Map projections?

[davebridges]

Will it be possible to select different map projections when building a new route in ORTS? I really hope that there will be some alternative to the one that MSTS uses. Here in the Midwest, most roads and towns are laid out on a north-south/east-west grid pattern, and it isn't possible to build a true grid and follow the GPS coordinates at the same time. It's especially bad if you're building a north-south route, like CSX's CE&D Sub, or Conrail's Dow Secondary/Marion Branch. If you follow the GPS, all of your towns will be skewed and stretched into a weird parallelogram shape.

[longiron]

yes, we are investigating a number of tiling strategies that will minimize or eliminate distortions when mapping geospatial information into a new OR route.

[lindsayts]

Some thoughts on this issue.................

Now this is a complex issue with some REAL good mathematics amongst the solutions and one could discuss this for a good while..........

The problem in this is the shape of the earth is not regular. The earth being a slightly flattened sphere with major diameter (the equator) being of an irregular radius. The only way to model it without distortion is to model its correct shape. I did some work on this but even a simplified routine was way slower than treating the earth as flat and it still produced an error of around 0.5% in distance, which I regarded as to high.

All projections produce some kind of distortion as it is not possible to flatten out a spherical surface without distortion. For the sim I am working on I am using "transverse Meractor projection" for the earth model. With this one can create a route on any distance north south but are restricted east west to between plus or minus 300 to 400 kilometres either side of the origin latitude. Much testing on this model showed it had an error in distance of around 0.1 % in an east west direction at around 400 kilometres east west. So one could produce an East west route of around 800 kilmetres with a distance error of +- 1 kilometre.

Co ordinate conversion is an allied issue, there being some quite complex maths to convert "geocentric' (latitude and longditude) to "Transverse Mercator" but I did manage to come up with a routine with the required accuracy. After producing that I came across a library of functions that did the trick with great accuracy.

So far my prelimary Route editor for my sim is working well with this solution (Transverse Mercator).

On the whole a problem with some good solutions avaiible, OR should have little problem.

Lindsay

[wacampbell]

Some thoughts on this issue.................

Now this is a complex issue with some REAL good mathematics amongst the solutions and one could discuss this for a good while..........

The problem in this is the shape of the earth is not regular. The earth being a slightly flattened sphere with major diameter (the equator) being of an irregular radius. The only way to model it without distortion is to model its correct shape. I did some work on this but even a simplified routine was way slower than treating the earth as flat and it still produced an error of around 0.5% in distance, which I regarded as to high.

All projections produce some kind of distortion as it is not possible to flatten out a spherical surface without distortion. For the sim I am working on I am using "transverse Meractor projection" for the earth model. With this one can create a route on any distance north south but are restricted east west to between plus or minus 300 to 400 kilometres either side of the origin latitude. Much testing on this model showed it had an error in distance of around 0.1 % in an east west direction at around 400 kilometres east west. So one could produce an East west route of around 800 kilmetres with a distance error of +- 1 kilometre.

Co ordinate conversion is an allied issue, there being some quite complex maths to convert "geocentric' (latitude and longditude) to "Transverse Mercator" but I did manage to come up with a routine with the required accuracy. After producing that I came across a library of functions that did the trick with great accuracy.

So far my prelimary Route editor for my sim is working well with this solution (Transverse Mercator).

On the whole a problem with some good solutions avaiible, OR should have little problem.

Lindsay

... or, we could just make the earth round in Open Rails. All this is a problem for map makers because maps are flat, but OR's world doesn't have to be.

[lindsayts]

... or, we could just make the earth round in Open Rails. All this is a problem for map makers because maps are flat, but OR's world doesn't have to be.

Depends on how accurate you require it to be. As I said the Earth is NOT a sphere and using an earth model with a constant radius means the distances will ONLY be correct in a small number of places. I did an earth model as a sphere the radius being the average radius of the earth. When this was used to model a line in Southern Australia produced an error in distance of around 0.7%, which I classed as to much. Also the routine to display terrain of this kind of model was way more complex than using the flat earth model so I turned to a mapping projection. This produced an accurate model (with some limitations, one being the max east west distance) with all the difficult calculations being in the terrain generation leaving the display routine very simple (and therefore fast).

[Railfan727]

Map projections?

Great to see you're already working on a route editor for OR! Will it include a utility that will automatically import elevation data and/or satellite photos and create markers, or will we need to continue to use separate programs such as USAPhotoMaps?

It would be really nice if there were a way to project satellite photos directly onto the terrain, if for no other purpose than just to get everything aligned properly while building... then perhaps change to a final permanent terrain texture once everything was ready. Any thoughts?

[twpol]

Will it include a utility that will automatically import elevation data and/or satellite photos and create markers, or will we need to continue to use separate programs such as USAPhotoMaps?

We're considering all features at the moment; obviously, with limited development time, not everything will be possible when we'd like, but reducing the number of tools route builders need to use is certainly an important area.