SBUILDER X Mesh changes

[JohnC]

Can someone describe how SBUILDER X gets its inputs to change the mesh? I read somewhere that one can smooth out the sharp edge airport flattens , etc. I can (and have ) done similar stuff with grises50. Does sbuilder make a series of nested polygons around the flatten that will all run in one bgl? Or do you simply define a polygon and tell it what altitude to draw it at?

 

[rhumbaflappy]

Hi John.

Each vertex of a "flatten" polygon can have it's own altitude. You just select the propeties of each vertex, and assign the elevation you want.

This isn't a trick of SBuilderX... this is the way FSX does it.

Dick

 

[JohnC]

SBuilder_Mesh_Repair

Of course...so I think I could use an FSX-KML polygon and do the same thing...i'LL TRY IT......i AM WELL ON THE WAY TO RESHAPING A VERY COMPLEX SURFACE USING GRISES50 ..The creation of the countour map is the big time consumer here , especially if you have an altitude range of 50 meters and you want 1 meter vertical resolution..if you are careful, can you create a non-flat runway using hi-res polygons?????

 

[rhumbaflappy]

Hi John.

Search this forum and others for "sloping runways" to get the idea of non-flat runways.

The real problem with that is the AI aircraft will not be able to correctly use the runways or aprons. As far as all versions of FS, runways are flat and level internally. So if you are going to publish your scenery, and you want the AI aircraft to use it, you'll need to keep traditional flat runways and aprons... those can be textured using the FS2002 style groundpolys, as referenced in the gmax forum.

To reshape land using FSX flattens, I just cover the area with triangles, and assign altitudes to each vertex as required. Then you can cover the area with a resampled image if desired. Resampled images in FSX can go to 7cm per pixel ( about 2-3/4 inches per pixel ).

Trying to make a custom mesh BGL with grises50 seems like a lot of overkill. I'm guessing you have a runway that is on a plateau or in a depression. The first thing to check would be is your runway height the actual real-world height? Once that is resolved, then SBuilderX, or FSX-KML can produce flattens that would smooth the depression or plateau as needed.

Perhaps the area you're working on has poor default mesh? Then possibly an upgrade to a better mesh would resolve the problem without trying to make a custom mesh.

Dick

 

[JohnC]

Mesh Repair

If you fly an approach to runway 15 at Burlington Vermont,
you will get an idea of how complex this mesh problem is directly in view. I have got some access to an academic ArCInfo GIS system and am working on finding a way to extract a subset of DEM file for complicated geometries of which this is the worst but not the only example. The DEM file looks like very high resolution up to date SRTM.

FSX KML does allow one to redefine vertex altitudes and reshape the mesh.

As an aside, I like to fly with 20 mile visibility, at dusk or dawn with 7/8 cumulus at 3000 or so feet for VFR flying with a little rain. Much more realism for my taste. I was only speculating about non-flat runways, but never thought about the AI aspect.

If you have expertise with Paint Shop Pro, dividing a topo map with 20 foot spacings into 6 separate color bands is not as hard as it looks. You have to understand how to efficiently do selecting and painting, and control the tool options. If anyone cares, I will go into detail.

 

[GaryGB]

Hi John:

If you get some time while your current protocol is still fresh in the mind, it would be great to have a detailed treatise on this specialized topic for us to better learn from should others have occasion to investigate that approach in terrain development.


Perhaps you could post it as a Wiki here at FSDeveloper:

http://www.fsdeveloper.com/wiki


...and/or as a step-by-step tutorial over at Scenery.Org:

http://www.scenery.org/tutorials.htm



Those who dabble with the Grises package might initially be uncertain about some of the wording and procedures in the included Tutorial (partly due to what may have been lost in translation from that innovative and generous author's native language).


Certainly as Dick points out, there may be alternate methods to address the blending of terrain at an airport, but I'll bet there have been times many of us have wondered if using the Grises tool might be able to achieve a particular terrain development goal, or solve a problem via a learning curve "different" from other protocols the FS SDK might otherwise involve!


Thanks for any help you might be able to share with the FS Community on this specialized topic... which to my knowledge is rather unique among scenery design tools for FS!


GaryGB

 

[GHD]

If you fly an approach to runway 15 at Burlington Vermont, you will get an idea of how complex this mesh problem is directly in view.

The problem at KBTV is not with the mesh, but with the elevations of the river hydropoly:

 

[GaryGB]

Hi John (Cillis, I presume?):

To further detail what George points at in his screenie, on my system with the default terrain mesh set at 100% complexity and 2 meter mesh resolution (I left it set that way as I am working on a project in Europe with 4.75 meter mesh resolution ), the indicated segment of the Winooski River Ultimate Terrain X water polygon seen running alongside KBTV is on the ground at a consistently flat 144.4 ft. / 44 m elevation.


This segment of the Winooski River water body starts below an upstream "terrain mesh-clinging" water body segment with varying elevations that is "stepped-down" to a water level of 252.6 ft. / 80.6 m at the spillway of the Dam by Cascade Park.

That Winooski River segment is then continuous downstream at the said elevation until the West side of "Main Street" bridge, where it appears to again take on a "terrain mesh-clinging" attribute of varying and descending elevation.


[EDITED] The KBTV terrain configuration can also be seen fairly well in either MS Virtual Earth ("VE") or Google Earth ("GE") using "3D, aerial imagery" mode with a terrain view tilted to include the top horizon.


This is a very interesting airport in a beautiful area... with multiple nearby hazards to the pilot, and "significant challenges" for the FS scenery maker!

BTW: I became acquainted with KBTV from flying wildly about in the Decathlon in the first release of Flight Unlimited... great fun for its time!

Also, I believe we once conferred in the past on a KBTV terrain issue in an earlier version of FS, probably at AVSIM forums.


Would I be correct that you were wanting to drop the terrain off the edge of the airport flatten more in keeping with the lay of the land as Dick had asked?


It is indeed possible (by several methods) to replace the existing airport flatten with a smaller custom-shaped polygon that might make blending of the airport into surrounding terrain a bit more to your liking.

If you wanted to keep using the default airport textures rather than a photoreal textured custom ground poly, you might wish to adjust the apron textures with Airport Design Engine (ADE) to better accommodate changes to the airport flatten poly before the terrain fall-off to surrounding mesh and textures.

And if it is of interest for you to implement such a change, I believe there should also be a way to change the attribute of the Winooski River 144.4 ft. / 44 m elevation water poly from "flat" to "terrain mesh-clinging", regardless of the actual elevation as a function of that water polygon being from UTX or FSX default.

FYI: The USGS has seamless NED 10 Meter resolution DEMs for the USA (although many areas appear to be resampled from 30 meter source) which are theoretically higher resolution than the reported FSX default terrain mesh based on 30 Meter source data; possibly this may be more to your preference for the area of KBTV.

[EDITED]: I just checked the USGS NED server, and the data around Burlington and Winooski is apparently genuine 10 meter aka 1/9 arc sec. data (not resampled 30 meter); you should get even better results with that DEM data in ArcInfo (and FSX with a non-default custom-made or other 3rd party terrain mesh).

This data set is reportedly what FSGenesis used for source data in FSG's "Version 2" FSX terrain mesh of the USA.

http://www.fsgenesis.com/Merchant2/merchant.mvc?Screen=CTGY&Category_Code=FSX


I see also that FSDreamscapes now has a 4.75 meter/pixel resolution FSX terrain mesh for Maine (based on InterMap's NextMap low-altitude aerial scans with much higher accuracy than the SRTM data which is IMHO still best regarded as only 90 meter resolution source data); it may not be long before such a package is available for Vermont, as I recall Dean Mountford mentioning here and elsewhere that he has a goal of eventual complete coverage for the USA in high resolution mesh for FS.

http://www.fsdreamscapes.com/store/


PS: I'll be curious how your modifications turn out, as I'd like to do some flying around there in your modded scenery after you're finished, so please keeep us posted on how things go; also, I was wondering if you ever finished the photoreal Lake Champlain ground textures you were working on a while back?


Hope this helps!

