Using the B channel from the normal map to define the roughness would be more correct, as it stores the "depth" of the normal map. Using the albedo is totally wrong as it has way to much variation in value and is not related to the roughness. Better to use a generic light scratches and dirt map for roughness, as it represents everything you cannot feel while touching the surface
Hey John, thanks for your comment! As you could see, I wasn't using normal maps in this video, I'm quite often using just a single RGB texture to generate all inputs from the RGB channels :-) I think as long as the final look is "good enough" it's all good man!
Hmm, I think that the end result you are getting is likely wrong because texture maps like Normal and Roughness are designed to be used with no further color space conversions. Hence, the corresponding 'Image Texture' nodes should have their 'Color Space' option set to 'Non-Color Data'. Another note I would like to add is that Principled BSDF shader (and its underlying physical calculations) has been devised to more realistically represent and visualize real world materials and their interaction with light. Plugging random textures into the shader's inputs might give interesting results, but should be avoided if you are aiming for realism. Otherwise, you might just as well use any other shader besides the Principled shader ;). Regarding the Specular input, the correct value is dependant on IOR (Index of Refraction) of the material in question and is calculated by the following formula: SpecularValue = ((IOR - 1) / (IOR + 1))² / 0.08 In any case, thanks for the tutorial. It is interesting to see and learn different techniques.
Hey Marach, thanks for your comment! From what I've tested, if you set the bump or roughness map to Non-Color Data and compare it with te RGB, the only noticeable difference is the image input smoothness, nothing else. I'm using Principled BSDF because it's a decent, overall shader base that allows creating decent looking materials. I've never treated it as a "realistic" nor "physical" way to simulate real world materials, that's why my approach to texture inputs is more "output focused" :-)
As you said, the indentations should be more rough, meaning that it won't be as "shiny" as the rest of the surface, so why then invert the roughness texture if you had white values at rough/indented points? As far as I know, the black value (0) is totally reflective and white (1) is rough... Also, doesn't the specular value in principled shader act as an IOR for non-transparent materials (the default 0.5 on the spec. should be about 1.45 I think)?
What chocofur means, is at 2:07 he isn't looking at an isolated black and white texture, this is the rendered view. Because the cracks were reflecting more (white) light than the rest, they appeared white, and vice versa.
I think the specularity is not really dependent on albedo. The screen of a smartphone is covered with some special glass to suppress reflections I suppose. Try for example to compare two things made of black and white plastic. The dependency becomes real when you are dealing with something like cavities, not just darker zones
Just observe the materials around you. In general, dark surfaces tend to have less specular (or less bright reflections if you want to put it in different terms). The "old" way of making shaders when you mix the Diffuse and Glossy nodes still allows you to control the reflection brightness manually by adjusting the glossy color.
@@Chocofur Just checked with two glossy plastic things: black and white. Reflections at sharp angles are absolutely of same brightness. So glossy component seems not affected by albedo
@@ВолосатыйДжо at sharp angles (the closer to 0 angle the more) every surface has bright, mirror-like reflections. That's what I've covered in my other video and we even have an addon that creates that effect in Blender (as the Principled BSDF doesn't support that effect by default).
@@Chocofur Yes, I saw that video. I think the nature of this "mirror" phenomenon is same IOR dependent reflection that is (was) controlled by fresnel node. It doesn't work out of box because we substitute the true surface relief with fake things like roughness and normal map (but maybe I'm completely wrong here). For glossy dielectric surface we have that mirror-like reflections without additional shader modifications.
so I have an object spinning in my scene, it has a white line that pops up on the bottom edge and traces the edge as it rotates, should I make the face that is affected separate from the rest of the mesh and just have reflection on that part? or should I dim some lighting? I'm at a loss
You forgot to change you're maps to non color data and putting roughness into the specular is very unprofessional, as it produces unrealistic results. Specular is the same for most materials you shouldn't change it.
Both roughness and specular values change depending on the material. A rule of thumb - the darker the color the lower the specular value (or reflection brightness if you like).
@@Chocofur I get what you are saying, but this topic addresses "Specular Highlights" not Reflection. This title might confuse beginners. In Blender to modify Reflection, you have to adjust the "Metallic" property (in the Principled BSDF). The Specular property is used to modify the "Brightness" of the highlight (which are only aggregating intense light values of scene elements), and the Roughness property is used to modify the smoothness of the surface. You could simply name the tutorial "Material Specular Highlights the RIGHT Way, in Blender 2.8" Nevertheless, the tutorial is very good.
@@Chocofur Sorry, Because I'm coming from a Maya workspace where we use don't use specular to manage reflection, I will defer to your explanation. I noticed that other Blender Tuts use your approach, as well. And If you want me to delete my comments, I will.