Loading *multiple* DEM files

[TheBadger]

oops look down

[TheBadger]

[attach=1]
[attach=2]

Ok,
So, I have four .img in four files. I believe that they are the four corners of the terrain area I selected. Does that sound right?

Anyway, I did get this to work with one file from another location. But when I set in on this terrain, (the other was just a test) I got what you see in the image above. There are also four files in that test location, but they appear only to overlay.

So what am I doing wrong now?

[jo]

Hi Michael,

Check the "Replace NODATA values" checkbox. If you look down below the "Height range" text there is a "Min. valid height" field. This is telling you the lowest height in the DEM which is not a NODATA value. So in this case 0 is the lowest height in the DEM and if you use that for replacing the NODATA values all of those will match the lowest valid height.

Your DEMs are most likely overlapping because you have georeferencing turned off. This means they're all placed at the position set by the heightfield shader, which is 0, 0 by default. I'd suggest turning georeferencing turn on and then moving the planet origin as described in the Geog heightfield load docs:

http://www.planetside.co.uk/wiki/index.php?title=Geog_Heightfield_Load

Regards,

Jo

[TheBadger]

Hey, jo. Thanks for responding.

So what you wrote sorta worked. I mean I see that what you wrote (and the instructions in the wiki) should have solved the problem. But there are some issues that are showing up still.

See here:
[attach=1]

There are some intense displacement issues happening, despite that there should not be simply because the location I downloaded is a place in Utah USA that dose not have cliffs like this (it looks more like AZ, which I also looked at. Maybe they sent me the wrong data?)

Anyway, here is the TGD with the 4 files loaded. You or Oshyan or Matt may want to have a look. One to help me, but there is something else too. When I clicked 'up-date' I lost my terrain. I had to unclick all 'georeference', then re click them on, to get the terrain back. Is that normal?

And then just if its interesting, I also have 112 zip files of Orthoimagery to go along with this terrain ((34GB zipped)(Not included below))  took forever to download, and no idea what Ill do with it. But I ain't playing! 

Zip file of terrain and files (42.3 MB on disk).
https://mega.co.nz/#fm/W1oHiJwD

[RArcher]

I didn't download your file but i'm guessing you downloaded the 1/9 arc second DEM data.  It is fairly common that there are large sections of missing data in this data set because they have only surveyed certain bits and pieces that they thought were important or needed at the time.  It looks like you have huge canyons all over the place because this is the location where there is no data present so it defaults to the Min Valid Height which is obviously set lower than the actual DEM height.  Try again with the 1/3 arc second data.

[TheBadger]

Hi Ryan, very cool to see a post from you!

After reading your post and looking at my data with what you wrote in mind, I think you are exactly right. Actually that's to bad! I wish that I had known that before, I wasted a lot of time getting that data. But glad I know it now
So thanks much!

So now I know how to do this all finally! Hurrah!

But I have one or two remaining questions on the basics. If you look at this image:
[attach=1]
There is that really nasty seem between where the two sections meet (the channel running up the mid of the capture). What if anything can I do to get rid of it, or even just mitigate it?! 

And then,

I remember some discussion about how to capture and or translate a hightfield into a map or something, so that I can use it as a simple displacement.

So what I want to know is can anyone remember the thread where Oshyan told us how to copy out a hightfield and then bring it back in, in none hieghtfield form (of any kind or way)?.. I KNOW that I remember that Oshyan talked in a post with me about doing something like what I'm asking about ((quite a long time ago) (maybe having to do with erosion stuff)), but I'm sure I can't remember the right words to phrase my question clearly.
If anyone here can understand what Im trying to ask about please speak up!

I'm trying to work through and then document a super complex workflow having to do with both whats in this post, and whats in that big Vector displacement thread. But I need a little more help on this part of it before I can move on. THANKS!

[TheBadger]

Ok, except for the questions in my last post, I am making progress!

[bobbystahr]

Ok, except for the questions in my last post, I am making progress!

looking promising

[jo]

Hi Michael,

The seam is coming from the border blending parameter in the Heightfield shader. You can probably set it to 0. I did that in my tests and didn't notice anything too bad from the edges.

If you want to use a DEM as an image map for displacement I would experiment with the Geog image map shader. This can load various DEM file formats as image maps, although it's main purpose is to load geospatial images like satellite photos.

Regards,

Jo

[TheBadger]

Thanks!

[bigben]

The seam is coming from the border blending parameter in the Heightfield shader. You can probably set it to 0. I did that in my tests and didn't notice anything too bad from the edges.

Just curious, does that mean you did notice something at the edges? It might be worth combining the DEM files first into a single geotiff so that you'll at least get interpolation of any height differences at the borders. Then you only have to load a single file as well. I think 3DEM (free) does this, and while it's no longer under development it's still available. http://freegeographytools.com/2009/3dem-website-is-gone-but-3dem-still-available-here

I saw some minor differences between heightfields when I was converting DEMs to TER and tiling them. Not a problem when viewed at a distance but enough that I expanded the size so there would be overlap for blending. Might not be as bad with original data though.

[jo]

Hi Ben,

What I meant was nothing bad jumped out at me. I wasn't really looking for it per se, more for how things were lining up across edges, but I don't recall any problems. That doesn't mean there might not be if you were really looking for them, or in different situations.

Regards,

Jo