GaryGB

 

[GHD]

It is indeed possible (by several methods) to replace the existing airport flatten with a smaller custom-shaped polygon that might make blending of the airport into surrounding terrain a bit more to your liking.

Hi Gary,

No amount of playing with the airport flatten will help. In fact, using the default mesh, most of the "skirt" on this flatten is less than 0.5m from the default height.

 

[GaryGB]

Hi Gary,

No amount of playing with the airport flatten will help. In fact, using the default mesh, most of the "skirt" on this flatten is less than 0.5m from the default height

Hi George:

I am assuming here that John is trying to more realistically render the KBTV airport and surrounding terrain, and that he is alluding to the diversity of elevations and features nearby that make scenery modification a challenge.


Indeed the river undergoes numerous elevation changes along its course from above the dam to below the Main street bridge; the area your screenie points to in the Winooski River water poly is in the "flat" area described above, and I wasn't clear on whether this was the specific area and/or feature that was of concern to John.


If I understand correctly, John is trying to achieve a more realistic fall-off of terrain from the existing airport flatten poly.


Considering how flattens work in FS in relation to John's project, in addition to "pushing terrain down" to a fixed elevation in a classic 1-elevation flatten poly (or even under a variable multi-elevation flatten poly), the flatten will "pull terrain up" to meet the edges of the poly, often creating cliffs above/below the edge of the poly.

Certainly one could assign various elevations to the vertices of a flatten poly where they extend downwards to meet the terrain of the surrounding hillsides around KBTV, allowing one to minimize (or perhaps even avoid) the appearance of "cliffs"; this would yield artificial slopes or "embankments", and as you know where we have slope... we have potential FS local landclass slope texture change, which may or may not be to John's liking.


But if this actually is a project by John Cillis, I would consider it more likely (based on his prior expression of interest in maximizing the realism of his FS experience to better reproduce his real-world flying in that area) that he would want to render the surrounding terrain with good accuracy "sans cliffs"!

However, I may be mistaken here as to the nature of what John (?) is trying to achieve, and what specifically he is concerned over... given the brevity of descriptions in his posts here so far !


If I were NOT mistaken, I'd be speculating that John is trying to achieve a more natural-looking terrain fall-off at the edges of the KBTV flatten poly by implementing a custom terrain based on (real-world?) contours from that area in Grises, which he anticipates being better fitted adjacent to the edges of the KBTV flatten poly after generated as a custom local FS terrain mesh file via resample, and when ultimately rendered in FS.

In some such scenarios (with an airport on a raised area surrounded by descending hillsides), I have found that reducing the extent of an airport flatten size sometimes makes it easier to avoid cliffs at the edge of the flatten poly; and, in general the results with this approach varies, of course, depending on the elevation pattern of a given airport's surrounding local terrain.


In John's case at KBTV, I believe it may be that he could perhaps better fit the custom contours for the surrounding "fall-off" areas he is creating for his local terrain mesh if he keeps the airport flatten poly from getting too close to the edge of the surrounding hillsides (that mostly appear to descend from the airport area) so as to avoid "pulling up" the terrain and forming "cliffs"... rather than the somewhat rounded off slopes we can see in FSX or a tilted GE/VE 3D view.

I suppose we might all achieve a clearer understanding of precisely what John would like to achieve if he would be so kind as to elaborate on what he meant by:

If you fly an approach to runway 15 at Burlington Vermont, you will get an idea of how complex this mesh problem is directly in view.

Perhaps John could clarify further by posting screenie(s) here with some labels as George has so kindly done?


Thanks for this interesting exploration of a challenging terrain project!

GaryGB

 

[GHD]

Considering how flattens work in FS in relation to John's project, in addition to "pushing terrain down" to a fixed elevation in a classic 1-elevation flatten poly (or even under a variable multi-elevation flatten poly), the flatten will "pull terrain up" to meet the edges of the poly, often creating cliffs above/below the edge of the poly.

Fortunately this is not how my flatten works in FSX. Each of the outer vertices has a height identical to the underlying mesh. This was achieved by interrogating the mesh height via SimConnect after excluding the default flatten.

