Anonymous at Tue, 9 Apr 2024 11:31:58 UTC No. 980043

cris

Anonymous at Wed, 10 Apr 2024 04:40:41 UTC No. 980115

>>980040
i know this is a schizo thread but why is this real advice? can someone explain this to a retard like me?

Anonymous at Wed, 10 Apr 2024 08:39:30 UTC No. 980124

>>980115
i think it's because of the disney shading model that everyone uses. under the hood the math does the diffuse component of non-metal materials is completely separate from the math for the metalness portion of the shader.

in practice however, you can have grey values:
https://twitter.com/ArtOfPilgrim/status/1749902580359065640

Anonymous at Wed, 10 Apr 2024 09:36:59 UTC No. 980127

>>980115
Because in physics atoms are either in the metal group or they're not and some retard that was clever enough to do damage
put this knowledge out there without enough context, and retards took this knowledge as gospel and ran with it.
Claiming you could reduce 'metallness' to a on/off switch because materials are either conductive or dialectic.

Metal atoms have an electron shield that reflect photons soon as they touch them this is why metals looks so shiny and mirror like.
Insulators/dialectics don't behave this way and appear diffuse or translucent as photons escapes into the surface thru the tiny surface imperfections and get's absorbed and remitted scattering every which way, why they don't appear as mirrorlike as metals no matter how much you polish them.

However what this naive understanding fails to to consider is how optical phenomena we see every day don't work like that.
Metallic surfaces are rarely pure or completely uncontaminated. If they are oxidized, has any sort of residue or oil on top or is a
compound material existing both of conductive and dialectic atoms and therefore exhibit optical properties of both at each pixel.

Pic related is steel viewed under a microscope. You can see just how much of the crystalline surface of steel is occupied by carbon, which isn't a metal.

Anonymous at Wed, 10 Apr 2024 10:08:37 UTC No. 980129

>>980127
There are no known materials that are both ferromagnetic an transparent. Likewise, the Indium-Tin-Oxide that's used for TVs and solar panels is such a rarity in that it's both somewhat conductive and also transparent to visible light.
The point is that PBR is not realistic anyway. It's Disney-realistic. It's a chinese plastic simulator essentially.

Anonymous at Wed, 10 Apr 2024 10:29:09 UTC No. 980130

>>980129
>The point is that PBR is not realistic anyway. It's Disney-realistic. It's a chinese plastic simulator essentially.

I disagree, PBR is very much on the right track, it's bad rep is from how it clashes in art styles and implementations by people who use a
trashy variant of it and lack the finesse to make the most of it. Instead of going for a subtle realism they go for hyper realism where they
over-emphasize the characteristics of materials to present what ends up reading like a cartoonish variant of the thing they depict.
It typically looks liek the material they intend it to be but it comes across as sterile and artificial like something from a photo-studio of
a surface that been wiped clean from all imperfections inside a clean room.

I've implemented PBR myself based on the 'Disney Diffuse BRDF' and they're very open about what it is and isn't if you read that original paper.
You don't like something about the look you can just change it. What's genius about the approach over legacy shaders is how it get's
you thinking about shading in a very unified way. It ensure that you develop assets that will looks just as you expect and behave as expected
regardless of composition and light conditions.

For example I don't use the same specular lobe as they do and that goes a long way to remove that typical 'disney plastic' look.

If you know some HLSL or equvallent everything you need to know to roll your own is available here.
https://media.disneyanimation.com/uploads/production/publication_asset/48/asset/s2012_pbs_disney_brdf_notes_v3.pdf

Anonymous at Wed, 10 Apr 2024 10:45:36 UTC No. 980131

>>980130
Like people dislike PBR because implemented naively it allows someone who's bad at art make something that shades very naturalistic. If you combine subpar textures and models with very realistic light you end up with something that looks to us like toys made of various rubbers and plastics because those are the only surfaces in the real world that look anything like the artwork presented.

So PBR is a lot like the Marvelman, you have great responsibility to not fuck up because you're dual-wielding too much power.
with legacy shaders if you where shit your shading would look equally shit and that worked much like the drawing of a child
where the art and the shading goes together and it looks charming.

With PBR you get the drawing of a child where stick-figure man wearing triangles for pants now has meticulous depth and subtle shape revealing ambient light and falloff. But it no longer mesh, it's no longer charming as each element presented makes a mockery of it's counterpart.

Anonymous at Wed, 10 Apr 2024 11:33:16 UTC No. 980133

>>980130
To me and to most other hobbyists it doesn't matter because I'm not trying to color match real footage and rendered CGI. I don't want or need the complexity it brings in and whether it looks good or not in the end is a matter of debate.
>>980131
That's right and that's why I call it Patriarchy Based Rendering. With the OBJ model, you could throw anything at it it would come out as equally mediocre. That's inclusivity. PBR tries to appeal to the elites instead and it's trying too hard to the point that most modern stuff is fatiguing watch.
You have to wonder why after pushing PBR so hard, now everybody wants to do NPR instead. And they're trying too hard to do that too.

