I think normal map is actually way to go. I used normal map to create first sample image with spheres, second one was with bump. But I guess for this to work flawlesssly I need normal map that is not created from height map that usually has very little height difference, but rather painted or generated to properly direct fuzz strands, not sure how to say it.
Indeed @Eneen velvet material looks good, but the shader setup sounds like a hack. I’d intuitively try something like that, but in the sheen inputs
A normal input can be added to the sheen layer of the Principled BSDF if you’re using OSL. The Principled BSDF OSL code can be found here, copied to Blender, and be modified: blender/node_principled_bsdf.osl at main - blender - Blender Projects
Obviously, OSL isn’t a option for some people.
There are plans to make most, if not all of the layers in the Principled BSDF avaliable as nodes, with layering support (E.g. layer a material on top of another). Once that project is done, you could use that to create a custom Principled BSDF with a sheen normal input. But there’s no clear time line on this “every layer as a node” project.
I’d call it solution. Sheen inherits normals from coating and you usually don’t use coating with sheen so it’s good enough. Ofc separate normal input would be nice though.
Hi!
This is my first time on this forum and I saw a link specifically to your comment with weed, so even though it’s a long time ago, I’ll reply to you!
I believe I have got a shader that I think works identically, or at least similarly, to the octane solution.
same exposure
Add
Mix 50 50
add
mix 50 50
grass with add shader both image
Add
Mix 50 50
Seems like you add translucent to principled not subtracting reflected light at incident angles from translucent first. This needs thin-sheet mode to be integrated into principled or thin-sheet-translucent node with option for separate materials for both sides of geometry.
Yes, exactly. The correct setup exists but is extremely clunky. I pointed it over 6 years ago:
In those 6 years, we got all sorts of gimmicks in the Principled shader, but thin translucency, which appears in nearly every scene (foliage, thin fabric, paper) is still not implemented.
Overwhelming majority of asset packs for Blender released in the meanwhile do the thin translucency setup wrong, and it’s hard to blame them given how complicated the correct setup is.
Can I ask why? I guess I don’t know.
As far as I can see, diffuse and translucency are mutually exclusive, just like with mixnodes, only in the case of mixnodes the maximum value of each shader at a point in time is limited by the fact that it is truncated by the mix value.
If we use what I suggested - At each moment in time we have 100% of the shader, no more. And we will see either diffuse or translucency, but this will be controlled by the incident light, just like with mixnode.
I’d like to hear your opinion. What’s wrong with this case?
Add
Mix
Add
Mix
Add
Mix
Add+inv multiply
Only add
It doesn’t actually matter if it’s mix or add, it is “correct” only when leaf/transmitter surface is pure diffuse and that’s not the case here…
Yes, ideally you first have to mix translucency with diffuse, and then put reflection on top. All the more reason to have it integrated in Principled BSDF because the amount of knowledge average user needs to have to set this up correctly is overwhelming for many.
This is why the building blocks exist in the first place, together with manipulation of normals you can create a lot of effects that the Principled Shader will probably not do out of the box for a while yet (ie. dispersion). A big reason for the ‘Cycles’ look that people often mentioned was because of the tendency to have Principled be the only shading node in a material.
But yet, farther development on Principled to make it on par with industry-leading engines is important for ease-of-use (since like it or not, it is unrealistic to expect many to make their own setups in an age where people want results quick).
It should be that way. Both for the reason of speed/ease to use as well as compatibility with software other than Blender. Setup complexity is only justified for complex results. Not for essential features like thin translucency.
I just checked the octane now. I messed up when I plugged in the lightness dependency. If you plug in value, the result is similar. And there will be no more excessive energy.
Cycles
Octane
Octane Thin wall(diffuse)
Separate coating and sheen-coat nodes would partially solve this problem (still no anisotropic thin transmitter though). AFAIK there will be separate metal node @Alaska made, so maybe those are possible also in short term?
There is a plan to provide sheen and glossy material layering in Cycles, and it’s already avaliable if you’re using OSL (which is supported on the CPU, and NVIDIA GPUs through OptiX, although with a serious driver bug at the moment). But it seems other things have been deemed higher priority so layering with nodes has been delayed.