[BillC]

Actually this has little todo with TSRE5, and more about Goode-Homosoline. When viewing the thread TSRE5 newbuild problems, the OP (NickonWheels) asked about the problem with GH skewing, and ending with user FrannDzs discussing obtaining DEM data for an Argentina route. Between the OP post, and the ending post. There have been replies ranging about obtaining DEM data, TSRE5, and map projections. Below are my ramblings about the subjects.

In reply #4 Genma Saotome (Dave Nelson) commented about the GH visual effects. Dave has commented for many years about this and in the forum “Open Rails Development (Private)” draw a diagram illustrating the problem. A few days ago I asked Dave if the thread The Goode-Homosoline Projection, could be made available for public viewing. In a PM he said that if the ORMT would agree the move would be made. Subsequently the thread has been moved to “OR - Developing Features”. Thanks for doing this Dave.


The above thread has in addition to Dave’s diagram, has a discussion about using a better projection. The GH is projection is an equal-area map projection, actually two equal-area projections rolled into one. From Wikipedia Goode homolosine projection is called a “pseudocylindrical, equal-area, composite map projection”. How ironic the use of the word pseudo. What we need as discussed in the above thread about is a conformal projection, which for a world view (small scale map) the lines of latitude and longitude orthogonal to each other. Allowing roads, buildings, etc. not to be skewed, the familiar Mercator projection. To be able a degree of accuracy on a (large scale map) there is a need for UTM.

For now I will end my comments. Latter I will comment of Goku’s words regarding TSRE5 and map projections. In addition I will discusses when having coordinates from different projections why you need to align them to the same projection (georeferencing)

[NickonWheels]

Pseudocylindrical? Should leave lots of space for imagination. :dance3:

One other thing I just want to mention about projections is that the map TSRE5 uses as default looks like OpenStreetMap, but there is actually OpenRailwaysMap where you also can see many historic lines which may be of interest to model for someone.

[Genma Saotome]

Map projections have been discussed for many years to no firm conclusion besides the G-H method sux.

Whatever consensus might occur at some future date do know that any implementation of even a perfect solution will require someone to volunteer their time to implement it.

That could have happened at any time in the past ten years but it has not, perhaps in large part because no one in the OR team had the combination os skill, time, and personal interest in route building. So both a realistic solution and someone able to take the time to code it are needed.

A long time ago DEMEX had a feature to not use G-H. AFAIK I may be the only person who has used it (Goose Island rte in Chicago) and I did so because Chicago's street grid is very rectangular and the route was scoped to a couple of square miles.

I still had problems... no markers really matched up with the few actual terrain features and so in the end I really had to just wing it.

My point here is a good solution will require more than creating terrain... is has to include an accurate reference to available "location-ing" tools, whether that is some marker function or aerial photos. Is that Google Maps? Open street map? Something else? It might turn out to be as hard to figure out as finding the better method to project terrain.

[PerryPlatypus]

Genma Saotome, on 11 August 2022 - 08:54 AM, said:

A long time ago DEMEX had a feature to not use G-H. AFAIK I may be the only person who has used it (Goose Island rte in Chicago) and I did so because Chicago's street grid is very rectangular and the route was scoped to a couple of square miles.

I have used this feature as well, but on two projects that so far have remained just personal projects that didn't go very far. One of those was the Alaska Railroad project I started. Compared to where I am used to making routes down in the lower 48, the G-H skew in Alaska was much worse, to the point that it was gonna cause some very awkward non-uniform horseshoe curves in the particularly loopy portion of the ARR in the mountains south of Portage, AK. DEMEX successfully un-skewed the terrain, but in the process the aerial imagery that TSRE pulls in from Google became completely useless, which as your alluding is a *big* drawback. The only silver lining is that DEMEX was also able to un-skew my marker file to land in the correct place...

The only other project I used DEMEX's un-skewing on on was a route just for my own testing purposes - the Ohio Match Lumber Co's "Burnt Cabin Railroad" in the mountains northeast of Coeur d' Alene, Idaho. Again, the fact that TSRE's aerial imagery tools no longer functioned properly was a big deterrence.


As a result of the TSRE issues, on subsequent route projects I have opted against using DEMEX's "un-skewing" feature, and I highly agree that a fix to the projection issue *must* be done hand-in-hand with the background aerial/maps tools. A huge can of worms, unfortunately...