 

[GaryGB]

Fortunately this is not how my flatten works in FSX. Each of the outer vertices has a height identical to the underlying mesh. This was achieved by interrogating the mesh height via SimConnect after excluding the default flatten.

considering how flattens work in fs in relation to john's project, in addition to "pushing terrain down" to a fixed elevation in a classic 1-elevation flatten poly (or even under a variable multi-elevation flatten poly), the flatten will "pull terrain up" to meet the edges of the poly, often creating cliffs above/below the edge of the poly.

Certainly one could assign various elevations to the vertices of a flatten poly where they extend downwards to meet the terrain of the surrounding hillsides around KBTV, allowing one to minimize (or perhaps even avoid) the appearance of "cliffs"; this would yield artificial slopes or "embankments", and as you know where we have slope... we have potential FS local land class slope texture change, which may or may not be to John's liking.

Hi George:

Your method does seem like it could render a rather nicely aligned, "skirted" flatten, especially if it uses a poly with an abundance of vertices to avoid a noticeably planar "LWM3-look", terraced, or "pineapple" surface appearance.

I'll have to explore that option further with a higher vertex count flatten poly configuration to see if I can achieve a good blend of some cliffs which arise upwards (to nearby higher terrain) from the edge of a flatten poly in a project I'm presently working on!


My only concern in John's case is that, at higher terrain FS mesh resolution and complexity settings, there might be some noticeable variation in terrain morphology from what might otherwise be rendered by the contours at the edge of a flatten poly from a custom-aligned terrain mesh.

I'd want to be sure that such a rendering in FS compared favorably with a more precise shape and slope rendition of what Jon is most accustomed to seeing as he flys at KBTV in FS (and the real world!).

If an artificially-created terrain surface "zone" between the flatten planar surface and the surrounding hillside(s) is made at KBTV with flatten polys, and it varies significantly from the shape and slope that John is most accustomed to seeing, the pattern of automatically induced local land class "slope texture" changes (which FS will render at the selected run-time mesh resolution and complexity in addition to superimposed ground shadowing if he also has that turned on) may look substantially unfamiliar.

My thought, assuming John might be likely to share a similar perspective to my own, was this may significantly affect visual experience of the RWY 15 approach more than a pilot (intimately familiar with the real-world KBTV) might be accustomed to seeing there associated with changes in season and time of day.


But, perhaps this might just be of particular importance to someone like myself, who would usually (but not always!) prefer to reproduce a terrain configuration that I know first-hand from the real world... in the most accurate manner feasible, with the most precise source data, available/affordable design tools (even obscure ones!), and according to my present level of ability and interest for a given project.


I'm hopeful that this helps clarify the basis for my perspective on a possible FS flatten blending approach to consider when a higher-precision rendering of a known terrain area may be the objective for a developer according to their personal preference.


Thanks again for sharing your ideas and techniques in this ongoing discussion (...and I hope John joins back in here soon, too) !

GaryGB

 

[GHD]

I have been to Burlington Mas but never to Vermont so I don't know the topography.

This is what rwy 15 approach looks like in FSX with photo textures. The hydropoly creates a waterfall next to the bridges

In fact there is an ox-bow bend which is not depicted by the river poly.



As you can see, the land around the airfield is almost flat.

 

[JohnC]

Wow, let me get started here

I just visited yesterday the University of Vermont Map department, which is very helpful in matters like these. They pointed me to the Vermont VCGI website and I downloaded a recent and accurate set of contour maps from an aerial survey with a laser interrogator, and some number crunching to weed out artificial artifacts. They can be viewed in ArcExplorer(free) , and a composite map created from screen captures. The spacing is 10 feet, which translates to 3 meters more or less, close enough for me. In an hour or two here is where I am with producing a color map prior to dividing the present colors into 3 bands, then producing a gray scale map.If you understand PSP painting and selecting this is not a serious challenge. You do have to be a bit of a grinder for this hobby anyway.
The altitude of the water in feet according to this accurate dataset is 100 feet short of the first dam (called the Salmon Hole) 140 at the top of the dam in central Winooski, 160 feet just short of the Green Mountain power dam east of the interstate, and 190 feet at the top of the dam. Directly in line with the runway is a gorge which is shaped like a V, with tall towers supporting the strobe light. This is a very steep gorge.

