some more experimental rock!

[inkydigit]

...

[inkydigit]

...
finally I got this...

the base terrain for the last series uses functions (3d Perlin)
with approx 300+ shaders!!
I will tinker some more

[Jo Kariboo]

Hi Jason,
I like much these experiments. Very massive with original forms. My preferred are: Reply N. 3-4-6-11-13-15.
Some have a recalling aspect of the animals of prehistory.

[Tangled-Universe]

Nice experiments, keep on going
I wonder: 300 shaders? for what?

Cheers,
Martin

[inkydigit]

thanks Jo, I was going to call one of these Mammoth!
and Martin....don't ask!!!
probably overkill on my part, there are a lot of different scales of displacement/warped and stretched functions and power fractals...I may have been able to achieve similar with less, but was fairly methodical in the structure of scale/displacement ratios etc....
I will keep going with this and try to complete a full scene...

[Tangled-Universe]

Yes I think you could do with less....say 20-30 nodes? What I observe is a kind of crack function, a heightfield/simple shape shader as terrain base and a redirect to push it outwards into different directions. Then on top of that at most half a dozen of texture shaders. That's what I think I'm seeing

Oh well, doesn't matter, as long as it's fun and it surely is

[Oshyan]

Some of these are quite nice-looking and somewhat unique from other things I've seen on the forums here. From your description, is it safe to assume this is all procedural, no image maps? I'm particularly interested in:

rocky_one-06_11_1.jpg
rocky_one_06_11_02.jpg
rocky_one_06_11_03.jpg
base_perlin_rocky_one_06_11_01.jpg
base_perlin_rocky_one_06_11_03.jpg
and
base_perlin_rocky_one_06_11_07.jpg

- Oshyan

[inkydigit]

Hi Oshyan, thanks, and yes, 100% procedural, no image maps anywhere...I may strip this back to the base terrain and laterally displaced functions(voronoi) and redirect shaders and see what happens, it is easy to get convoluted when reworking different parts, to achieve uniformity of randomness!!!
I will keep you posted!
btw the 'rocky_ones' use calico's canyon build as a base, which I modified quite a bit, the 'base_perlin' ones as I said use 2 3d perlin functions and a colour adjust shader and displacement shaders for the main terrain structures, with some redirect and more stacked voronoi and power fractals for the progressively decreasing (often lateral) displacements...

[Oshyan]

Thanks for the info. Some of these look very much like Dax Pandhi's rock work in Vue, which is partly why they interest me. I've always known that stuff was achievable in TG2, but few people seem to go in that direction in developing shaders. So this is nice to see. Looking forward to seeing further progress.

- Oshyan

[inkydigit]

thanks, here is a higher res/quality version of the last render, with some added clouds, no post work just jpegged...

[Kadri]

Different and nice rocks , inkydigit!

[Dune]

Cool tests, Inky. I really like some of them. But as Martin says, you should be able to use less then 300+ shaders. The 3D perlin is very uniform (1 scale at the time, as far as I know), if you replace that with normal perlin, you'd have more variation at once.
The voronoi would be a very interesting addition if it could be built as a multiscale modulator, instead of having to pile all scales together in blue nodes.

[inkydigit]

thanks Dune and Kadri,

300+ shaders may be a bit misleading, 300+ nodes would be more accurate (lots of merged voronoi etc as displacement inputs for individual surface shaders, of which there arent that many), the 3d perlin functions I wanted to create a simple 'blobby' landscape which would be easier to displace without broken looking parts, I found some problems when using ridged and billowed perlin especially before a redirect...
anyway, as I said this is a work in progress and I will try some more...
cheers
Jason

[Tangled-Universe]

I figured you meant 300 nodes and not shaders, hence I suggested it could be achieved with 20-30 nodes.

Anyway, you can avoid broken parts in displacements by making sure that the displacement amplitude is respecting the scales of the fractal.
An exaggerated example would be displacing 10m scales by 100m. That's asking for troubles.
Also, not having too many octaves helps too to avoid breaking up and other unwanted artefacts.

Same goes for using the redirect shader.
You can use a reasonably high octave noise as long as you keep the displacement amplitude lower to equal at most as the feature scale.

Another nice new feature which allows a lot of tricks with a minimum amount of fractals is using the updated transform shader.
You can now besides moving and scaling, also rotate noise.
Merging a differently transformed powerfractal can give very interesting results, especially if you use mild scaling differences and 45 degree rotation on x- and z-axis. I've found out.

[inkydigit]

thanks Martin, yes that sounds interesting...I will check that out,
cheeers
J