Height Maps from the APK

  • Thread starter Outspacer
  • 217 comments
  • 33,778 views
Just started practising elevation maps, learned at you can alter over 20m instant jumps/drops with TPE on Eifel, learning bit more to understand logic fully, posting more later.
In exemple this track has unwanted drops, few minor changes and those are gone.

That is starting point, soon better out..
 
@eran0004 - yeah, I guess something like that must be going on. I can't get even a close approximation to the backdrop for Eifel yet. Some of the finer details show up and match, but apart from the well it looks flat by comparison - and this is with about 5x more contrast than the Death Valley one I did!

normal_3b06-1.png
 
@eran0004 - yeah, I guess something like that must be going on. I can't get even a close approximation to the backdrop for Eifel yet. Some of the finer details show up and match, but apart from the well it looks flat by comparison - and this is with about 5x more contrast than the Death Valley one I did!

View attachment 466119

That is odd. Can you try a logarithmic scale and see what happens? Like, if lowest point = 0 and highest = 1, then the true height is indicated height squared. That way the lowest and highest points would remain where they are (1^1 = 1, 0^0 = 0), but everything in between would change in scale so that the lower areas becomes flatter and the higher areas become peakier.

It's a shot in the dark, but could be worth trying.
 
Just started practising elevation maps, learned at you can alter over 20m instant jumps/drops with TPE on Eifel, learning bit more to understand logic fully, posting more later.
In exemple this track has unwanted drops, few minor changes and those are gone.

That is starting point, soon better out..


I very interested in what you find out Ode, I've been messing around with elevation aswell. I recreated CotA on Eifel for elevation, (not perfect as it's hand drawn) see below.
q5IOZXLejHNf8kVI9_0.png
.
I've managed to smooth out most of the elevation change's but can't seem to do anything with turn 15 which has become a very steep up & over tight turn.
2013-motogp-the-first-circuit-of-the-americas-impressions_1.png
.
Feel free to keep me posted with your findings as I'll do the same with mine if it helps.
Here's the track if you'd like to take a look.
https://www.gran-turismo.com/us/gt6/user/#!/friend/Mr-6rumpy1/course/1512664/
 
Last edited:
That is odd. Can you try a logarithmic scale and see what happens? Like, if lowest point = 0 and highest = 1, then the true height is indicated height squared. That way the lowest and highest points would remain where they are (1^1 = 1, 0^0 = 0), but everything in between would change in scale so that the lower areas becomes flatter and the higher areas become peakier.

It's a shot in the dark, but could be worth trying.

It's not that exactly, but right direction. I did notice that it emphasised the roughly circular creases you can see in the last pic I posted. I'd thought they were just some artifact of my crude lighting code, but then I remembered seeing something like it before...

640px-Log_by_aliasing_to_int.svg.png

The values in the heightmaps are half-precision floating-point

Which gives a nice familiar backdrop for Eifel :D


normal_3b06-fn.png


(black patches are just a glitch in my code, and the light's from a different angle, but yeah).
 
Last edited:
@Mr Grumpy deciphering wicked Japanese humor.. There is some "what goes up has to come down" pattern, if you take one Bezier downhill, then next Bezier even going same downhill has to rise/go up, at least on start of Bezier. There's logic, but something still missing. Also starting point and ending point of Bezier are crucial how TPE adapts heights, same identical part of track on "identical"(Japanese humour, you have to lay track down on separate pieces, it adapt height differences only on first time, and when moving anchors, adding anchor doesn't do a thing) course line with different anchor points can make over 20m or more height difference.

Too twisted to easily translate all findings yet, need more understanding of them.
 
It's not that exactly, but right direction. I did notice that it emphasised the roughly circular creases you can see in the last pic I posted. I'd thought they were just some artifact of my crude lighting code, but then I remembered seeing something like it before...

640px-Log_by_aliasing_to_int.svg.png

The values in the heightmaps are half-precision floating-point

Which gives a nice familiar backdrop for Eifel :D


View attachment 466214

(black patches are just a glitch in my code, and the light's from a different angle, but yeah).

