Are your DEM gradients wooping you ... Take back control !

Well, that sounds good - but I also need my track plan overlaid on the real world imagery as I am creating it, which poses a problem. To my knowledge no CAD program lets me do this. TransDEM is made to, only its editor thinks solely in terms of straight lines, not arcs.

Do you know of a workflow I can use for several hundred miles of track, to overlay it in CAD over sateĺlite imagery?
Sorry for the delay but Trainz is not a high priority in my life right now. It's setting on the back burner and has been for 6 months or so and probably will continue to be for a few more. I did find some time recently to finish some new especs for an old friend's DHR locos and I have committed to some forensic scripting work for another Trainz content maker but apart from these I have other more pressing things to work on.

Your quote: "To my knowledge no CAD program lets me do this." - As to your personal knowledge I can't attest to that but personally I have done this using various cad programs over the past 15 to 20 years so I'll have to disagree with your premise at least.

In any case many cad programs do allow you to import an image file as a background and "trace" lines or any other drawing objects over it. I use turbocad - an old ver 9.0, also an old version of Autocad, and am testing out a free open source cad program called FreeCAD. While I haven't used FreeCAD for any production work it has a ship design module and as a retired naval architect I'm hoping I find it rather interesting to play around with. CadRail is another small and limited 2.5 D drafting program I use a lot. It's designed specifically for model train layout design so has a few features tailored for track work design. I use it for some basic 2D drafting and for Trainz I can make full sized drawings of real railroads with it. All of these cad programs I've mentioned allow you to import an image file as a background and then trace out a drawing over it.

I've done my home Trainz layout, the EBTRR in it. A small 3'ng railroad that stopped commerical service in 1956 almost 80 years after the main was built. Most of the main line track is still on the ground but unusable. With 32 miles of main line and 10 branch lines and spurs, it's over 100 miles of track lines, 3 main yards and 1 with several miles of standard, dual and ng track. I have to admit I did not trace the track route (current one I'm using) overtop image files in the cad program(s). I created the dxf files from data take from the EBT's 1919 survey maps since they inidcate the station number (feet from along the main from a reference point in Mt Union, PA) of each curve start and end point, each turnout point and every grade change, crossing an bridge locations along the main and branch lines along with other interesting things like the radius and central angle of each curve and elevations at each grade change and vertical transition curves.

However when I 1st started the route back in TRS2004 I did import images of the Tiger data into my cad program and used that to trace out the track route using only straight tangent sections and circular arcs since that is what the EBT used when laying the track on grade. However any drawing tool was available and transition curves could have been added if I wanted - for some reason - to add them in. Now I limit tracing image files to primarily working with the yard trackage in some locations. I have modified the traced drawings based on my own limited surveys taken in the field or descriptions of other peoples work.

The approach I used is outlined in general terms below and I've used basically this approach in many application where I've traced over images - not just for track work in Trainz:
1) obtain an image file that is geo referenced to the terrain defined by the DEM you are using. A UTM projection is the most suitable. I'm not sure how ortho imagery compares with this projection.
2) import the image file (must be in a format supported by the cad program - conversion utilities do abound though)
3) geo-referencing is a problem typically you will have to position the image properly so the coordinate grid used in the dwg aligns with the UTM grid of the image for the zone in question. Cad programs typically allow you to scale, translate and rotate complete drawings or any objects within the drawing which sometimes is necessary to correct alignment problems
4) the dwg units should be in meters but this can be corrected after the fact. If the image file is a full size UTM projection you won't have to make any major adjustments later.
5) using the tools available create the track route over the image file
6) scale and units aren't too important since the cad program allows you to scale the drawing afterwards.
7) export a dxf file of the drawing and import it into TransDEM

"TransDEM is made to, only its editor thinks solely in terms of straight lines, not arcs." Well the cad capabilities are I'll grant you not up to the standards of a 2d drafting program. But I don't believe that was Roland's aim here. It does allow you to pegout a route over the displayed DEM or an imported image of the area that is sufficient to allow filtering of the trainz baseboards created so you can limit their number and confine them to areas where the track is to be laid. I think you're asking a lot more for the 30 USD or so you paid for TransDEM than you're entitled to IMHO. The little cad program for model railroad layout I mentioned above is IIRC 95 USD. Updates are cheaper and the 1st version I got was probably less than 50 USD. So TransDEM with a better cad package in it should be worth a bit more than 30 USD you paid.

I'll also note in passing: Tracing over orthro pics or such won't necessarily allow you match the radius or start and end points or where the transitions start and end either to the accuracy you might be thinking of. I've tried using the push pins in google earth pro (I've an early post in this thread about this) and it's just an approximation as good or maybe better than many use in Trainz route generation. So I'm not sure why you need to match some mathematical cuvre to the degree of accuracy you think you might be able to achieve here. You'd probably have to do your own calculations for the lengths of the transition curves to use in your drawing. These could be based on a "known" speed limits, superelevation and curve radius. Without track diagrams some or all would of this be based on estimates or maybe just "guesstimates" of what you see in the images.


