Middle Earth DEM Project

[Redrobes]

Yes, same place - just called MindolGlaciers2.ter or something like that.

[monks]

Cheers, I did a wee erun last night. This looks a LOT different to the last one. I had a render coming down but Windows decided to install an update and rebooted my computer grrrr!!
I'll try another today...

monks

[monks]

I'm currently talking to Frank B on the TG boards about the glacier terrain. Hopefully we'll sort something from that. don't see why not.

While I was waiting fro someone to bite I've started to add vegetation in. VERY basic right now. Still gotta control placement via image mask:

http://www.skindustry.net/medem/file...on test_01.png


http://www.skindustry.net/medem/file...on test_02.png


http://www.skindustry.net/medem/file...on test_03.png


monks

[monks]

some population tests:

closeup. This is the grass clump. This is not a great model. I've got better ones than this knocking around.

http://www.skindustry.net/medem/file...on test_04.png

full coverage of the area

http://www.skindustry.net/medem/file...on test_06.png

coverage controlled via imagemap shader.
http://www.skindustry.net/medem/file...on test_07.png

monks

[Redrobes]

Y'know on test 06 the snow does look very odd - like its in places on the low lands... When I ran the glacier creation stuff I didnt put in the thermal adjustment layer for the volcanic ridge - whats that called, the one that separates Mordor. So there was Mindol on one side as a splat and a long line of snow on the other side. You haven't got the texture reversed so the line of snow is being put on Mindol ? Just a vague thought there because it shouldn't have generated snow in low lands.

The grass looks alright but its a bit uniform. Is there a way to say put a little more here and a little less here instead of the all or nothing mask. If its a B&W one only then maybe we should take GTS veg maps and create a series of bands of veg quantities out of them. Then you can put in different types of veg depending. Of course trees would look just dandy.

EDIT -- actually looking at it some more I can see its the right snow for the mountain but the mountain has been smoothed out on the right lowering it drastically from the original.

EDIT2 -- Just been reading the planetside thread. You have to remember that the ice sheet height is probably a +ve amount of height all over but its very thin in places and thicker in others - esp the flow zone - as shown in the Wilbur screenie. The point is that if your trying to get TG2 to give rock poking through then I expect that its a problem in the colour bit of it not the height bit of it. Presumably you have to set up a colour shader program for how white you want snow depending on depth. In my screenie above from DF that colour texture was generated from GTS and it uses a min snow depth before it shows any snow and then just linearly ramps up white until it reaches a preset max level. That quickie tex generator from GTS is quick and dirty. To do the snow properly in GTS you need to use the scriptable texture compositor program thing where you set the depths and the graph of colour vs depth up yourself. I expect that you will need something akin to that in TG2. So where the fluid is less than 0.1m say, just make the white amount zero and where its 2m or something like that make the white full and expect that in between it will be rendered as thin ice in some non physics kind of way. Perhaps the actual height map with snow added is not so bad. Maybe it needs to be a little less depth on the snow that is shown in the wips above tho. I dont think you would be able to see that front edge of the snow on the mountain in the background. Just my 2p Ideally we should be able to export the height GTS generated into TG2 accurately but going via .ter is not so easy for me. Does TG2 do HF2/HFZ yet ?

[monks]

Yeh, I mentioned this previously. I think Seer has done something with the dem. Well for now it's only a test. The better imagemaps should come from GTS anyway based on the original terrain. Exactly, the veg maps could give some very pretty results. The grass can be broken up with some fake stones. Not sure what the computational demand is of fake stones, but if it's high then there should be a way of limiting to the fov anyway- just like the veg populations. It could easily be varied procedurally in colour based on slope, altitude I'm sure. There's some heather maps in there too so that will use a different plant model; the heather will simply correspond to GTS moorland I reckon.
The grass coverage in the last shot should actually be trees anyway.

Had os e more feedback on the glacier. I'm trying another approach as suggested n the TG boards. No luck so far. I'm using a distribution shader instead of the image map shader for the ice- still using the terrain tho- I was querrying the need for both a terrain AND an ice mask generated from the ice terrain...seemd a bit unnecessary.



monks

[Redrobes]

Ok cool. I am not familiar with TG2 so I dont know whats in a distribution shader or whether that can deal out height and color at the same time. Your right in that it seems like it kinda could.

Keep us posted and let me know anything that I can help or push it forward with. Really want to see Mindol get some extreme TG2 goodness.

I have the height map in HF2 format. Would it help if I gave you a contour map of the fluid / snow height for a reference ?

The problem was that I had to go via PNG16 to get to ter so I lost the real world values in it.

[Redrobes]

Contour bands at 100ft intervals. I guess I let a lot of snow fall

[monks]

Hey Robes, just a quick update. this is two plant models using different population distributions:

http://www.skindustry.net/medem/file...on test_09.png

I'm not sure what you mean in your last post...?
I haven't made any more progress on the snow but I'll get it. I'm learning all the time here. I'm trying to get trees in at the moment- lots of texture maps to set up. It may require that the heightmap be respanned in TG. Shouldn't be too difficult at all in theory.

monks

[Redrobes]

Cool.

I guess you didn't understand about the png bit ? Well HF2 has heights in floats where the values mean real world stuff - its a calibrated format. When going to .ter I have to go HF2 -> PNG16 -> TER. By going through PNG16 the values of that format are just 16bit grey so we lose the real world calibration. I dont even know whether TER files have real world calibration but whatever, I have lost it by that point so the only way to recover that is to get a contour map of the HF2 and try to make the import height match up with what I had originally. Because GTS uses real world heights and is doing a fluid sim then its quite important that the height is the same from GTS to TG2 in order for it to look right - probably some leeway but you get the idea.

So, if you can make the import of the TER file about 600ft of snow at maximum depth then it would be about right.