Checking Tile Resolution

[landnrailroader]

Friends,

While we have determined that we can use the higher resolution 10M DEMEX files at least as long as we have computers old enough to support MSTS & DEMEX, this brings up another question. Is there a way to tell what the terrain resolution is for a completed route, completed at least as far having terrain generated? I want to go back regen the terrain on routes that I know were done with 1 arc-sec data such as my Kentucky Div. which also needs to have DM tiles created and I have a pretty good list of others to fix as long as the old XP machine tolerates such abuse but if in doubt, is there a way to determine the resolution. We know that the terrain mesh is at 8m from comments that were made, so I guess 1/3 arc-sec terrain would have much more dense wire frame views than would 1 arc-sec data?

Jerry

[wacampbell]

It doesn't matter what resolution was used to DEM the route; its all converted to 8 meter in MSTS/OR. So all you will see in your old routes is an 8 meter terrain grid. 1/3 arc-sec and 1 arc-sec will look the same resolution once its imported into MSTS/OR. Although admittedly the routes DEM'd with the 1/3 arc-sec data will have more accurate elevations on its 8 meter grid. But to go back to your question, I don't think there is a way to tell if an old route was DEM'd with 1/3 arc-sec or 1 arc-sec DEM.

[scottb613]

So all you will see in your old routes is an 8 meter terrain grid. 1/3 arc-sec and 1 arc-sec will look the same resolution once its imported into MSTS/OR.

Hi Wayne,

Hah - I thought I had a grasp of what's going on - [1/3 arcsec is 10m] - and [1 arcsec is 30m] - - -[ORTS is 8m]. Since the mesh in the sim is 8m - wouldn't the 10m source impart far more detail to ORTS mesh than the 30m?

Regards,
Scott

[landnrailroader]

Scott,

That is what I wondered. In any case, I only have a few routes that need to be redone and it doesn't affect any object work that has been done. A few will also need to have DM generated. I am going to download all the DEM modules I will need and then move them over to the old PC. Routes were all created there, so if I have not done work on one yet, I can just update what is there and save it, otherwise porting by way of thumb drives is SLOW.

Well, I just looked at the terrain grid of the Kentucky Div. (higher resolution data) and the Camas Prairie (low resolution data) and there is no visual difference. So elevation values seem to be the most obvious change. For example, the flat ridge to a correct ridge I mentioned earlier and then another thing I noticed is that railroad or road fills are much more pronounced with the higher res. data. Now they may or may not fall exactly centered on our digitized line because there is variance between differently dated aerials in Google Earth, so I just choose the date that has the best image and stay with it. The lateral variation is usually not more than a couple of meters and the only other issue is if the different images are not smoothly joined together but in those cases I do the best I can using track chart data at that point and live with it.

Jerry

[wacampbell]

Hi Wayne,

Hah - I thought I had a grasp of what's going on - [1/3 arcsec is 10m] - and [1 arcsec is 30m] - - -[ORTS is 8m]. Since the mesh in the sim is 8m - wouldn't the 10m source impart far more detail to ORTS mesh than the 30m?

Regards,
Scott

When you import 30 meter DEM into the MSTS 8 meter grid, the importer ( DEMEX or whatever ) has to interpolate the data. eg on a certain MSTS grid point it finds the nearest DEM elevations and figures out a weighted average to guess what elevation would be at that point on the 8 meter grid. With 10 meter DEM, this interpolation is more accurate, because the nearest DEM points will be closer to where you need. So a route DEM'd with 10 meter data will have more accurate elevations than one DEM'd with 30 meter data, but either way the elevations will be on a 8 meter grid. I guess you could say that is more detail, but not more resolution. BTW, remember that DEM data's vertical accuracy is +/- 1 meter, so terrain shape won't ever be right on.

Wayne

[Coonskin]

Maybe I was seeing things, but after learning (at this forum) there was supposed to be a difference between 10m and 30m, I was convinced I could see a difference between 10m and 30m DEM when the terrain was quite rugged.

I mean, IF I'm understanding correctly, the difference between 10m and 30m is that with 10m, the elevations are sampled every 10 meters (32.8') and 30m was sampled every 98.4'. That's quite a discrepancy. I think I recall way back when I learning route creation, I was creating a route in a rugged portion of Colorado using the less meg-hungry 30m SDTS DEM.

Upon reading there was a difference, I went back and downloaded 10m DEM. (Anyone remember MapMart?) The relief detail in the rugged portions (more vertical w/crevices/etc) was more apparent via the shadow feature from the lighting in MSTS. I concluded at the time that 30m DEM resulted in terrain that was noticeably "smoothed". That's fine for modeling gentle flat lands, but wasn't so fine when modeling rugged terrain. Also, over the decades, I "think" I have noticed that with 10m I end up with terrain that shows traces of cut/fill renders that are not present with 30m? In fact, I think I Of course, I did not chronicle.

However, what do I know? (I'm not tech savvy.) I've merely been a putz route creator for decades!

WAIT! I've had a brain awakening moment!

WAY back in the 2000s, I was hoping to create a commercial 1893 era Colorado Midland upon which I could build the first division and get that to market, then set about building the next 100 or so miles, and selling that as an add-on, etc.

I originally used 30m SDTS DEM. Upon learning about 10 NED Seamless terrain, I reformed my existing work terrain with 10m NED Seamless. There WAS a difference, in fact I took pics of it and posted my findings here at Trainsim back then. Digging through my old route archives I found the old route AND the comparison pictures. Check this out:

Terrain created with 30m SDTS DEM:

AE_10mNed_cf_30mSDTSc.jpg

The SAME terrain created with 10m NED Seamless:

AE_10mNed_cf_30mSDTSa.jpg

And here's a pic after I realigned the previous laid track to match the rendered road bed and tunnels/etc that were a result from switching to 10m NED Seamless. Such detail was NOT there using 30m DEM.

RE_UtePass5.jpg

In view of the above, from some 2 decades ago, I'm again convinced there IS a difference in detail and accuracy of 10m vs 30m.

Andre

[eric]

Is math the only reason we are stuck with an 8m grid? Why not a 5m grid so that 30m and 8m have a common denominator?...

If a route definition (i.e. TRK file) or Terrain Definitinon (TD) could specify the terrain resolution, ORTS should be able to adjust accordingly when rendering the tiles. And yes, TSRE would need to handle it as well, which is problematic since nobody has stepped up to help figure out the math involved there...

[scottb613]

Hi Folks,

All good - thanks for the explanations - I have it now.

We still sorely need better tools to sculpt terrain - as it seems impossible to recreate effects we see in DEM created mesh - with our current sculpting tools.

Yeah - this is what I noticed - this limestone quarry was basically nonexistent when I used 30m DEM - - - I love the detail that comes out at 10m. When painted - this should look pretty good.

Image1.jpg

Regards,
Scott

[landnrailroader]

Yes, I have noticed that cuts & fills render pretty close to required routes when using 10M DEM. I don't adjust the track to the terrain, though, I go by the digitized path. In any case, I wish we had a better way of doing cuts & fills. That was one area that MSTS is far superior to TSRE in doing and the parameters make more sense too. But I dream on.

Jerry