Nahhh ... That just aint' gonna happen to easily get a program to lay Trainz track for you ... Get used to laying track the good ol' fashioned, taking 6 years of your life, way !
I guess easy is in the eyes/mind of the user but I've been doing this - creating track files outside surveyor - for many years.

Bob Pearson
 
Last edited:
Thanks for the detailed reply. Ill have to respectfully disagree on the ability to match turns to the terrain over ortho imagery. With TransDEM, the only issue is the inability to create curves of specific radii. Straights get wobble added from operator inability to track a straight line. I made no comment about the appropriateness of TransDEMs pricing, and I dont have any comment to make about it being higher priced if more useful tools were added to it. Certainly I concede that few users are interested in such a tool, in any event.

But then, few users are interested in rigorously realistic track geometry, too. Im guessing I might be better off looking at a train simulator rather than a train game.

Thanks again.
 
Thanks for the detailed reply. Ill have to respectfully disagree on the ability to match turns to the terrain over ortho imagery.
I'm not saying you can't match curves to the terrain I'm just saying especially if it includes transition curves you'll have trouble picking start and end points and therefore an accurate curve radius without some other calcs or construction work on the image. I had this trouble tracing over imagery in google earth pro.

With TransDEM, the only issue is the inability to create curves of specific radii. Straights get wobble added from operator inability to track a straight line.
Wobbles are an inherent problem with the Trainz track spline. If you're concerned about them I think you'll need more than just a circle/arc tool in TransDEM. There are many other places apart from "off line" vertices along a long stretch of tangent track. I started another reply where I went deeper into this and ended up with a long discussion of the spline Auran picked and N3V still uses for the track. If I get some time I might finish it up and post here.

I made no comment about the appropriateness of TransDEMs pricing, and I dont have any comment to make about it being higher priced if more useful tools were added to it. Certainly I concede that few users are interested in such a tool, in any event.

No problems here. I fully agree. I was just pointing out it would add to the cost. This is a concern for many trainzers who consider using TransDEM. And it would be difficult for many users to understand "why", since as you point out, there are very few of us in the Trainz community that think any of this "lunacy" is worth talking about let alone taking the extra time effort to incorporate in our routes.

But then, few users are interested in rigorously realistic track geometry, ...
Thanks again.

Well I count myself as one of these few and have been since Trainz 1.1 or so.

Bob Pearson
 
Last edited:
~Certainly I concede that few users are interested in such a tool, in any event.
Thanks again.
I'd be willing to bet that a majority of Transdem users who paid for the program, never master it, and never actually make a successful DEM, and discard the program, amongst with the pile of other programs they never mastered
 
Fortunately for us, enough people have bought it, taken the time to learn how to use it and built and continue to build great routes for Trainz with it.

Bob Pearson
 
I measured the metric heights in Google Earth, and applied those GE metric heights into Trainz on a Transdem DEM ... Nope doesn't work ... The GE heights are way off by + 5m

I used to meticulously guestimate, and apply guess work gradients ... which was VERY time consuming, and was mostly WRONG

I lay my MB Quad tracks on the straightaways, then connect curves, then add easements, then measure the terrain heights in the topology tab, and use those heights and apply them to the track spline point heights ... it sometimes is a little rough, and inaccurate, and needs fine tuning to the heights/gradients ... as one grade can be -0.05%, and the next gradient +1.95% ... but at least this method gets you close to in the ballpark, and stops rollercoaster floating track ... I have @ 30 some miles to get to Pitcairn Pa ... and another 30 some miles to get to Conway ... then I will work backwards from Mount Union to Harrisburg/Enola

I swear my route has half a million baseboards, and is 5 h 29 min @ 250 miles ... Then my life will be complete

It runs on my ancient integrated graphics Toshiba Satellite 105 laptop ... just blank gray baseboards, and track
 
Last edited:
I measured the metric heights in Google Earth, and applied those GE metric heights into Trainz on a Transdem DEM ... Nope doesn't work ... The GE heights are way off by + 5m

I used to meticulously guestimate, and apply guess work gradients ... which was VERY time consuming, and was mostly WRONG

I lay my MB Quad tracks on the straightaways, then connect curves, then add easements, then measure the terrain heights in the topology tab, and use those heights and apply them to the track spline point heights ... it sometimes is a little rough, and inaccurate, and needs fine tuning to the heights/gradients ... as one grade can be -0.05%, and the next gradient +1.95% ... but at least this method gets you close to in the ballpark, and stops rollercoaster floating track ... I have @ 30 some miles to get to Pitcairn Pa ... and another 30 some miles to get to Conway ... then I will work backwards from Mount Union to Harrisburg/Enola

