Cycles's GSoC 2018 project: new hair shader! Expectations?

General Forums Blender and CG Discussions
21

Weird, I set up a debian jessie based build environment precisely to avoid this issue.

Could you post the error somewhere?

22

Sure, here’s a cl screenshot.

error1
error1917×118 150 KB

as text:
./blender: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version CXXABI_1.3.11' not found (required by ./blender) ./blender: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version GLIBCXX_3.4.22’ not found (required by ./blender)
./blender: /lib/x86_64-linux-gnu/libm.so.6: version GLIBC_2.27' not found (required by ./blender) ./blender: /lib/x86_64-linux-gnu/libc.so.6: version GLIBC_2.25’ not found (required by ./blender)

23

Hi.
With last commit of yesterday, Rendered preview works but Render image does not work (you get only black result in hair).

24

Hey @YAFU,

It may be because of the Diffuse fallback - I had to turn it off because it wasn’t compatible with OSL.

Can you (or anyone who’s getting the black hair error) git bisect the branch and tell me the commit where it stopped working?

25

I forgot to say that I’m in Linux just in case

I have no idea how “git bisect” is used. But

Hash 07696962c4d: Render image works well.
Hash 25677d77385: Render image only gives black hair result.

In both Rendered preview they seem to be working well.

26

Hm, in 25677d7738530d8f827e5ac960ccded6b128d897 I added the last missing feature (Chiang’s primary reflection roughness modifier), so I had to (in layman terms) take all the curve geometry data out of the shader’s main data source. This change should only affect OSL, though.

Can you try with 4774e906e19816b720c20271b9f9db28bb89debe and see if it’s fixed? If not (or it crashes outright), please send me a .blend with the bug (and the stacktrace if your OS gives it).

27

Hi.
“Render Image” still does not work with new commits, hair looks black (but works in Rendered preview).
It happens with any file, anyway here one:
http://pasteall.org/blend/index.php?id=49628

But better you wait until Lukas or another developer can confirm the problem, there is a possibility that I am doing something wrong.

28

I’m seeing the same error too (Linux Mint (18.3)). Seems like the GLIBC 2.23 Mint comes with isn’t good enough (ldd --version). But to be honest, I don’t know if there is anything we can do about that.

29

No need @YAFU - I could reproduce it (I was using a complex node tree to test, you used the shader alone).

I’ll come back later in the week and tell you when and how I fixed it :+1:

1 Like
30

Hey all,

Brecht fixed @YAFU’s bug in 1fac3b7da05ada27d8a41c114283394b5a505ebb. (I didn’t know I had to tell Cycles how many data sources I was using :sweat_smile:).

I also added the following features:

  • OSL shader
    • (I finally understood how OSL works in Cycles. I’ll certainly add docs to Blender-Wiki)
  • Chiang’s Primary Reflection Roughness modifier (i.e. “shiny coat on rough-looking fur”)
  • Support for Cycles’s Filter Glossy modifier
  • The default color type is now set to “Direct coloring” (ie. plug and play).

Please, have a look at my branch and tell me if you find any other bugs. As previously, I’ll release a set of builds on Monday.

1 Like
31

Can we expect a more user-friendly way of altering melanin levels akin to some of the other production renderer shaders out there?

1 Like
32

Hey @m9105826,

Do you mean, like Arnold’s Standard Hair?

About the melanin - I can certainly add two extra sockets, like Arnold’s, to set the melanin quantities (and use them if you set “Melanin concentration”).
As for the rest - I think they may be added too.

In any case, I’ll ask Brecht and Lukas what they think about it.

33

Here’s a new Linux build based on the latest commit (7780d306).

I think I solved the problem that caused the build to not work on older distros, now it should work on every distro that can run official builds.

34

The melanin controls certainly should get their own dedicated sockets, putting them in the color field was just a quick hack I think.

1 Like
35

Hey all,

Builds for this week’s commits are now up.

Let’s recap the added features:

I also added:

  • Color, Melanin and Absorption coefficients now have their own sockets.
    • Choose your parametrization and the right socket will appear
  • Help tooltips on each parametrization

Build links:

2 Likes
36

Is constrained melanin randomization possible with the new setup? It can go a long way towards creating a realistic look, especially for fur.

1 Like
37

I’m not sure whether this should be part of the node itself, I’d rather add a per-hair random value to some input node to give more flexibility.

38

Melanin randomization is physically based part of the shader, it should be in the node. On top of that, it could be useful to have some hair ID attribute for artistic control.

1 Like
39

Something like this?

ss
ss1920×1050 306 KB

40

Yep, that looks like what I was expecting!