I was doing some experimenting with rock surfacing, and this is the first one I'm moderately happy with. Still early days though. C + C welcome.
It looks like some of my stuff when I really take time to make a good surface. (Better actually) What I found was that it comes down to scale and masking. Where does the surface need to be rough, like where something has broken off. Or smooth, like where it has been eroded away by water. Getting it just right is like looking for the Holy Grail. You can get close, but often end up in terrible peril in Castle Anthrax.
Hey, BEAUTIFUL!!!!!! 8)
Put this in a Cafe and you would have 'Hard Rock Cafe' ;D
Really though, this is very nice and natural looking.
really nice looking - almost igneous!
Thanks everyone, for your comments. njeneb is certainly right that getting the scales and the masking right is important.
Quote from: Thelby on January 16, 2010, 12:06:02 PM
Put this in a Cafe and you would have 'Hard Rock Cafe' ;D
Lollerskates falling out of a roflcopter. :D
This rocks! Did you guys ever try to set up a network of rocky displacement shaders (twist, strata, etc) and then feed this into the terrain generate (from shader)? I did. It wouldn't use the sideways distorted areas, I figured, but you can add erosion then and might end up with a nice terrain.
---Dune
Thanks Dune. I'm not quite clear what you're asking re the rock shaders though. Are you talking about taking a procedural terrain with other "sideways-displacement" shaders added, and then generating a .ter from it? IN which case, why not make the .ter straight from the procedural, then the erosion, then add the "sideways-displacment" shaders in later? Or am I missing something? I've made a .ter from an Alpine fractal, which already has erosion built in, and I'm planning to use this surfacing on it.
We're on the same line, but once you have the erosion and then displace sideways, it wouldn't be 'real' erosion anymore. Very interesting, though. What I just meant was that it would be interesting to put a whole lot of nodes before the shader input in generate terrain, instead of just one fractal. And then apply erosion.
Quote from: Dune on January 18, 2010, 03:50:40 AM
We're on the same line, but once you have the erosion and then displace sideways, it wouldn't be 'real' erosion anymore. Very interesting, though. What I just meant was that it would be interesting to put a whole lot of nodes before the shader input in generate terrain, instead of just one fractal. And then apply erosion.
I've tried this before, it's dead slow, but cool :)
@Dune - I see what you mean - so eroding a terrain with overhangs etc. isn't possible (I can kind of understand, as the program will effectively find two different values for the height at the same spot, which probably throws it into a loop). I wonder - do GeoControl or World Machine allow this?
@Martin - When you say "it's dead slow", do you mean generating the terrain is slow? Because presumably once the heightfield is generated, it renders pretty quick?
Probably the render. I have done some very complex displacement work, and it takes forever to render on my machine.
No, the render and heightfieldgeneration is just fine.
I meant that the erosion operator is very slow.
It often crashes here, and if it works, an hour or more is not rare.
Quote from: domdib on January 18, 2010, 08:10:37 AM
so eroding a terrain with overhangs etc. isn't possible (I can kind of understand, as the program will effectively find two different values for the height at the same spot, which probably throws it into a loop). I wonder - do GeoControl or World Machine allow this?
No they don't, for exactly those reasons. These systems generate verticle displacement only. Of course you can then push/pull these things in TG to create overhangs. But another good aspect of these programs, is the ability to export masks for each layer of erosion. So for example you could use a fake stones shader only on a generated scree slope...
OK, I added strata and fractalised Voronoi cracks. Next, I may add light displacement to the base colour.
This is much better. The texture is more 'real' looking.
The only problem I have with that is there's a huge spike in the upper left part of the image.
@TBH - what, you mean the interesting landscape anomaly? :)
Anyway, here is the next iteration - I've now added some small-scale displacement. Ignoring minor render glitches, do you think this is believable? (apart from the obvious lack of water erosion - let's just say this is in an arid country ;))
I think a surface layer should be added to make some of the flat and large slope areas smoother. Use the intersect underlaying with 85% smoothing.
It's really coming along.
Dominic, I find the previous image more interesting. I think in the last iteration there's just too much of these "downwards flowing shapes". Like, the rocks had been liquid once, and then froze suddenly. That being said, I often struggle with this sometimes unwanted effect, too.
Personal taste perhaps, but sometimes less detail is better. Instead, I would try to add small fake stones everywhere.
Cheers,
Frank
@njeneb - I've done something similar to your suggestion in an image below.
@Frank - I think I know what you mean re the frozen shapes, and the level of detail. It's partly because I'm seeing how far I can push things right now - and there is a small amount of breakage. From now on, I'm going to be dialling back the displacements a bit. Although this began as pure exploration, the seeds of an actual image are beginning to grow, so I may add in some fake stones once I'm happy - and I'm also planning to add some of Martin's sand.
After there's some fake stones, can you post a tgd?
