[bbb]

Hello everyone!

For about a month now, I've been using my free time to dive back into TSRE and start working on a new route project.
But as I progressed with the modeling, I stumbled upon something both surprising and a bit frustrating.

As a regular user of TSRE’s geo-data features, I mainly rely on:

- HGT files for terrain relief

- Satellite imagery (Google Maps Z17/Z18) and OSM data as overlays to accurately recreate the real-life track layout

On paper, everything seems perfect. But something just didn’t add up…
While working on the platforms of some stations on my route — some of which, in real life, can accommodate up to 15 to 17 Corail coaches — I noticed that in-game, even when precisely following the satellite overlays, my platforms were too short. They could only fit 14, maybe 15 coaches at most.

Why the discrepancy? After several checks, a thought came back to me: I had already noticed during synchronized tests with cab-ride videos (at matching speed) a growing offset between the real footage and the TSRE/OR simulation, particularly:

In a small yard north of Toulouse-Matabiau
And during a simulation in the Lot region, where the mismatch becomes noticeable over time

And that’s when it clicked.

I directly compared — using a ruler — the platform lengths in TSRE and in Google Earth… The result? Differences of 50 to 60 meters! Digging further, I compared the width of a rail in the satellite view to that of a 3D TSRE rail: the rails and trains in TSRE seem oversized relative to the satellite overlays.

 capture2-min.jpg (173.21K)
Number of downloads: 3 Measurement of the platform length in TSRE,
 capture.jpg (192.79K)
Number of downloads: 3Measurement of the platform length in Google Earth, making sure to measure exactly the same distance

➡️As you can see from the comparison, there's a 40-meter discrepancy — which is really substantial.


That’s when I realized: after laying down over 350 km of track, I may have been working from a map whose scale doesn’t fully match the 3D world of TSRE. And when a rail appears perfectly aligned on an OSM map, but its 3D rendering is noticeably wider… it becomes clear that everything can be off: lengths, platforms, timings, distances.

So now I’m turning to you:

🟡 Have you also noticed this scale issue between satellite/OSM maps and the TSRE 3D environment?

🟡 Has it affected your route-building work?

🟡 Have you found any methods to fix or work around this problem?

🟡 And more importantly: what real impact could this have on a simulation spanning several hundred kilometers or a 6-hour journey? Is it possible to estimate the time or distance offset?

Lastly, do you think HGT files (used for terrain elevation) might be influencing this scale issue, or be somehow related to it? Have you ever noticed a correlation?

➡️ One puzzling observation: when comparing "game length" and "geo length", the latter seems to better reflect real-world measurements… but strangely, it’s never fully visible or reached in-game or in TSRE.

Thanks for reading! I believe this is a valuable topic for anyone using satellite/OSM data in their TSRE creations. I’m looking forward to your insights — hopefully we can better understand, or even solve, this scaling issue.

See you soon,
bbb

[Weter]

Hello.
First, such a great issue-report. Thanks for efforts to write it so well!
Second, Do You know about different geo-projections? (what does it mean)
Definitely, a satellite with camera doesn't hanging exactly above the spot, we are interested in; Lens of camera is spheric, leading to distortion, which is greater towards frame's peripheria. In map making, film is translated in special ways, according to complex calculations and accurate data about aircraft's or satellite's attitude at the moment, the photo was taken. General-purpose representation on services like google Earth, likely doesn't use "transformed" images from cartographic manufacturers; instead - just stretching images, they have (taken from different satellites, in different times from different height, angle, inclination and focus) to match their edges (not always well). Note also, HGT data has shift in sides and height (6-8m).

[Weter]

TSRE provides us with ability to compensate such distortions somehow. Not the perfect way, but better, than nothing.
First, author should decide, which and what kind of distortion should eventually satisfy his expectations.
Second - he should examine all his sources to note, which geo-projection is used on each one of them.
Third, it's needed to determine the route's center geographic co-ordinates and use that, telling TSRE: from which point to compensate.
TSRE will stretch map's image (and HGT) to match objects' positions better, but do note: map is always flat, so no way to put spheric surface of the real Earth on it, without any distortion. Did You notice diamond-shaped representations of rectangular in reality structures? So will be ORTS route, built in TSRE - compromiss should be determined first. Note also, HGT's height is being manually corrected separately.

Quite good tutorial is written by Darwin Smith here. Good Luck!

[Kapitaen13]

The projection is explained in this topic.
New TSRE Map projection


Jan

[Weter]

May I show You a picture of initially - built tracks (yet without correction used) and re-built (after correction applied)?
https://www.elvastower.com/forums/uploads/monthly_05_2023/post-11269-0-92216500-1685556550_thumb.jpg
Note up to 1.2 times(!) exaggeration (stretch) in longitudinal direction for initial attempt: 1 km was like 1.2 or more.
While real latitudinal lengths are actually longer, than in that first version. Corrected sizes are still not real, but better.
And note, that this map's span is 40km! (having so many to re-place) While Your route is 350...

[eric from trainsim]

Can you clarify if you used the default geo Projection, or if you changed the projection to match your route's center point?

Example:

TsreGeoProjection ( 48.8485 2.29143 -12340224 30290944 )

If you, you're simply experiencing skew in the projection that was introduced 25 years ago.

