TUTORIAL Race Track Builder - Inject elevation data and terrain texture

[Les Neilson]

... and the localhost page shows

 

[Wurstkoffer]

I see. Everything after the "?" is not necessary. I doubt it'll fix the problem, but try it. And send me some coordinates where you get the -999999 values.

 

[Les Neilson]

Okay,

55.6186272568359, -5.4533306776263

55.7746347714844, -4.58570556276981

 

[Wurstkoffer]

There we go. The scotland WMS indeed sends a -9999.0 value. That's not nice.

Do the following. In config.php around line 66 add this:

// There are WMS servers out there which sends extreme fake values
// for areas outside its coverage. To compensate that, specify
// a range in which the expected real value should lie.
$wms_elevation_value_limit_min = "-300";
$wms_elevation_value_limit_max = "9000";

In maps\api\elevation\json\index.php change line 184 from this:

else {

to this:

elseif($elevation_wms > $wms_elevation_value_limit_min && $elevation_wms < $wms_elevation_value_limit_max) {

 

[Les Neilson]

Okay, let's try that... I'm actually finding this PHP stuff interesting, by the way. I've downloaded a couple of pdfs. Now to find time to read them...

 

[Wurstkoffer]

Yeah, relatively easy to learn programming language for web stuff. There are a really good documentation at http://php.net/docs.php. A good editor (like the free Komodo Edit) helps with code completion. I think there are also addons for Notepad++ which upgrades code completion.

 

[Les Neilson]

Last night it was working well at last, although slowly. Unfortunately RTB crashed so I lost it all, but I'll try again this evening. If all goes well I'll build a short stretch to see how it looks and runs. I'll post a link if anyone is interested. Even subdivided to around a metre or so I wasn't seeing any stepping and the road bed and edges were pretty well defined. So far, very positive!

 

[Willy Wale]

Thanks again, it seems to be working now... On to the next problem I'm getting a lot of negative numbers outside of the sampled area (-9999) Where in the code can I limit the elevations to zero? (or only slightly negative)

UK LIDAR data I have looked at contains -9999 for any data point that is a body of water (lake/river/sea). It may also be the generic value for no data. So I think the 'correct' values are being returned (ie same as the source data) and as you wrote above you need to post process somehow.

 

[Wurstkoffer]

Last night it was working well at last, although slowly. Unfortunately RTB crashed so I lost it all, but I'll try again this evening. If all goes well I'll build a short stretch to see how it looks and runs. I'll post a link if anyone is interested. Even subdivided to around a metre or so I wasn't seeing any stepping and the road bed and edges were pretty well defined. So far, very positive!

Oh nice, no steppings. Maybe they use some interpolation in these DTM/DSM layers. The layer i used is pixel based with a resolution of something like 1 pixel = 2x2 meter.

RTB crashes might be related to RAM? But I'm not sure. I have 32GB and rarely crashes. In some sessions I may have been subdividing for 2-3 hours without interruption and RTB has retrieved elevation data from the WMS during that time. At some point I went to sleep and left RTB to itself. The next morning I saw that it was receiving elevation data for 4-5 hours after i left. And there was no crash.

I'm really interested in your results, so of course send links!

 

[Les Neilson]

The lack of stepping might just be that I haven't subdivided enough to see it yet. I'll let it finish and try it out as is. Then I might try to see how far I can subdivide... In the meantime, another couple of hours and I should have something to share

 

[Les Neilson]

Okay, here's the test track

https://www.mediafire.com/file/ls36cz79zz43kez/testlidar.zip

 

[Wurstkoffer]

Nice, that was fast. Terrain looks really detailed. Just needs some flattening. And you already placed some trees, buildings and objects.

 

[Les Neilson]

Yes, fast because I only subdivided close to the road. But no visible stepping, and I was able to conform the track to the surface (well, 25% in the centre and 95% at the edges) and it seemed pretty good. I haven't driven that bit of road for maybe 20 years but it looked pretty realistic

 

[Les Neilson]

It's a shame that there's no freely available lidar data for my pet project (the Isle of Man TT circuit). It's on sale though, for £7000

 

[Wurstkoffer]

damn

 

[Les Neilson]

Another question: what happens if you save a piece of terrain and reload it before all the subdivision has been completed? Are the unsampled points eventually fixed or do they get left as they were when you quit?

 

[Wurstkoffer]

I'm not sure, if RTB is saving the requests and resumes after restart. I don't think so.

 

[Les Neilson]

It doesn't appear to, but that doesn't surprise me. Although I notice this: when I reloaded a track and tried to add some more detail to a section, it was okay at the beginning, then I started to get only -9999 again. So I quit out, and realised I'd hit the edge of the sample area. When I restarted RTB again it went on sending the requests that didn't get a valid response before

 

[Les Neilson]

What would be really useful: in my case there are two layers covering different areas. If the php could try one layer - if it gets a null response try the next - and if nothing works default to google data.

 

[Wurstkoffer]

Oh, good to hear it resumes. Multiple layers for different areas shouldn't be a problem to code. Send me the URLs and some locations and i'll take a look at it.