2
$\begingroup$

It started with making a skin shader. I am still convinced that we see a rough colored gloss in skin a little bit. Which is in contrast to the theory that a material is either dielectric or a metal.

I even got more convinced when making a shader for fabric, there are some fabrics that appear quite metallic to me like Satin, some sports-wear, new carpets, and some more. It feels like I can tell now how much of the rough colored gloss is present in a material like grass 5%, sand 10%, bricks 10%, satin 70%, skin 5%, some plastics 10% etc.

But this is in contrast with that theory that a material is either dielectric or a metal. Any idea what I think I'm seeing then ?

Maybe a theory is that for example in Satin, the light bounces in the grooves of the fabric, and this causes the reflections that reach my eye to now have some color influenced by the color of the fabric.

So I tried to simulate that in Blender: A diffuse with a PBR reflection node (Cynicat Pro), and gave it Bump with a distorted wave texture with a very high scale (1000). The color of the gloss from the reflection node didn't change toward the color of the diffuse shader which was pink.

Then I tried to make the gloss more rough and that didn't work either. I tried that setup with 16 gloss bounces but that didn't have any effect.

So I am a bit confused, I hope some one is able to clarify what I believe I'm seeing. I made a thread on BA as well

The property rough-colored-gloss in dielectric materials

$\endgroup$
8
  • $\begingroup$ Could it be some sort of translucency or Sub Surface Scattering that you are seeing? Most materials are not totally opaque, and fabrics being generally thin sheets tend to let some light through them. Have you tried mixing it with a translucent node to give it a softer appearance? $\endgroup$ Jun 17, 2016 at 17:07
  • $\begingroup$ Blender Cycles is physically accurate so it could be that your node set-up needs some tweaking. $\endgroup$ Jun 17, 2016 at 18:56
  • $\begingroup$ @Duarte Farrajota Ramos That was what I thought to in the first place and translucency is an important property in fabrics, though doesn't explain the metallicness in fabrics like satin. When a satin bed sheet is on the bed, translucency will do nothing ( No light is shinging at the back and we won't see much of its translucency, while the satin effect remains the same on bed or as curtain. ) $\endgroup$ Jun 17, 2016 at 21:22
  • $\begingroup$ @David Prentice Hmm, If Cycles was physically accurate I wouldn't have problems with normals facing backwards when using bump ( The black artifacts when the bump is to strong), and I could do the double slit experiment. As far as I know Cycles implemented some formula based on physics, but doesn't mean that Cycles is physically accurate. I know that there are quite some problems with the gloss shader and rendering of normals when using bump. The question is why does satin ( the satin curtain in my real life bedroom I.E. ) looks so metallic to me while it is a dielectric. $\endgroup$ Jun 17, 2016 at 21:25
  • 1
    $\begingroup$ @DuarteFarrajotaRamos Thanks for the suggestions. I will play with the desaturation and coloration of the gloss shader. I will do a bit of experiments of the effect of different weaves on how they have influence on the reflections. $\endgroup$ Jun 18, 2016 at 10:38

1 Answer 1

1
$\begingroup$

The theory I had seems to be correct: the light bounces in the ridges of the fabric and give the gloss a color toward the diffuse. We cannot simulate this in blender with a bumpmap or a normalmap. So I did an experiment with real geometrie. Here two illustrations. First illustration shows that we cannot achieve this with a bumpmap. Second illustration shows that we can achieve this in blender only with real geometrie and enough light bounces.
The light is not bouncing in the ridges of the bumpmap and therefore doesn't return a colored gloss The light bounces in the ridges and return a colored gloss which makes the material look metallic while it is pure dielectric

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .