I discovered something interesting....

Started by digitalguru, November 06, 2018, 05:17:40 PM

Previous topic - Next topic

digitalguru

... though I'm not entirely sure why it's working  :P

I've got a stack of fake stones shaders piped into a network via a surface shader, but some of the stones in shadowed areas look unnaturally dark like there are NaN's in the color values or something:
[attachimg=1]
So I thought I'd try to split the displacement and the color so I could clamp the black values and try to fix it, but the fake stione shaders are contributing both the dsp and color so I thought I'd pipe the output through two merge shaders, each Mix to A set to 0 and Merge Color turned off on one and Merge Displacement turned off on the other (not even sure if that would work!)

When I plugged the color merged only node into my surface layer - it fixed the dark stones:
[attachimg=2]

It's a complete fluke and I don't know why it's fixing that, but I'm not complaining :)

anyway here's the merge node that is the last node in the tree plugged into the surface layer
[attachimg=3]
[attachimg=4]




digitalguru

oh well, don't think it was that much of a discovery after all...

thought I still had some variation of color in the stones, but I've lost a lot of that - think it's more down to find a good median point for the base stone color and balancing that with the color variation

look likes it's easy to push the color values into an area that reacts unnaturally in the shadowed areas


WAS

This is really interesting, and I think explains some issues I've had using the colour variation where I've just opted out and done a surface layer masked by the scalar of the stones with the white point set to 0.

But this has me thinking, could you capture the stone colours with Colour to Vector and than clamp/augment them with a duplicated fake stone shader, and than pipe that as the surface of the second (original) fake stones shader?


WAS

#3
Example of what I was talking about if I didn't make sense (I know I struggle explaining myself)


Matt

The "colour to vector" is redundant in this case. The clamp colour node expects an input of colour, not vector. It can get the colour from the shader, so just pipe the shader in.
Just because milk is white doesn't mean that clouds are made of milk.

WAS

Quote from: Matt on November 06, 2018, 10:27:53 PM
The "colour to vector" is redundant in this case. The clamp colour node expects an input of colour, not vector. It can get the colour from the shader, so just pipe the shader in.

I continue to forget this, yet, I literally run into it all the time when manipulating colours and forgetting I need to add a surface layer to feed displacement through the Main Input and do colouring with the Colour Input or Child Input.

*Repeats in head*