Testing geotiff

Started by bigben, June 26, 2014, 10:33:21 AM

Previous topic - Next topic

bigben

Not a particularly pretty image, but I finally got around to playing with geotiffs in TG3 and couldn't help getting a little nostalgic. Hands up if you remember being impressed by the 513x513 TER of Cape Royal in TG0.9  ;)   I went on to play with tiled terrains and managed to create some large padded terrain sets by masking multiple TER files... but the handling of geotiffs in TG3 is freaking awesome compared to that.  Most of this terrain set is out of frame, but pretty much most of this frame is 10m data along the length of the Grand Canyon, with the left and right edges being padded with 120m resolution for another 50kms or so and finally 500m resolution centered on the canyon and extending all the way to the west coast.  3 geotiffs exported from GlobalMapper and not a mask in sight. Certainly makes preparing variable LOD terrains along a flight path very easy.

Tangled-Universe

Very cool!

Is this rendered from the georeferenced position or did you transform everything closer to the origin coordinates?

How large is the dataset in total?
Is it at share-able proportions? ;) :)

Cheers,
Martin

bigben

#2
The two low res images total 60Mb, the main canyon is 475Mb, going as far west as Vulcan's Throne. 1Gb RAM to load, another to render. I relocated them to the top of the planet.

I made another image for the area west of that but that's another 410Mb (they're 32bit, LZW compressed).  I may have to be a little more methodical and split it up into more manageable squares which would also let me crop a little closer to the canyon edge. That would also let you use less high res data if you're only interested in a specific area.

It is very nice being able to simply load higher res terrains on top of the others without having to mask.  Must dig out a global DEM model with bathymetry to see if it's practical to do the full outer space to surface animation.

[edit]  Had a quick play, pulling down a 0.05° global geotiff (105Mb) and the 2 higher resolution images. Had to remove the outer image from the set as it had gaps at the edges. Should be able to share the data as it will now be a mixture of multiple (public) data sources. Definitely needs the intermediate terrain to improve the transition around the main terrain but it's looking pretty good.  Blue surface just altitude limited to 1m, but I'd use an ocean mask  to exclude Death Valley and other low lying areas.

Also had to turn off ray tracing objects for the high altitude render to avoid black patches.[/edit]

bigben

Digging into my archives I found my bathymetry dataset and a 20k blue marble texture.  There's no ocean/atmosphere, and the ocean bed is coloured with a clip I pulled out from an old file based on http://www.planetside.co.uk/forums/index.php/topic,2968.msg30137.html#msg30137
Global elevation and texture at 0.05°. Full res texture (0.0083°/px) around the Grand Canyon

Dune

Hi Ben. Interesting stuff, good to see this develop!

Mahnmut

That´s what I need for Mars!!

bigben

#6
Quote from: Mahnmut on June 27, 2014, 04:16:13 PM
That´s what I need for Mars!!
Texture map as geotiff? or planetary DEM as geotiff?
Having a couple of issues making the latter but this is 0.05° per pixel on a Mars-sized planet

bigben

Ok, I think I've got Mars sorted. The DEMs were loading with 180° long. at the centre which was causing a few difficulties.  I also started with the 128 dataset which has a 2° gap at the poles.  So for DEMs I've loaded both the 64 and 128px/° sets and  got a 32K texture from Celestia and montaged it with Imagemagick.

Still at 0.05°/px with the added texture. For a planetary scale 0.025°/px is pretty easy for TG to handle

mhaze


bigben

The previous Mars images were a long way from the origin (I'd put the north pole at the top). The detail is OK from a distance, but up close it's pretty degraded. I could barely make out the crater at the top of Olympus Mons let alone the smaller craters inside it. This image has the volcano moved to the top of the planet and has an extra terrain and texture at 0.001°/px (although this will have some interpolation in the terrain). Much better :)

I'm running a quick animation to see how the detail holds up in an approach to the surface.  180 frames to go.

TheBadger

Quotemars-render_05c.jpg
winner.
It has been eaten.

bigben

I cut the animation short because there were some obvious issues with the terrain. I should have used a fixed multiple of the resolution of the DEM for resampling it.  I've fixed that and am running a different animation down to 2m altitude to test the limit of the terrain and image map.

TheBadger

looks great to me. Why is it blue though?
It has been eaten.

bigben

#13
The texture map was slightly desaturated by the person who compiled it, but not all of Mars is "red". http://nssdc.gsfc.nasa.gov/image/planetary/mars/marsglobe3.jpg. That and possibly one of my atmosphere settings. It was just a quick compilation. I've tweaked the atmosphere a bit and will probably increase the saturation a little.

Oshyan

Nice work Ben. Now just integrate HiRise data into it and you can go orbit-to-ground no problem. :D By the way, there's another person around here who has been working on a Mars scene with that kind of scale range, running into a few issues along the way. Perhaps you and he could put heads together. His nickname is Mahnmut:
http://www.planetside.co.uk/forums/index.php?action=profile;u=2662

- Oshyan