Nice, there it is! :D

Can you create a heightmap from that? I mean, one that is linear like Andalusia?
 
Nice, there it is! :D

Can you create a heightmap from that? I mean, one that is linear like Andalusia?

Sure, I'll do them this evening. Andalusia isn't quite linear either, but it only uses two exponent values so almost is. I think a 1 in half-precision would be 15360 (if read as an 16-bit integer), so the values in all files are all less than 1. I might as well scale them so 0..1 becomes 0..65535 in a .raw, then see how they look in 8-bit .png.

Can we work out some reference points for determining the vertical scale? Ends of home straights perhaps?
 
Ok, I just want to make sure I'm understanding all of this;

Are the visual terrain maps provided by the app actually matching up with the existing terrain in each theme, or are they odd forms of "placeholders" to draw over?

I've got a thread where I'm testing terrain and laying methods, and the stuff in this thread could drastically change some of my more recent findings.
 
Are the visual terrain maps provided by the app actually matching up with the existing terrain in each theme, or are they odd forms of "placeholders" to draw over?

Still not sure how accurate / up-to-date they are, but I think reasonably so, at least within the valid area.
 
@Mr Grumpy deciphering wicked Japanese humor.. There is some "what goes up has to come down" pattern, if you take one Bezier downhill, then next Bezier even going same downhill has to rise/go up, at least on start of Bezier. There's logic, but something still missing. Also starting point and ending point of Bezier are crucial how TPE adapts heights, same identical part of track on "identical"(Japanese humour, you have to lay track down on separate pieces, it adapt height differences only on first time, and when moving anchors, adding anchor doesn't do a thing) course line with different anchor points can make over 20m or more height difference.

Too twisted to easily translate all findings yet, need more understanding of them.

After a little bit more of a fiddle I've managed to smooth out turn 15 a little better, it's not perfect but after deleting a anchor within the turn at the top of the hill (which affectively deleted the turn) then dragging another anchor from the bottom of the hill to create a new turn 15, it's still got elevation, just not a severe drop like before, more of a steady incline / decline now.
https://www.gran-turismo.com/us/gt6/user/#!/friend/Mr-6rumpy1/course/1515872/
 
After a little bit more of a fiddle I've managed to smooth out turn 15 a little better, it's not perfect but after deleting a anchor within the turn at the top of the hill (which affectively deleted the turn) then dragging another anchor from the bottom of the hill to create a new turn 15, it's still got elevation, just not a severe drop like before, more of a steady incline / decline now.
https://www.gran-turismo.com/us/gt6/user/#!/friend/Mr-6rumpy1/course/1515872/
That my track on video I had to draw long S and J shaped Bezier lines to keep unwanted mini mountains away(after that video version, on video there is just 1:1 turns from real), avoiding anchoring near turn on several places to keep elevation changes away from middle of corner, stretching turn Bezier to J to go across next "high rise" area of terrain, getting smooth uphill etc.
Doing track on such terrain/start point on Eifel is really really much tryout thru errors and compromises on track.
I'm having elevation map of that track and trying to hit that, much much work with this editor. Changing to bit smoother elevation area of terrain map would help, referencing terrain maps here in this thread. Good practise start point what I'm now using.
autopolis00.jpg
 
I wonder if it would be possible to replace the height maps with our own height maps? Or are the land scapes a set geometry?
I wonder this by myself from the beginning I heard that this editor will be on tablets.
 
Sure, I'll do them this evening. Andalusia isn't quite linear either, but it only uses two exponent values so almost is. I think a 1 in half-precision would be 15360 (if read as an 16-bit integer), so the values in all files are all less than 1. I might as well scale them so 0..1 becomes 0..65535 in a .raw, then see how they look in 8-bit .png.

Can we work out some reference points for determining the vertical scale? Ends of home straights perhaps?

The most accurate way of determining vertical scale is probably to make a track from the lowest point on the map to the highest point of the track and look at the elevation change in the game. For Eifel it seems to be about 500 meters from top to bottom, maybe a bit more if you go to the bottom right corner, which seems to be slightly higher than the top right corner.

