Have anyone figured out how new Curvature Smooth node works in Designer?

gnoop · Sep 1 2025

So far I find it best for doing various edge masks just on top of a final normal map . Tried to make it work in Painter but as always it's weird and never same , works only on square textures i.e a dead end.

Want to recreate it for Blender shader editor.

I see it sort of shift iteratively in x and y positive and negative detection the dot products of normal XY channels vs plain XY vectors and then subtract from before shift . No actual blur involved probably which is extremely hard to do in Blender. just loops of fading shifts. But this is mostly my guess. I couldn't understand some parts at all when open the node in Designer . Am I on the right path ?

Image: https://us.v-cdn.net/5021068/uploads/editor/wb/ez32vnqt9f2n.jpg

I would want to give it a kind of a wind effect to have brighter slope down to certain direction . Any ideas how to do it?

poopipe · Sep 2 2025

it's a bit beyond what you can do with a stateless pixel shader effect afaik. standard curvature is doable with a sobel (like) filter but this needs to consider a much larger range

if you want to mask by a specific normal direction though - just dot(normal, other normal)

gnoop · Sep 3 2025

poopipe From what I see in the node Designer just makes copies scaled down images till 2x2 pix where it perform roughly same and upscale back . Instead of doing blur or Sobel like . But it's extremely hard to read and figure out form their math nodes really as if they specifically tried to make it cryptic vs Blender for example so I am not sure . Some parts I don't understand at all. Like where they set their variables and what for . I hardly see any "set" nodes.

poopipe · Sep 3 2025

I (naively) assumed they'd done it properly
Hacks like that explain why it breaks down under some circumstances

edit:

ok so I've had a look . Hopefully this helps with some of the weird bits
The only variable they seem to be using is the builtin $sizelog2 -which is log2(pixel size of the input) == log2(512, 512) == (9, 9) etc..

The constants they declare are mostly the log2 of image size - this function which sets the image size on one of the blends ensures that the final output is 32x32 px. if you pass a 64,64 px node in, it will set the relative size to -1 : (5,5) -(6,6) = (-1, -1)

Image: https://us.v-cdn.net/5021068/uploads/editor/sj/ves0ovwmki53.png

In terms of what it does..
You're basically right...
They're splitting the normal map into x and y components, making several copies, blurring them to various degrees and blending them back together with less and less influence as the sharpness (image size) increases (The blends set opacity with (1.0 / $sizelog2) ). This is a hacky/ingenious way to pretend you're sampling multiple times over distance and is a similar principle to building fractal noise.

Things i noticed...

I don't think this will work properly at over 4k.
The very small (16x16) images are treated slightly differently to the >= 32x32 images but the principle appears to be the same
There are a lot of nodes (the switch blends on the right side of the graph and the nodes that feed them) that don't appear to do anything but guard against data it doesn't like - you can delete quite a big chunk of the graph without changing the output
This could probably be implemented more elegantly but in terms of output I think its as good as you'll get from designer. I find it slightly disappointing that they're not implementing some of these nodes as a c++ plugin (the vector/directional blur would be a good candidate)

Have anyone figured out how new Curvature Smooth node works in Designer?

Replies