[bbb]

Hello everyone 🙂
And thank you so much for your detailed and insightful replies !

➡️ About the projection system:
I did read through the topic that was kindly shared with me, and while I think I got the general idea, I'm still not entirely sure I fully understand how the compensation system in TSRE works
If I got it right, it’s a built-in function in TSRE that can "stretch" or "adjust" maps and HGT data? (Still a bit fuzzy for me...)

🗺️ About map distortions:
It’s now clear to me that satellite images can introduce some distortions (due to lenses, angles, altitude, etc.).
But I’m wondering:
🟡 Are OSM maps also affected by this kind of distortion?
I don’t know exactly how OSM works internally, but I always thought it was based more on third-party numerical data and less on raw satellite imagery — which led me to believe it might be more accurate in terms of scale
Any insights on this?

➡️ About the track offset:
I was really surprised by the comparison image you posted — that visible shift to the left is quite striking
Beyond just visual alignment, I’m curious:
🟡 Did you notice any significant changes in actual route distances or travel times once the correction was applied?
That would be very useful to know, especially for long routes like mine (~350 km).

Thanks again for your help and shared knowledge 🙌

All the best,

[Weter]

Hello.
Let's try to sort it out...

Let's say, TSRE always "wraps" route's terrain to osm image, but for last official version (2021 or so), author has opportunity to specify correction coefficients, based on coordinates of the point, he prefers (usually, route terrain's center) to minimize distortions (by image stretch).

Earth's surface could be "projected" on cylinder's (for whole globe), or cone's surface (for hemispheres), which then will be "cut" and straightened to be placed on flat.
These are simplest cases of projections (there are many other, much more sophisticated) and it's obvious, how much they differ form reality, especially on maps, showing large pieces of surface.
On small "plans" (that's why they are so called), distortion is nearly minimal, as sphere of Globe is so large, while terrain piece represented is relatively small. But when many plans are put together to form a big map (as if they would be tiles), that appears problematic.

Any REAL map (on paper) has so-called topo-base (I don't know actual english term for this). These are proportions, distances, directions to astronomic, geographic and magnetic poles, many more. The key for proportions, distances and distortions is geo-projection, used. This is a law, how spheric surface with all it's natural and antropogenic landscapes will be represented on flat of map sheet. There are different projections for different purposes and different locations, having their own strong and weak sides, specific distortion's localizations and appearance - so for concrete purpose, specific projection type is preferable (to demonstrate minimal distortions for it's aim).
For example, large topographic maps, representing multiple of rectangle "plans" of small parts of Earth's surface comprise of squares, but in some places, there are triangle fragments, compensating true spheric nature of Earth's surface.
At the same time, MSTS (and ORTS) world comprises of uniform square "tiles", and even not 2000x2000, but 2048x2048 meters, because the system is based on binary calculations of computer data, where 2^11 is 2048.
Therefore, we have totally different coordinates and dimensions system, still using meters and geographic coordinates (globe section's degree measures) to determine, where objects and elements of the virtual world, we being build, are to be placed.
And, that never will match real world, neither geographic map, nor satellite image the perfect way. it's impossible.

I suspect, TSRE shows osm map, as distorted initially, but Author can specify, how to distort it better for his goal.
I have no idea, what topo-base osm actually uses, but as You can see, that almost doesn't matter for us in this case.

To Your bottom question: Don't You believe, that when train should "ride" 1200m instead 1000 per every kilometer of route in East-West direction, either a bit less - in North-South, the time, needed for that will differ? :)
I can see that by laying tracks: now (after correction) 2x500m straight sections, put together are closer to distance between markers of 1000m, than before. Switches of 3.5 degree match route's image on map overlay better (before, I couldn't lay turnouts in accurate way).
The route is still unfinished, so I didn't have some exact timing observations, but I expect the result to be more accurate.
I ran out of dynamic track sections limit currently, while re-placing my trackwork, so can't simply finish the route.

[BillC]

Google what is open street maps ESPG code, and you will get deatils on the projection. Which BTW is the same as Open Rail Way Map

[jtr1962]

Is it possible for TSRE to modify an existing route to the new projection? I would guess it's not difficult to recalculate new positions for objects on the tiles. The real problem is the track, and maybe roads. You could recalculate the .tdb and .rdb to reposition the tracks/roads where they belong. Obviously once you do that you'll need all procedural track (except for switches) as standard track sections will no longer fit the new .tdb. Maybe you can leave the roads as is given that they're just relatively short segments near the tracks. It would be nice to have closer to real world distances in existing routes. In PRR East NYC to Washington is short by nearly 21 miles compared to the real world distance, for example.

[Kapitaen13]

jtr1962, on 17 May 2025 - 11:53 AM, said:

Is it possible for TSRE to modify an existing route to the new projection?

No, unfortunately this is not possible. The TsreGeoProjection parameter in the .trk can only be used to start a new route.

Jan

[Weter]

Hmm...
I'd say, this is not a simple question. That depends...
Currently, I don't know, does it work, or no (but no, much likely), meaning, TSRE would re-compose *.w-files with corrected co-ordinates for objects positioning, doing all routine for author in automatic way... Result could be unpredictable, but this is just a first glance guess (I didn't think about this before). Tracks... I had to re-place them manually section by section in my case.