I threw together a quickgrises50 some time ago and I will be glad to share my step by step procedure for using this and will put it in fsdeveloper.

As to my photoreaL STUFF, IT SUFFERED FROM RUNOUT BECAUSE IT IS IN STATE PLANE COORDINATES. I wrote a post some time ago on how to get around this w/o $250 globalmapper and wrote an XL program to calculate how to extract rotate and distort to produce a a lat-lon image. Rotating a half degree was not good for the resolution but it worked very well to eliminate runout Maybe someone with globalmapper would volunteer (privatelY)to do the state, but "I hate to ask". I am not at home so I will post again with a screenshot which
should show where I am. Much obliged at all the input and a very interesting discussion. More soon.

 

[GaryGB]

Hi John:

Good to hear back from you, and I'll look forward to reading your further details when you get a chance to follow up again!


GaryGB

 

[rhumbaflappy]

Hi John.

As George indicates, the problem is primarily caused by the flattening of the water polys. You would need to exclude this default water before any additional mesh would work. And, as I mentioned, you'll have to live with the airport flattening, if you want the sim to behave as designed regarding AI planes.

Then the only problem would be getting a better mesh set, if desired. The NED 1 arc second elevational data is free and corrected for errors. It is a 30 meter set...meaning the points are spaced aout 30 meters apart, with a vertical resolution of 1 meter. This will produce an LOD 10 mesh, and could be forced to LOD 11 with a little fantasy involved. I don't know why you would need more than LOD 11 in the sim.

I sense that I may still not understand what you see as a problem here. If its the river, then just exclude the waterpolys, and make new mesh-clinging polys.. or no polys at all if you're doing photoreal. This will reveal the underlying mesh, which is a feature of FSX ( excluding water returns the mesh to it's original state ). If that mesh is not refined enough, then the easily available NED dataset should help you create a very good mesh BGL without needing to convert contours.

The sim can only show terrain to a limited extent... an LOD 11 mesh can only be seen at a very close distance. Then smaller LODs are used, which would defeat any point in using more accurate data. I think Steve Greenwood had worked out the distances a few years ago... I don't think LOD11 is seen at more than 300 yards, and at aircraft speeds of 50 kts. for landing, it would go by in an long eyewink.

I just went to the USGS seamless server:

http://seamless.usgs.gov/website/seamless/viewer.htm

and it appears there is an NED 1/9 arc second dataset for the Burlington area! That's about a 5 meter dataset... good for LOD 13.

Dick

 

[GaryGB]

Hi George:

Indeed a picture is worth a thousand words; I can now see what your earlier screenies were pointing out !

In my posts above, I was referencing an FSX scenario using default FSX terrain mesh (at 2 Meter resolution and 100 % complexity settings), with Ultimate Terrain USA for FSX ("UTX-USA") as the vector source for water bodies and their assigned elevations for the Winooski River segment between the Dam by Cascade Park and the Burlington Main Street Bridge near the KBTV RWY15 approach under discussion and as depicted in your screenie immediately above in post # 13.

I am wondering if your Winooski River water bodies are FSX default vector data, rather than that I've referenced here from UTX-USA ?


Here's a screenie from VE (oops... I meant "Bing" ... Pull Up, Ballmer, Pull Up ) showing the KBTV RWY15 approach; it shows the lay of the land as the Winooski River elevation ranges downstream between 57 meters / 187 ft. and 55 meters / 180.4 ft where it forks underneath the US Interstate 89 bridge.

Here's a screenie from FSX showing the KBTV RWY15 approach in a top down view of the same area.

NOTE: The elevation in this screenie area compared with what your FSX system shows as a waterfall is the same "flat" 144.4 ft. / 44 m elevation segment of the Winooski River above and below where it forks underneath the US Interstate 89 bridge.

