer chance at learning the code. I've heard several people try, then give up. I've been there myself.

Here is the wiki I started a couple weeks ago. I've been busy coding, so I haven't gotten to the good stuff yet...


http://wiki.flightgear.org/TerraGear_Documentation

I know we are headed in a different direction, but it would be so really helpful if this was completed. Especially as a fall-back in case the newer approach falls through.
I would do it, but unfortunately I don't know the code details.
What has already been documented was really helpful
Kansas(2-27-15)/Ohio/Midwest scenery development.
KEQA (2-27-15), 3AU, KRCP Airport Layout
User avatar
wlbragg
 
Posts: 3930
Joined: Sat Aug 25, 2012 11:31 pm
Location: Kansas (Tornado Alley), USA
Callsign: WC2020
Version: next
OS: Win10/Debain/nVGT640

Re: Next World Scenery build

Postby statto » Tue Jun 13, 2017 10:08 pm

I don't have the architecture needed to generate the scenery on my own. There's a lot of data that's on a mapserver that would take the continental U.S. and give it the Europe treatment... there's also other data I have for Australian cities and some other Caribbean islands to add in to the mix.

I also have bash scripts written to run TerraGear easily, but my TerraGear copy doesn't work any more (see the other thread I just posted...).

We also don't need to generate everything at once, especially if there are going to be random failures. Australia honestly would be a good starting point - it's an island, it's fairly large so something will break, it's now got data for nicer custom scenery, and I've checked to make sure there wouldn't be any edge problems along degree lines. Perhaps New Zealand would be an even better starting point since it wouldn't take long to generate. Maybe Madagascar or Iceland as well as a good test to make sure everything works?

You could also split the Americas off from the rest of the world during the generation process so the entire world wasn't generated all at once. We don't really need to refresh the Antarctica scenery, nothing has changed there, and nobody really cares all that much anyway.

Anyway, I'm interested in helping with this process, especially since there seems to be a void here.
Custom Scenery available from http://www.stattosoftware.com/flightgear
statto
 
Posts: 2076
Joined: Fri Jan 25, 2008 9:57 pm

Re: Next World Scenery build

Postby wlbragg » Tue Jun 13, 2017 10:40 pm

There's a lot of data that's on a mapserver that would take the continental U.S. and give it the Europe treatment...

This isn't the "rounded" North American" shapefiles is it?
Kansas(2-27-15)/Ohio/Midwest scenery development.
KEQA (2-27-15), 3AU, KRCP Airport Layout
User avatar
wlbragg
 
Posts: 3930
Joined: Sat Aug 25, 2012 11:31 pm
Location: Kansas (Tornado Alley), USA
Callsign: WC2020
Version: next
OS: Win10/Debain/nVGT640

Re: Next World Scenery build

Postby statto » Tue Jun 13, 2017 10:56 pm

I believe it is.
Custom Scenery available from http://www.stattosoftware.com/flightgear
statto
 
Posts: 2076
Joined: Fri Jan 25, 2008 9:57 pm

Re: Next World Scenery build

Postby D-ECHO » Wed Jun 14, 2017 7:16 am

I think we should start with a list of Landclass data sources:
Vmap0
CORINE
Custom:
Azores
Canarys

Please write if you know more
User avatar
D-ECHO
 
Posts: 1327
Joined: Sat May 09, 2015 12:31 pm

Re: Next World Scenery build

Postby Thorsten » Wed Jun 14, 2017 5:08 pm

We also don't need to generate everything at once, especially if there are going to be random failures. Australia honestly would be a good starting point - it's an island, it's fairly large so something will break, it's now got data for nicer custom scenery, and I've checked to make sure there wouldn't be any edge problems along degree lines.


My thoughts precisely.

If all people who are interested in this get together, pool resources and knowledge and start tackling some islands first to test and then move up to Australia, we have something useful and the experience needed.

And after that, maybe the rest of the world, maybe not.

But start with limited projects, see where it goes.
Thorsten
 
Posts: 9302
Joined: Mon Nov 02, 2009 8:33 am

Re: Next World Scenery build

Postby Torsten » Wed Jun 14, 2017 6:53 pm

I could try installing the terragear chain on scenery.flightgear.org. It has enough storage, a postgres database and almost everything needed. It also runs Jenkins to create the objects from the scenemodels database.
Maybe, i could create a Jenkins job to generate those scenery islands automatically, eventually triggered by some source changes?

Would that make sense?

Torsten
flightgear.org - where development happens.
User avatar
Torsten
 
Posts: 604
Joined: Fri Feb 01, 2008 9:22 pm
Location: near Hamburg, Germany
Callsign: offline
Version: next
OS: Linux

Next

Return to Scenery

Who is online

Users browsing this forum: No registered users and 1 guest