[dajones]

I'm using Transverse Mercator for the route I'm building. I was unable to get TSRE to do anything useful on my previous Linux machine, so I gave up on it and decided to try building a route using home made tools. I use proj4 to do the coordinate mapping. I use C and javascript versions of proj4, but I think there are c# versions also. I put a projection.json file in the route directory to document the project used by a route. Here is a sample:
{
"proj4Str": "+proj=tmerc +lat_0=44.67912590032705 +lon_0=-72.57981532105902 +k_0=0.9999878417038954 +x_0=0 +y_0=0",
"centerTX": -10957,
"centerTZ": 14572,
}
The proj4Str value is a string used to initialize proj4. CenterTX and centerTZ are the MSTS tile in the center of the route. I selected these to match the lat_0 and lon_0 using the usual G-H project.

My DEMEX like program uses osgEarth to process elevation data files. That code is a bit of a mess and would be difficult to share.

My main editing program draws the route on internet maps adjusted to match the route's projection. That code is written in javascript. I've never tried running it on Windows, but it might work. But most of the features are very different from the usual route editors.

Does anyone know what projection the alternate DEMEX feature uses? I might be able to modify my code to match it.

Doug

[Weter]

@NickonWheels
Thank You for important note about Historical lines, shown on map!

[BillC]

 Demex.htm (59.99K)
Number of downloads: 4

dajones, on 11 August 2022 - 12:06 PM, said:

I'm using Transverse Mercator for the route I'm building. I was unable to get TSRE to do anything useful on my previous Linux machine, so I gave up on it and decided to try building a route using home made tools. I use proj4 to do the coordinate mapping. I use C and javascript versions of proj4, but I think there are c# versions also. I put a projection.json file in the route directory to document the project used by a route. Here is a sample:
{
"proj4Str": "+proj=tmerc +lat_0=44.67912590032705 +lon_0=-72.57981532105902 +k_0=0.9999878417038954 +x_0=0 +y_0=0",
"centerTX": -10957,
"centerTZ": 14572,
}
The proj4Str value is a string used to initialize proj4. CenterTX and centerTZ are the MSTS tile in the center of the route. I selected these to match the lat_0 and lon_0 using the usual G-H project.

My DEMEX like program uses osgEarth to process elevation data files. That code is a bit of a mess and would be difficult to share.

My main editing program draws the route on internet maps adjusted to match the route's projection. That code is written in javascript. I've never tried running it on Windows, but it might work. But most of the features are very different from the usual route editors.

Does anyone know what projection the alternate DEMEX feature uses? I might be able to modify my code to match it.

Doug

Doug,
I added a Demex.htm file which is Demex user manual of sorts. which has some discussion about UTM, and datum shifting.
I will PM you in the next couple of days about my thoughts for some code development, and copy Dave. In a trditional developer environment Dave would be the customer. He already has used 1/9 data with Qgis, and spoon feed it into Demex. I have CLI code both in C, and C# to covert spherical coordenates to IGH and the inverse. My problem is that I don't have a good feel of file structure of MSTS terrain files. With your knowldege of Open Scene Graph, or with other OR developers perhaps we can cobble something together.
Bill

[Goku]

dajones, on 11 August 2022 - 12:06 PM, said:

I'm using Transverse Mercator for the route I'm building. I was unable to get TSRE to do anything useful on my previous Linux machine, so I gave up on it and decided to try building a route using home made tools. I use proj4 to do the coordinate mapping. I use C and javascript versions of proj4, but I think there are c# versions also. I put a projection.json file in the route directory to document the project used by a route. Here is a sample:
{
"proj4Str": "+proj=tmerc +lat_0=44.67912590032705 +lon_0=-72.57981532105902 +k_0=0.9999878417038954 +x_0=0 +y_0=0",
"centerTX": -10957,
"centerTZ": 14572,
}
The proj4Str value is a string used to initialize proj4. CenterTX and centerTZ are the MSTS tile in the center of the route. I selected these to match the lat_0 and lon_0 using the usual G-H project.

TSRE has the same function build in. You can define it in route TRK file and it works with terrain generation, maps, satellite images etc. There are how to somewhere on the forums.

[eric from trainsim]

Glad to see you back, Goku.