Sure - I was actually thinking of doing a beginners/intermediate tutorial on this stuff, once I've finished exploring.
those are some really cool/good displacements! :)
This last image is what I had in mind. From here it should not be too hard to get many types of igneous rock. Color variations and changes in the reflective index would do much of this.
Changes in the reflective index - done! (Actually, it's tiny fake stones)
A very nice looking rock surface!
This seems nearly finished, isn't it?
Only thing left perhaps is a more clean crisp render, then it will probably be quite photorealistic :)
What were the rendersettings for this one, including lighting and atmosphere?
I like 'smoother.jpg' the best, really perfect. The colours and displacements work together really well. I wouldn't have added the tiny fake stones.
both smoother and shiney have their merits...very good work here, love to see a tgd/node shot?
I think the small fake stones may be too much. I would have used a PF for a mask which is scaled very small for a slight reflective shader. (I never do, render time...)
TGD, TGC and brief explanation now up at http://forums.planetside.co.uk/index.php?topic=8668.0
Here is the final version. Thanks to everyone for your comments.
@Martin - the render settings for this one : Detail 0.9, AA 6, GI 2/4, Catmull-Rom. Previous render was 0.6, 6, 1/2
Lighting and atmo on both are just the standard out-of-the-box ones.
Here is another version, with the addition of negatively displaced fake stones - an effect that I'd probably tone down with a painted shader if I did it again.
Perhaps a tad bit too rough, but I still like this very much!
About your rendersettings, very well chosen in my opinion. The GI 2/4 really pays off I think. Hence the lighting and coloring in the shadows.
You mentioned that you didn't alter the atmosphere settings and that's probably the culprit why it isn't looking so clean.
I suggest you render this with at least 48 atmo samples, preferably with 64 or 80 for perfect clean results, but that's taking quite long.
If you need renderhelp then just let me know, I'm happy to do it for you.
Cheers,
Martin
Wow - that's come on a lot! I prefer the second last one, though :)
@Martin - well, you might have already tried this yourself now, but I tried your suggestion of upping the atmo, and didn't really see much difference. I then switched GI Surface Details on (pleasantly surprised that this only increased render time by about a third - perhaps an optimisation in 2.1??) and I think I can see some subtle differences. See what you think.
The next two images also had GI Surface Details on. Moved to a higher POV...
Then I switched the sun heading by 180, and was surprised by the difference this made to the rock colour. N.B. that's the ONLY setting I changed.
Hmmm...the result is not as spectacular as I would have expected.
You have changed two things, upping atmo samples and enabling GI surface details, so it's hard to say which is most responsible for the improvement.
I haven't looked at it by the way...after seeing this my first guess is playing with the haze settings (less dense) along with increasing atmo samples slightly and see if it then looks better.
These kind of color-schemes together with heavy displacements tend to look noisy in the shadowy areas very fast. Hence some of my canyon-renders and my "inside deep forest" render. Suffering from the same difficult conditions for the renderer, I used 256 atmo samples there.
Martin
Finally, two closer renders - the first one shows a really weird displacement - no idea how it came about. The second one, from a slightly different POV, has Intersect Underlying switched on in the Roughen Surface node, and shows an interesting almost polished finish in the foreground.
Quote from: domdib on January 21, 2010, 05:33:02 PM
Then I switched the sun heading by 180, and was surprised by the difference this made to the rock colour. N.B. that's the ONLY setting I changed.
That's not such a big surprise honestly.
First of all you changed the POV dramatically which makes quite a difference.
Second you added a lot more direct lighting by moving the sun behind the camera.
I hope you agree with me that this looks dull, flat and much worse this way.
If you prefer to solve the lighting problem by moving the sun, instead of tinkering with the atmo settings, then I'd get back to your original POV and try moving the sun around in 10 degree increments to see if some extra direct lighting will give better results.
I might try upping the atmo samples further, as I didn't notice any big difference in render time. And reducing haze is also a good idea - I often switch it off altogether. Thanks for your suggestions.
EDIT: And I neglected to mention earlier that I made a render with just the atmo samples increased (to 80), and could see no difference - that's why I switched the GI Surface details on.
OK, switching off haze and upping the atmo samples to 200 does make a difference - the shadows are much bluer, and arguably 'cleaner'. Is this the sort of thing you were driving at, Martin?
Martin? ;D
Oops...thought I had replied to this already, sorry :)
Anyhow, this was indeed more or less where I was "driving at"...though the shadows are bluer and lots cleaner they still have a strange hue/tone.
Sometimes I find it easier to use TG's gamma and contrast settings and then render straight to final with barely doing postwork afterwards.
With surfaces and lighting like this I often run into this same kind of challenge of lighting the shadows without getting a strange hue etc.
My suggestion for this now would be to actually lower the GI strength and color on surfaces somewhat and then render at slight higher exposure, or adjust that in post.
Too strong GI on surfaces can cause a strange hue. My first strata wall image is a perfect example of this.
Bedtime now, if needed I will contemplate more about this :)
Cheers,
Martin