Bathymetry/terrain test

Started by bigben, September 27, 2007, 06:39:22 PM

Previous topic - Next topic

bigben

Here are a few images from a bathymetry/terrain test. The terrain has a flat ocean to which I applied a water shader and several image masked layers to simulate different water depths. The image masks consist of a single TIF derived from 1km bathymetry data, which is then fed into a colour adjust shader to adjust the "depth" distribution of each surface.  The effect is greatly exaggerated (visibility down to 4-5km below the surface) and falls apart completely when you get closer.

[attachthumb=#5]

From here I'm experimenting with making adjustments to only the animatable settings of surface shaders to see if I can come up with an acceptable version for both mid and close range renders with the aim of determining a relationship between the value of the settings and camera altitude so that they can be set via a database generated CHAN file for animations.

The image map used is only 8bit and it's looking like I may need to either use a 16bit image or increase the separation in the highlights of the image to provide a greater variation in shallow water. Also adding some additional fractals to add some extra subsurface detail, but I needed to see what variation I was getting from the masks first.

The third image is just a closer view of the terrain which is a set of 6 TER files all at 90m resolution. The masking was a lot simpler for this set because it's split into 2 separate sets of 3 so I only needed a mask for the middle terrain in each group with an inverse of this mask being applied to the other two.

[attachthumb=#4]

Detail 1, AA4, No GI, render times about 1 1/2 hours each.

Lucio

Your work is always very interesting bigben :)

nvseal

This is amazing to say the least. Masterful work you are doing here bigben. Can't wait to see more.

bigben

Here's a modified version of the mid range render. A single dark blue fractal has been added as a quick addition for extra subsurface detail. Apart from that, the main tweaks have been in the colour adjust shaders (black level and gamma) and coverage values for the surfaces. Still a little rough, but the comparison with the previous version demonstrates the principle.

I'll definitely have to get some more highlight detail in the mask image to stop the shallow water looking too muddy. I left this image in its raw form for these tests to see how far I could push things with colour adjust shaders. Most of the gamma adjustments were fairly hefty so I'll probably use these as a base adjustment for the original image. Some unsharp masking may also be in order but it's looking very promising so far.

A similar principle may work for clouds, but of course they're a lot more complex than colouring a flat surface.

efflux


cyphyr

Your really getting somewhere here (well NZ & OZ actually :D ). I'd say there was a job in here somewhere :)
Richard
www.richardfraservfx.com
https://www.facebook.com/RichardFraserVFX/
/|\

Ryzen 9 5950X OC@4Ghz, 64Gb (TG4 benchmark 4:13)

rcallicotte

Wow.  Now this is looking cool. 
So this is Disney World.  Can we live here?

sjefen

Your work is very interesting. Keep it up.
ArtStation: https://www.artstation.com/royalt

AMD Ryzen Threadripper 1950X
128 GB RAM
GeForce RTX 3060 12GB

dhavalmistry

"His blood-terragen level is 99.99%...he is definitely drunk on Terragen!"

Volker Harun


bigben

#10
On holidays, so I'm not online often. I ran a second render to see things "a little bigger".  The render below has the atmosphere turned off to see the full effect. Bathymetry data is about 1:1 on screen.  The bathymetry image was run through adjust levels - shadows/highlights - levels - shadows/highlights - levels to get a more aesthetic result. With the atmosphere on the effect is a lot more subtle and looks quite good.

At lower altitudes however the data is inadequate to produce a good looking result so it may be necessary to hack a fake depth for shallow water using a blurred terrain mask and merging that with the low res bathymetry data.. but that's potentially quite difficult.

There's a small glitch along a seam east of NZ (look near shallow water at top left) which I've since fixed. This was due to the distortion of low res data when converting to UTM projection.

I've also created some extra terrains for the bigger islands to the south and east of NZ (no overlap = no masking needed  :)) Additional subsurface "detail" was removed for this render.  Now for some extra surfacing on the terrain....

dhavalmistry

"His blood-terragen level is 99.99%...he is definitely drunk on Terragen!"