Blasted

Started by efflux, January 20, 2009, 03:47:44 PM

Previous topic - Next topic

efflux

#15
Cyber-Angel is right about the clouds. I tried changing them but it spoilt the sense of scale. This planet is stuck with these clouds. If I change them it changes the whole character. However, I have changed the terrain which is better now. I'm after an other world look but those big bulbous hills in the distance were too much. The reason those appeared is that I used altitude to displace the whole hillside. I could show you the graph for that but if you think it through it's quite simple. You have a value coming from altitude node which obviously increases with altitude so you can use that to change various functions. The benefit with this basic displacement is that it pushes the whole hillside smoothly before any rougher displacements over the top. I've used this a few times. This was clamped to not displace any further beyond a certain altitude but it still displaced too much on the higher hills. This is sorted because the terrain now rarely climbs higher than was intended. New render coming.

Thanks for comments.

efflux

#16
New render.

EXR again. No separate treatment of atmosphere and surface this time. The scene was not so much in shadow so it wasn't such an issue. Maybe I am overcorrecting exposure? The trouble with EXR is that the tweaking becomes even more dependent on monitor. Cinepaint has exposure and gamma warning but is not set up for my monitor. I need to look into these issues more deeply. Last image was tweaked on the Linux system where the Quadro driver was defaulted to turn the monitor to high brightness then when I switched to the Mac (also plugged through the Eizo - dual DVI) it was not set to full brightness and the image didn't look too cool.

efflux

One problem with this planet is that I did a lot of detail on the lower ground but I have found that POVs from these low areas don't seem to work too well. I'll have another search with this new terrain arrangement.

Esgalachoir

I agree with everyone else, totally awesome...

RArcher

I really like the composition of the second one.  The surfacing is fantastic on both and the clouds are in a better position now as well.

rcallicotte

Quite lovely.  Bravo.  Great work.
So this is Disney World.  Can we live here?

Mohawk20

The second one is a great image, but you lost the best feature: the light from within the rock. I liked that the best, and it doesn't show much on this one.
Howgh!

Volker Harun

Oh I love this rim ... did you use 'smoothing' for cutting into the spires?

efflux

#23
Yeah. I know I lost the glow. It only really shows up where the rock is in shadow but then I get the exposure problems in the shade. I'm thinking of setting up a planet where I have control of all the surface lightness as a kind of exposure hack.

Volker. I used elements of smoothing just about everywhere so yes if I understand what you mean then smoothing was used as you describe.

The terrain that the rocks are on is actually simply a perlin billows powerfractal. There is another perlin fractal for terrain which supplies some lower undulations but I've found that you can tweak all the parameters of the power fractal to get kind of spires but not lots of nasty sharp spikes because you can use the spike control and also the smooth control in compute terrain. You don't really want spikes but more smooth bulges. Ironically I did use the other spire technique for some very small rocks which you can faintly see in the first render. Discussed here:

http://forums.planetside.co.uk/index.php?topic=2530.0

Except in files on that thread a colour adjust shader was used to get something to mask with. I replaced this with a conditional scalar.

The reason for using this technique was more to do with the fact that I needed to clamp and mask what looks like small stones so they had different materials (can't really be seen clearly in these first two renders). Fake stones are cool but they are voronoi and always look so. Clamping and masking perlin gives you some other options.

The ideal set up (or really compromise) for this kind of extreme displacement is to set up terrain that is peaky and steep as possible while keeping it smooth with low detail then the smaller overlying displacement provide that detail without having lots of underlying terrain detail clashing with it. Smoothing in the surface layers is also effective. This is one of the great things about TG2's displacements. There are lots of various ways to smooth them out. Other apps tend to get masses of artifacts with nasty clashing angles.

You also find a problem at the hill peaks. This is where big displacements tend to get seriously messed up. The altitude drive I use blows the terrain up like a balloon and so smoothes out this top area = smoother overlying displacements.

efflux

Funny. I just read that old thread. The problems were so easy and obvious. Cut to the chase and go to the end of that thread where some files are provided if you are interested.

The main issues were caused by lack of documentation on exactly how TG2 works! Had to be done the hard way.

efflux

Here is a shot of the entire graph for this planet. You won't be able to work out details from this. Node names are shortened and it's a mass of connections but it shows a general data flow that I am going to continue with. The blue groups are handling much of the displacements in a kind of separate branch which hooks back into the first layer which is a sort of base layer.

Seth

Oo'

woah ! now I understand your profile's pic !

efflux

He he. I should go back to an old profile pic that I used to use but I didn't like the idea that people think I'm smoking drugs.

The graph isn't as complex as you might think. Since I build stuff on planetary scale then I guess this means lots of procedural tweaking.

Mohawk20

Quote from: seth93 on January 24, 2009, 08:17:41 PM
Oo'

woah ! now I understand your profile's pic !

An uncanny resemblance indeed! 8)
Howgh!