I wish TG2 could do this:

[Hetzen]

The problem is that TG lays a texture grid over the top of any displacement to generate texture co-ordinates, almost like laying a blanket over your terrain, and only looking down on the undulations.

If Matt could work out a Voxel way of defining texture co-ordinates, a lot of problems would be solved, like slope populations thinning out, creating truely square forms, stretched noise on hard slopes etc.

[Dune]

Try smooth terrain and small patch size in the compute terrain...

[Hetzen]

Try smooth terrain and small patch size in the compute terrain...

That's the work around, but you still see the issues described in my earlier post. As anything gets near to being verticle, theres not enough points in the patch size to fully cover the facets.

[mhaze]

That's very clearly put Hetzen, it's made some of the problems I've had more understandable.

[mhaze]

Right I give up.  This is as good as I can get.

[mhaze]

For those who may not understand whats happening here and to make the problem clearer, I've made these three quick images based on the tgd above.

The first shows how textures are distorted on a vertical face
The second shows how using smooth step adds some slope to the sides and enables a better texture
The third shows what I call the tablecloth effect. Imagine a cloth is draped over the SSS you get wrinkles at the corners stretching up in a inverted v.

[Hetzen]

Good effort Mike. I had very similar problems with the Wyvorm's Realm pic I started. I had all sorts of weird population projection angles over my form to try and solve some of these issues. I realised that TG is not really the program to do these sorts of things. Still, it's fun trying.

[Tangled-Universe]

The problem is that TG lays a texture grid over the top of any displacement to generate texture co-ordinates, almost like laying a blanket over your terrain, and only looking down on the undulations.

If Matt could work out a Voxel way of defining texture co-ordinates, a lot of problems would be solved, like slope populations thinning out, creating truely square forms, stretched noise on hard slopes etc.

Exactly why I named this thread this way.
Despite that the noise functions are 3D internally, in the end they are in fact 2D or 2.5D at its best.
It also explains why stretching/compressing noise in Y often gives weird or no results as it's just projections of a 3D noise function.

[Tangled-Universe]

Right I give up.  This is as good as I can get.

Thanks Mick. That really looks more than decent. Why don't you populate it and try to finish it?
I haven't had the time today to dive into your setups and see if I can take it further from here.
I'll post my results as soon as I have them!

Cheers,
Martin

[Tangled-Universe]

From the same guy. The example I showed was a product of his entry for the Vue Environment Competition. This is his entry:

[bla bla 2]

Facile, après on peut multiplier le nombre de simple shape shader par le node mutli shader, en faisant copier/coller, plus cas rassembler un peu partout et ajouter un cloud layer.

Easy, then we can multiply the number of simple shape the shader node mutli shader, by copy / paste, if more reassemble around and add a cloud layer.

(google translate)

[bla bla 2]

Je peux pas faire plus de deux tours. 

[Jack]

Remember this from the tg2 gallery?



who did this again?

[Tangled-Universe]

That's from Marcello Deschino a.k.a. as Martchi.

[bla bla 2]

Et, voilà.