I swear my route has half a million baseboards, and is 5 h 29 min @ 250 miles ... Then my life will be complete

It runs on my ancient integrated graphics Toshiba Satellite 105 laptop ... just blank gray baseboards, and track

The height disparity is mostly due to us being stuck with fixed grid. We have a choice of 5m or 10m and no steps in between. Mother Nature has analog controls over her heights and set them to whatever she wants. This disparity is also seen with DEMs imported with TransDEM for the same reason, however, knowing there's that disparity there means we work within it and that +5m makes no difference since everything is relative to its base.

It's like a keyboard instrument being tuned to A @ 415 versus today's concert pitch which is A @ 440. The A above middle-c sets the pitch of the whole instrument. Sure the instrument is a semi-tone lower, but since everything is also, the instrument remains in tune to its self. A @ 415, by the way, is Baroque pitch by today's standards. There was also A @470 and A @ 392 or Versailles pitch which is a one full step lower!
 
Basically, it's all down to the DEM data source. Particularly, USGS NED 1/9 arc sec (where available) will perform quite well. Cuttings and embankments/high fills all nicely shaped. But one should use the 5m Trainz terrain grid for this. Terrain will only need minor levelling for the right-of-way.

Sent from Cape Town
 
anyone know how to merge DEMs in the latest version of microdem?

Make sure all your DEM sections are converted to TransDEM dem files first.

Once you do this, you open up your first DEM file then click on the file then add DEM. Repeat this until all your DEMs are placed.
 
is 1m lidar dem acceptable? my export seems kind of flat (although i am trying to build a city route)

I've used UK 2m Lidar, it works well with Transdem. Worth noting that you will only get 5m resolution on the actual terrain as that's the smallest size TANE can cope with, it's still very much more accurate than any other dem.
 
It would seem that whoever drew the lines on the DEM, you are trusting in their accuracy. If a line goes around a hill that is 20m wide, the center, or outermost edge of the line, away from the crest of the hill, is the actual grade plateau height to base your grade. This will mean that after you grade your track, a cut, will cut into the hillside, making a cliff. I have tried everything ... and still a DEM grade is totally wOOping me :eek: I have found that most 2.75% grades are in total error, and by using the outermost edge of the tracing line, gives you a 0.10% to 0.50% gradient ... rarely does a RR grade exceed 1.00%

It would seem that Google Earth heights are @ 5m higher, different than a DEM height, at an given point
 
Last edited:
I believe it is 1/9th arc data, but if you get down and measure any area just inch's away the heights are a real bumpy rollercoaster ... I have learned to work around it and just lay the track, and hit the smooth spline tool
 
I believe it is 1/9th arc data, but if you get down and measure any area just inch's away the heights are a real bumpy rollercoaster ... I have learned to work around it and just lay the track, and hit the smooth spline tool
This is an example for 1/9 arc sec in TransDEM, with DEM shading on:
.
Could you post a similar screenshot?
 
Never is it just plop down the track ... it is always -2.37% then up +3.75% ... even when I lay dozens of small track splines on a single straight away, it is never flat, and is chock full of leaps and dips

I took Transdem off my machine as my HD disc space was filling up ... I just work in surveyor now, and a 5 mile long branchline with a yard, takes days, and days, of questimating, and fiddling with track spline point heights

I only float the track a small portion of a meter, till it is good enough ... then hit the smooth spline tool
 
Never is it just plop down the track ... it is always -2.37% then up +3.75% ... even when I lay dozens of small track splines on a single straight away, it is never flat, and is chock full of leaps and dips
But when you look at my example and the various railroad right-of-ways shown, the cuttings and fills, you will probably admit that this looks pretty accurate and nothing like a roller-coaster. Therefore I presume that your DEM data source has been of a much lower resolution. I suggest you download a high resolution DEM and recreate the terrain for the baseboards in question. I'm sure you will get similar results to all the other people who successfully work with hi-res DEMs.
 
Not even one little bit of that, is ever, ever, going to happen ... I have a 1.09GB DEM route, from Harrisburg/Enola to Conway, written in stone ... I am NOT ... Going to start over again ... What ever resolution it is ... Is staying, as is ... I believe the DEM is a high quality resolution ... I'll just have to work with it

My mountains are not rounded
 
Last edited:
In case you change your mind at a later stage, that's what a hi-res DEM will produce (click on thumbnails):

Bare-bones terrain, 5 m baseboard grid:


With topo map ground textures (1:25k map):


With a 3D UTM tile (10m) as texture carrier for a large scale map (1:2500):
 
Back
Top