Here's the info from TMF Viewer on the "flat" 144.4 ft. / 44 m elevation segment of the Winooski River from the water poly in [FSX install path]\Scenery\UtUsaWater\Scenery\HP12816.bgl

Here's the info from FSX Terrain.cfg on the GUID for that same water poly

[FSX install path]\Scenery\UtUsaWater\Scenery\HP12816.bgl

[Texture.428]
Name=UT_INLAND_LARGE_ROUND_PERENNIAL
guid={6796C963-C25D-45f4-A5B3-203477C370F3}
Textures=OCEAN_SEA_LARGE_LAKE_0
FlattenMode=slope
FlattenPriority=41000
LandClassRemapType=none
ExcludeAutogen=Yes
RenderToTexture=Yes
RenderPriority=81000
Water=Yes


Here's the info from TMF Viewer on the "flat" 144.4 ft. / 44 m elevation segment of the Winooski River from the shoreline in [FSX install path]\Scenery\UtUsaWater\Scenery\HL12816.bgl

Here's the info from FSX Terrain.cfg on the GUID for that same shoreline

[FSX install path]\Scenery\UtUsaWater\Scenery\HL12816.bgl

[Texture.450]
Name=UT_INLAND_ROUND_LARGE
guid={09567D63-2578-4a7e-B198-E590A60DEC5A}
Textures=UTLAKE.BMP,UTLAKE.BMP,UTLAKE.BMP,UTLAKE.BMP,UTLAKE.BMP,UTLAKE_LM.BMP
Layout=3_PLUS_4
StripWidthMeters=32
FlattenMode=none
FlattenPriority=41000
LandClassRemapType=none
ExcludeAutogen=Yes
RenderToTexture=Yes
RenderPriority=90000
Water=No

NOTE: Although it is my intention to be loading FSX with the "Richard Ludowise / Luis Feliz-Tirado Terrain.cfg Fix" distributed some time ago, I'm not sure if any of these parameters were changed by an installer for any of Holger's sceneries that have written to this file.



I hope this helps explain the FSX terrain anomaly you see (and possibly what John sees?) compared to what I see on my system.


Hmmm.. now I too would like to change the attributes in a replacement water poly and shore line for the Wynoochie River where it is a "flat" segment at 144.4 ft. / 44 m elevation to match the "terrain mesh-clinging" segments above and below it !


Time to do some "tracing" in Google Earth using FSX-KML since we cannot "Append" FSX format CVX vector BGLs in SBuilderX (...yet ? )


GaryGB

 

[GHD]

With a little bit of work in SBuilderX and Paint Shop Pro, I was able to replace the hydro polys and add a water mask to the photo textures:



You can see the misplaced extrusion bridge where the loop was missing in the default hydro poly.

This is obviously a quick fix and will not be as good as John's rendition.

 

[GaryGB]

Hi George:

Impressively fast and fine work you do!


In the interim until John further undertakes his fixes here, FSX now has an enhanced "US Interstate 89 Missing Bridge Car Wash" !


Thanks to Dick, George, and John for exploring this interesting FSX development project, and I anticipate seeing further progress here !


GaryGB

 

[JohnC]

Hi all

Custom DEM from grises50 is on the way to Wiki..probably in the wrong place..hope Arno will relocate....

I am waiting for MegaScenery to produce hi-res photoreal for this area. To this end, I used FSX KML and gmax to add a bunch of things 350+ shapes into ONE bgl. The lake shoreline has been redefined, the Lamoile and Missisquoi rivers has been redefined with their islands, many islands missed by Microsoft have been added along with their shorelines, etc. and Colchester Pond has been repaired. Along the way, checking out remote features I have been questioned by suspicious police for going where no man has gone before..at least one is now interested in flightsims...I enclose my quick grises 50 fix since everyone is trotting out their views of KBTV runway 15. One shows the salmon hole and the other shows the grises50 result.
I also enclose a shrunk pic of my new contour picture of the area of interest on the way to a new and better view. I will check out rhumbas' DEM source, but except for grises50 I have never been succesful in this area. I will go back and go over your post more thoroughly rhumbaflappy..