Anonymous at Wed, 10 Apr 2024 12:20:22 UTC No. 980135

>>980133
>You have to wonder why after pushing PBR so hard, now everybody wants to do NPR instead. And they're trying too hard to do that too.

Think it's a couple things going on there, the novelty of PBR meant all the trend chasers flocked there at the inception, but now it's no longer
the hot new thing, we see it in every game and every show and it's just the new goto standard.

But because it's now recognized how demanding it is on the artist to make something appealing in PBR that stands out amongst the competition
you see the flight of a lot of people who can't pass the mark. The same crowd who very enthusiastically jumped into PBR because it was 'the quick way to make realism' now flee to NPR for the same reason 'the quick way to make appeal' but now under the guise of 'artistic reasons'.

I'm not trashing on either NPR or PBR, both are and will remain valid. Just pointing out how trendchasers are always running for what they think
is that low hanging fruit within their grasp. But thing is you need to be a good artist to make anything worth a damn whatever style you go for.

It's like how pixel art seem more accessible to newbies, and sure the barrier to entry is lower, but take what one think of when you're talking good pixel art from the peak of the era when legit artists was practicing it because it was at the cutting edge and all of a sudden pixel art is no longer easy at all.

Anonymous at Wed, 10 Apr 2024 13:25:22 UTC No. 980140

>>980135
>It's like how pixel art seem more accessible to newbies
I was thinking about Big Buck Bunny.
https://www.youtube.com/watch?v=YE7VzlLtp-4
As an hobbyist, Blender's Internal render is easy to use, really fast on modern hardware and that level of quality is attainable to me.
Let's assume modern PBR or NPR are 100x technically superior, that doesn't doesn't matter to me because there's too many many sliders I have to adjust in very specific ways to be able to get anything at all. And the $50 extra dollars in electricity per month because I have to use a video card that was mostly designed for mining crypto, not to do graphics. That's my main grief.

Anonymous at Wed, 10 Apr 2024 17:45:48 UTC No. 980157

>>980040
i like to put metallic in my painted surfaces to simulate automotive metallic paint

🗑️ Anonymous at Wed, 10 Apr 2024 18:13:27 UTC No. 980160

>>980127
>Metal atoms have an electron shield that reflect photons
that's not quite how plasmon surface resonance works but you've captured the spirit anon, repsect

Anonymous at Wed, 10 Apr 2024 23:03:23 UTC No. 980177

I put the sliders where it looks good

Anonymous at Thu, 11 Apr 2024 00:14:06 UTC No. 980184

>>980177
You can do whatever works for you but think about it for a second. What's the point of having "physically based" formulas if the parameters are chosen by feeling?
It's like if I gave you a physics simulator but I didn't tell you what the force of gravity on earth is and instead just told you to go outside and measure it yourself. First of all, who goes outside? I mean... Who does that? You understand?

Anonymous at Thu, 11 Apr 2024 02:47:58 UTC No. 980200

>>980184
What anon does is correct tho. Metallic paint has flakes of metal shavings suspended in in a coat with color pigment particles.
If you sample a pixel area of such a surface it will have both a metallic and a diffuse response at that point, what anon is doing is physically accurate.

>What's the point of having "physically based" formulas if the parameters are chosen by feeling?

If the model is correctly built you can't crank the settings anywhere such that it'll violate any physics.
A PBR model will conserve energy such that a pixel can't return more energy than what hit's it no matter how you drag the sliders.
Unless it's set to be emissive and would do just that in the real world as well.

No matter where you crank your spinners and sliders a physically plausible BRDF for a surface will be achieved.
If that surface has the correct settings for representing what you want it to display or not is another question but any combination of maps and settings
in the model are valid.

Maybe you made something more representative of a powder coated metal or a glossy ceramic where you intended to have PVC but it's not violating any physics.

Anonymous at Thu, 11 Apr 2024 04:29:01 UTC No. 980211

>>980200
>A PBR model will conserve energy such that a pixel can't return more energy than what hit's it no matter how you drag the sliders.
[Laughs in open domain albedo textures substantially above 1.00]

Anonymous at Thu, 11 Apr 2024 05:54:21 UTC No. 980215

>>980211
Run normalization on your color texture in case someone decides to stick a HDR-file in the diffuse/albedo slot.

I don't know what will happens in my shader if I do that, the thought to do this never crossed my mind. I guess it'll cause a nuclear runaway where the scene gets progressively whiter as I use ambient light probes that could amplify the light in a endless cycle.

Anonymous at Thu, 11 Apr 2024 06:12:06 UTC No. 980217

>>980215
Yeah I know I was just shitposting. But doing so does lead to some "interesting" results, particularly in Cycles. I was shining a spotlight onto a piece of Scene-Referred artwork I made in Krita, and the bounce off the bright parts of that image predictably led to the scatter of much more intense indirect light compared to a regular white surface. I also attempted to make a "super white" cube with a white that was 250x higher than display white and it kinda freaked Cycles out, leading it to slow down and hang.

