@LucaRood@Aeraglyx@kolupsy I guess either plug the numerical derivative into the bump node
or without the bump node using finite differences, like this one
Though the result isn't quite smooth without increasing the epsilon
Would this work for what you need?
@LucaRood@Aeraglyx@kolupsy Ah yeah it's the scale that messed things up, dividing the function by the scale fixes it now.
Thought it was a bug or something, never crossed my mind it's the scaleπ
@Aeraglyx@LucaRood@kolupsy It doesn't smooth it out, just scales/offsets the line in the y axis, it still retains the details of the function
The line was thin on flat areas for some reason that's why I added half of the thickness to the Z
That alone makes the line look even enough I think
@LucaRood@Aeraglyx@kolupsy Nice method!
I made a couple of improvements on it to make it look more even and closer to iq's method.
the sqrt(3)/4 is just a random number I found that works, idk if there's any math behind it π
Here's a comparison showing their differences:
idk why I didn't do this before
but changing the Rayleigh coefficients instead is better than just multiplying a base color at the end
gives you more control on how the colors appear altogether
Little update on my sky generator
Changed the sun settings, exposed more settings and added limb darkening
Also you can use images as the sun now :P
#blender#b3d#eevee
I'm still contemplating whether to keep everything in one nodegroup
or make the sun separate, as there'll be other texture mappings for it, which would make it modular instead π€
Oi you can take HQ screenshots of your nodetrees now!
How?
With PrintNodes, my new addon
And the beautiful tree in the demo, thanks to @cmzw_ for that!
#b3d#Mayterials#Nodevember
https://t.co/6jSWHBPKN8
https://t.co/f5xLVmqil5
@lorenschmidt Yep that's finite differences, hence why the step is a small epsilon
Should be the other way around for forward differences, f(x + h) - f(x)
https://t.co/c19b7L24CO