6 Favourites2 Comments108 Views
Got this working. My cavern mesh is too messed up to use it with, though. I'll have to make my own caverns.
This is up there with my convolution node setup in terms of complexity, I guess, and I'll have to make a scene out of it so that I can upload it to blendswap.
The node setup mixes between six different node groups on the basis of normal. Like the faces of a cube. Each face has its own texture, bump, and roughness map, and they get combined. It has three UV maps, just ortho projections, to read the textures. You could put some other stuff in there if you wanted to, mix it with some veins running through, create a color gradient on the basis of position, something like that.
So far, it looks pretty good. There's still too much influence from nearly parallel shaders, giving it too much of a projection-mapped feel, so I'll try reducing the weight from dot(normal, cubeside) to dot(normal, cubeside)^2.
Of course, this is bakeable to unwrapped UV. One of the reasons I wanted to do this was so that I could texture character models from renders rather than bakes, because the renders look a lot better-- they have color correction, they don't screw with the camera (uggh baking velvet shaders.) There would be problems with concavity, but nothing that couldn't be solved.
2560x1440px 805.86 KB