If you could create linear heightmaps for all three sceneries that would be awesome. We could create really accurate contour maps from that :)

Ok, I just want to make sure I'm understanding all of this;

Are the visual terrain maps provided by the app actually matching up with the existing terrain in each theme, or are they odd forms of "placeholders" to draw over?

I've got a thread where I'm testing terrain and laying methods, and the stuff in this thread could drastically change some of my more recent findings.

I'd say that they match. I made a hillclimb track at Eifel yesterday based on the heightmap and it matches perfectly.
 
The most accurate way of determining vertical scale is probably to make a track from the lowest point on the map to the highest point of the track and look at the elevation change in the game. For Eifel it seems to be about 500 meters from top to bottom, maybe a bit more if you go to the bottom right corner, which seems to be slightly higher than the top right corner.

Yep, a reasonable range would help, but the main thing is to make sure our x-y position is accurate at the reference points, so using the smoother areas around the HS's might help (so if we're off by a bit in x-y, we wouldn't be far off in height).

If you could create linear heightmaps for all three sceneries that would be awesome. We could create really accurate contour maps from that :)



BTW, I've only just realised that the forum resizes all these native res 1024x1024 pixel .png files down to 800 pixels wide, even when they're done as thumbnails... what a pain, for this :(

edit: Thanks to Gravitron for the tip, these now link to the media section for full-res :)

Eifel Flat


Eifel


Andalusia


Death Valley
 
Last edited:
I wonder this by myself from the beginning I heard that this editor will be on tablets.
Changing the maps won't affect the in game geometry since it's a set layout. The only thing it would be useful for is to provide a much more intuitive map to layout the tracks on.
 
Doesn't look like anyone has done it yet so... here we go :cheers:

Death Valley
View attachment 464934

Eifel
View attachment 464935

Andalusia
View attachment 464933
I tried to make a replica of Virginia International Grand West using track 2 on Eifel as the back straight. Those little ravines seen cutting across the topography create huge bumps. Will try again with track 5, for VIR the back straight is at 500 ft level and the low point horseshoe turn is at 370 feet, the track kind of hangs on a side slope.

Some great info with these topo maps.
 
Changing the maps won't affect the in game geometry since it's a set layout. The only thing it would be useful for is to provide a much more intuitive map to layout the tracks on.
Sorry, you missunderstood me. I dont think to change the maps, but the data for the survace layout.
 
And to think my land navigation classes in the US army were my only reliable source of making a good track on gt5 and gt6
You have made my job a bit lazier,now I have evidence to back up my method of track making
 
Finished the Eifel contour map based on the adjusted heightmaps. For better resolution, I attached a PDF file as well.
The image below is cropped at the borders, so it only shows the available build area. The PDF file shows the entire map. There is no vertical scale, but it should be roughly 525 meters from the lowest point to the highest point, which gives an elevation difference of about 7.8 meters per line. The horizontal scale is 1000 meters for the thicker lines, 500 meters for the narrower lines and 100 meters for the dotted lines.


Eifel_contour_web.png
 

Attachments

  • Eifel_cont.pdf
    2.2 MB · Views: 109
Last edited:
Just wanted to say great job by everyone on this thread 👍 This will help me greatly for making touge tracks :D The contour provided in the app is horrible for judging elevation. Eifel for example looks a lot different from what I imagined :boggled:

Bonus points if someone can make a 3D map based on the height values :D
 
Finished the Eifel contour map based on the adjusted heightmaps. For better resolution, I attached a PDF file as well.
The image below is cropped at the borders, so it only shows the available build area. The PDF file shows the entire map. There is no vertical scale, but it should be roughly 525 meters from the lowest point to the highest point, which gives an elevation difference of about 7.8 meters per line. The horizontal scale is 1000 meters for the thicker lines, 500 meters for the narrower lines and 100 meters for the dotted lines.


View attachment 467634
I don't understand any of this, do the colours represent anything ?
 
Back