2013-08-19

Shinines vs Rougness vs XXX poll

I have a question about preference, since this seems to differ between people I talk to... and I'm starting to be curious as what is actually most "widespread":

The question: For a glossy reflection shader, what parameter do you want to define how "glossy" reflectoins are... and what "direction" should it work?

Clearly the name "glossiness" (that mia_material) uses is suboptimal, because some people associate "more glossiness" with "more blurry reflections" whereas others associate "more glossiness" with "more polished i.e. closer to a mirror".

So there are a few choices, either go with "shininess" which is easier to understand that a "higher" value is "more like a mirror". However, Physically Based Shading is talking increasingly about "roughness" as the value (the MILA shaders use that, for example). Unfortunatly, the "roughness" as defined in physics goes from "zero to infinity" in a very visually "nonlinear" way, and hence is very hard to map a texture to. In the other end of the spectrum we have "specular exponents" like Phong, that also go from "zero to infinity" but the other way around.....

So .... I just want to hear from you, the user community, what you think is preferred.

/Z



/Z

16 comments:

Rawalanche said...

NEVER EVER!
EVER!
EVER!
EVER!
EVER!
EVER!
EVERRRRRRRRRRRR!

seriously...
EVER!
use roughness value...
EVER!

It is one of the biggest mistakes that were ever done in CG.

If you look at a black and white map, that would be considered a glossiness map, your brain would automatically identify brighter values as those more shiny ones, and darker as those rougher ones.

If you create a simple plane, look from an angle on it, and put an area light on opposite direction, shining on the plane. Then apply glossiness map on it, you will see the glossier parts of the plane as brighter ones. It is just way more natural and intuitive.

I have worked with several pieces of software where they did a mistake of using roughness value instead of glossiness. Even after spending several months adapting, it remained to be very counter-intuitive solution, and it took a lot longer to get the results tweaked the way i needed them. Of course other people may have difference.

That being said. There is already established standard, so changing this standard would mean all the glossiness maps i have saved on my hard drive will become invalid. Or that i have to feed them through invert node everytime i use them, and then tweak glossiness with absolutely inverted mindset, where increasing contrast decreases contrast and increasing strength decreases strength.

There is really no need to re-invent the wheel that rolls just fine. If this is related to mental ray, then you often get complains about falloff of the glossiness curve(i think it's called a lobe), but never about direction or range of the glossiness value, so there is no need to touch that.

Master Zap said...

So the problem (and the reason I ask) is that at SIGGRAPH, every single set of Physically Plausible Shading paper I saw discussed "rougness", and every single Physically Based Shading pipeline talked about "rougness maps" as a normal thing. I come more from the "shininess" camp myself, so I wanted to know.. well... how the world is divided here :)

/Z

Rawalanche said...

True :)

My response was so explosive, because i can not imagine how destructive effects would have changing glossiness just like that. I got quite scared when i saw roughness in MILA.

Also, i would swap the world "shininess" for "glossiness" as it is more familiar to people who use mainstream renderers. And those who vote very quickly may not fully realize what they are voting for ;)

ronviers said...

i agree with Rawalanche, i don't paint roughness maps i paint shininess or glossiness maps

Kevin Deguisne said...

personally : Glossiness with 1=mirror

it's he same for opacity/refraction/transparency
I don't see the problem ... we used to click invert check-box for those maps when we change renderer :)

GustavoEB said...

come on rawalanche, this is just the way you see things based on your past experience... dont make it into universal truth.

from a linguistic standpoint IMHO "shiny" and "glossy" are poor synonyms of the physical phenomena of "mirror like reflections" (a much more precise term, again IMHO)

maybe if you do need to have a one word name for a map where white means "so polished it reflects like a mirror" the best word would be "smoothness" but this, obviously, can also bear trouble with other cg terms

so:
*if everyone else is calling it roughness
*if it works well in practice (never had any problems with it)
*if it is a term that trully expresses a physical phenomena (assuming you are working in a physically plausible renderer)

why use a a different term?

i mean, of course it works, but why be the one using the imperial system when everyone else uses the metric system?

bnrayner said...

I prefer something that you can easily texture map, i.e. a scale from 0 to 1. Other than that, I don't really care if you call it "glossiness" or "roughness", or if a value of 1.0 is mirror or diffuse. It does appear that "roughness" is the more trendy terminology ATM.

Rawalanche said...

I do not care what is it called. I would be okay with roughness. But i am not sure you actually understand what this is about. This is not about terminology, it is about what black and white colors on the map represent.

Also, you can not just click invert on glossiness maps, they are very rarely usable as they come. They need to be tweaked, especially when almost every renderer has a different glossiness curve (Vray vs Mental ray for example), and having black areas on the image representing maximal shininess, instead of white one, makes entire process a lot more difficult.

ronviers said...

It makes sense for darker values to represent less and lighter values more. It also makes sense for lighter values to represent the parts of the material that will render brightest. So i prefer shininess or glossiness

Master Zap said...

Except that a shiny area doesn't *necessarily* render brighter.

At angles of light away from the reflection direction, the "rougher" areas catch more light than the "shinier" areas, and would actually be brighter. So it's not that "simple".

/Z

ronviers said...

even with the exceptions, it seems counterintuitive to use black to represent areas that are mirror smooth

ronviers said...

recessed areas of a surface are more likely to be occluded and less likely to receive wear. Typically when making a glossy map i will isolate the lighter values, apply a falloff and add noises at different scales. Then use this as a stencil/stamp. So if 'roughness' is used, i will need to visualize the occlusion as lighter, which means either working backwards or inverting my stencils/stamps

Unknown said...

I feel you! It's a tough call because, as you say, commonly understood English definitions don't map exactly to the 'scientific' definitions.

After reading the comments, I'm gonna go with "Glossiness" with 0=diffuse and 1=sharp for reasons of intuitive use. Maybe that's silly reasoning. (I'm pretty sure I voted for 'Roughness' in the poll so it's all a wash.) In any case, a big no-way to 'infinity' being a value for texture maps. That way lies madness.

Timothy Hess said...

I say we call it "bluriness" which is unambiguous and straight to the point.

Chris Ford said...
This comment has been removed by the author.
Chris Ford said...

But if it were "Blurryness" then you'd have to invert the way the glossiness slider worked, so why not call it "Sharpness"? this way you can keep the same 0 to 1 mapping the glossiness attribute has... 1 is a pure sharp reflection, 0 is the blurryest / pure diffuse reflection.