Terrain

[rjbradley]

I'd like to know how much improvement there is going to be in the terrain in the new sim? MSTS had its problems: sawtooth cuts and fills, distorted textures on faces in steep cuts and fills and the 8m minimum terrain spacing. Trainz has similar problems. KRS is yet to be seen.
The route building that interests me involves railroads that traversed through very rugged country similar to that of the White Pass & Yukon where Microsoft has visited. Would it be possible to recreate terrain like that on the WP & Y, especially the steep canyon faces and narrow benches that the railway was placed on?

[tdragger]

Take a look at FSX if you haven't yet since that will be the starting point.

[keber]

That's true for general terrain. But what about sharp cuttings and embankments, where at most one meter precision is required to llok it right? Or even ditches?

[Hack]

>KRS is yet to be seen.

Kuju's posting of the Cajon screens hints at a mesh not unlike what we have now with MSTS.

Cheers!
Marc - 3DTrains
http://www.3dtrains.com

[link:www.3dtrains.com/products/index.shtml|http://www.3dtrains.com/images/pro_funit_atsfmain.jpg]

Pro Series v4.0 SuperPacks

[muskokaandtahoe]

Shame. But the math makes a strong argument for it: There are 125,000 terrain polys / tile in MSTS with 8m triangles (32 sq m/poly). That's a lot of polys and it explains why the default for error bias is one, not zero -- they want to reduce the count by averaging things out. 4m triangles would require 500,000 polys / tile (8 sq m /poly). MSTS-X has to deal w/ those facts too.

On the flip side a modern, high end video card can deal w/ a billion polys / second... but not everyone owns those so some sort of compromise is required.

[b]Dave Nelson
SLW Route Design: The Cal-P, 1950.[b]
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/Dancing_Genma.gif
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/4ad3d633.jpg

[keber]

Why polys? Embankments and cuttings should be regarded as part of the track side profile and not part of terrain as it is now. With intelligent track system you could have perfect embankments, cuttings and ditches, independent of terrain mesh.

[dyvach]

>Take a look at FSX if you haven't yet since that will be the
>starting point.

There's a nice paper available at http://www.fsinsider.com/developers/Pages/GlobalTerrain.aspx but it doesn't really answer what's maximum resolution of accepted elevation data, nor how precise the mesh tesselation can get. Since available DEM data doesn't generally go beyond 10m accuracy at this point, is there plans for the simulator to support higher density mesh (and/or terrain data) than that?

[muskokaandtahoe]

That works well until the drainage ditch has to rise back up to ground level. How do you get it to cut off evenly with the terrain? And what if the edge of the poly doesn't align at all with the drainage?

[b]Dave Nelson
SLW Route Design: The Cal-P, 1950.[b]
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/Dancing_Genma.gif
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/4ad3d633.jpg

[rocky]

Well I am going to go with the simulator that has the best solution.

Would it not be possible to increase the mesh only where you choose?
The 'missing' dem could be averaged from the adjacent points and adjusted by hand.

If neither offers any improvement I am giving up as my route does not cut it on 30' highways. That's the Black Canyon on CP(left) and CN(center bridge) Thompson sub and south shore of Kamloops Lake (CP)at Cherry Creek.

http://forums.flightsim.com/vbts/up1/123247.jpg


http://forums.flightsim.com/vbts/up1/123248.jpg

[keber]

>That works well until the drainage ditch has to rise back up
>to ground level. How do you get it to cut off evenly with the
>terrain? And what if the edge of the poly doesn't align at
>all with the drainage?

Ditches maybe more difficult, but they are still solvable. But there should be no problems for cuts and embankments; design programs for railways work that way already for decades. And one of the future simulators (http://www.zusi.de) already has implemented described way of railway building - defining axis and side profiles.

http://www.zusi.de/zusi3/zusi3_01.jpg
(look at the right embankment)
That sort of rototypical railway should be standard for future sims.

[muskokaandtahoe]

> Ditches maybe more difficult, but they are still solvable.

If the problem of ditches smoothly joining the surrounding terrain really can be solved then I'm definatly in favor of being able to include such features in defining a ROW profile.

The function that applies the profile would have to know when to apply a cut or an embankment as well as when to include a ditch (it won't always be needed) and how to handle the rise of the ditch to normal ground.

There is a logic tree to that so it could be coded. Whether any game developer will bother to do that for us is a whole 'nuther question.

[b]Dave Nelson
SLW Route Design: The Cal-P, 1950.[b]
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/Dancing_Genma.gif
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/4ad3d633.jpg

[muskokaandtahoe]

> Would it not be possible to increase the mesh only where you choose?

Sure it's possible. Just select a poly and hit the function to divide in half... you see that in CAD software and route editing is CAD work. I would think a selection for all polys under and/or adjacent to track could also be possible. But will anyone code that for us?

[b]Dave Nelson
SLW Route Design: The Cal-P, 1950.[b]
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/Dancing_Genma.gif
http://i3.photobucket.com/albums/y51/muskokaandtahoe/Avatars/4ad3d633.jpg

[GTW6408]

>>KRS is yet to be seen.
>
>Kuju's posting of the Cajon screens hints at a mesh not unlike
>what we have now with MSTS.
>
>Cheers!
>Marc - 3DTrains
>http://www.3dtrains.com
>
>[link:www.3dtrains.com/products/index.shtml|http://www.3dtrains.com/images/pro_funit_atsfmain.jpg]
>
>Pro Series v4.0 SuperPacks

Would you have a link to see those screen shots of KRS version of Cajon?

Ben

[driver89]

I'm also curious to see how the sim will handle terrain. If they do go with a mesh similar to what we already have, I hope there are some good tools to be able to fine tune that mesh and allow for more detailed and smoother terrain.
I'm also curious to know if there will be an ability to paint textures onto the ground like they have in KRS and Trainz. The screen shots don't really tell us if the textures are tile based or you spray them on as you go.
I personally would like to see a combination of both, where you can have a general texture in a tile, but then you can also quickly and easily paint (or spray) on more detailed textures. This would be usefull for such things as adding gravel or dirt to the area around a signal box, or adding some rocks to a hill. Kuju's system of transparencies sucked, and having to create unique tiles is also a bit of a pain, especially as you end up with all sorts of added terrain texture files.
So I think the ability to use both would be great.

[boleyd]

FSX, in the USA is usually flown with 9meter mesh available from several sources. Higher resolution mesh is not universally available and is expensive. On the ground, where trains run, hillsides do exhibit the sawtooth effect. However, in the air when you are 3000ft above the ground it is not really noticeable. Trains present an entirely different challenge. There is a company called Dreamscapes that is selling mesh that allegedly has used a special interpolation (fill in the gaps in the teeth) that should smooth the display between elevation points in a realistic manner. If the mesh files for MSTSX are the same as FSX this may work. Or, Microsoft may provide 9m mesh, where available, with the elevation between points already smoothed.