Anonymous at Thu, 11 Apr 2024 10:33:52 UTC No. 980223

I _WILL_ violate the physics.

Anonymous at Sun, 14 Apr 2024 22:55:22 UTC No. 980593

>>980223
wtf anon NO

Anonymous at Tue, 7 May 2024 20:01:47 UTC No. 982730

>>980593
Yes.

Anonymous at Wed, 15 May 2024 16:05:16 UTC No. 983494

just do whatever works best for you

Anonymous at Thu, 16 May 2024 05:32:52 UTC No. 983538

What if I want my metallic surface to be a little bit tarnished?
Wouldn't it be a computationally cheap way to do that?
>Turn down metallic
>Have dirty looking texture
Or am I missing something?

Anonymous at Tue, 28 May 2024 21:54:24 UTC No. 984870

>>980115
The math makes no sense, and it's just people not understanding what it does.

The metallic value exists because they wanted two paths for both material models in the same shader. It's just a toggle.

You want an object halfway between metallic and dielectric, you make both materials and use a blend function or node, which is not mathematically or visually the same as putting the metallic value to 50%.

Anonymous at Wed, 29 May 2024 03:34:28 UTC No. 984893

>>984870
>which is not mathematically or visually the same as putting the metallic value to 50%.

But it is. Imagine you have a scenario where you have a metal that has a surface layer of some other material that has worn away, painted or powdered metal is a good example.

You can have pixel samples where either the paint is thin enough that the metal shines thru and the sample there for need to shade partially as dielectric and partially as conductor to capture the look of what is present in a real world reference that has a dual BRDF. Or what if a dielectric powder is covering some percentage of your pixel and the rest is bare metal, but such details are sub-pixel at your resolution. You end up with a sample that again has a dual BRDF response.

Any real world metallic surface that isn't freshly cleaned and spotless will never have a metallic value of 1. Pure metal objects that reside in the real world and
are covered in dust, finger prints, dried rainwater mineral deposits etc will have a 0.9x type metallic response over most of it's surface.

Thinking your metallic value is supposed to be used as a toggle if >realism< is the goal is retarded, because that is unrealistic.

Anonymous at Wed, 29 May 2024 03:48:52 UTC No. 984894

>>980129
>There are no known materials that are both ferromagnetic an transparent

Do you know why sunglasses mirror coatings looks like metals anon? It's because they are metals, good example of common material that is both metallic and transparent. When the metal layer is just atoms thick you can see thru it and it still shades like metal.

Stop thinking in terms of these lab sample thick monoblocks of material sitting in lab and think of it in terms of what sort of molecular matter the very surface layer of the thing you are depicting has, because that is what determines how it'll shade.

Anonymous at Thu, 30 May 2024 14:22:24 UTC No. 984988

>>984893
because it doesnt work like that
mixing two shaders =|= setting average parameters
imagine you have a shader that mixes a specular reflection with 0% roughness with a specular reflection at 50% roughness
the result would be a rough reflection with addition of some very sharp highlights
this result is not the same as simply having one single specular shader at 25% roughness

same concept applies to metal
metallic surfaces tint the reflected light with their color
if you take a grunge texture and make it metallic, you would tint the reflected light with the colors of the grunge texture which would be really inaccurate
the metallic material should only ever be a single base color to get accurate reflections
all grunge and texture detail should be mixed in as a completely non-metallic material

Anonymous at Thu, 30 May 2024 19:45:49 UTC No. 985007

>>984988
>if you take a grunge texture and make it metallic, you would tint the reflected light with the colors of the grunge texture which would be really inaccurate

Nothing is stopping you from writing your shader such that the metallic BRDF uses a different diffuseColor/albedo than your non metallic one so it's contribution is in the correct tint.

Saying you can't have a metallic value other than 1 or 0 is like saying you can't have worn paint sitting ontop of metals.

who came up with this weird ass claim? I've heard it parroted from several people over the years. Is there some foundational text out there that is teaching people wrong?

Eok at Thu, 30 May 2024 20:27:49 UTC No. 985010

>>980040
I can't care less about realism if it works

Anonymous at Fri, 31 May 2024 03:46:49 UTC No. 985058

>>985007
and the easiest way to do that is to just use a mix shader node
yes it would help to have a separate shader input that defines metallic color
do you know what else would be useful?
having a separate normal and roughness input for the "metallic component" so you can have effects like having brushed metal that is covered in a more regular grunge layer
oh wait
that would overcomplicate the shader and thats the whole point why we have a mix node in the first place

besides, nobody ever ACTUALLY said that your metallic surface cant be blended with a non metallic material. if you bothered to actually watch the in-depth tutorials they will tell you that IF you want to create a blended surface, you should mix two separate shaders that are at 0 and 1 metallic respectively, since it is not the same as having only one node set to a value inbetween

Anonymous at Fri, 31 May 2024 04:28:48 UTC No. 985062

>>985010
Another tripfag
Post your donut

Anonymous at Fri, 31 May 2024 21:30:55 UTC No. 985110

>>980127
>dialectic

Anonymous at Sat, 1 Jun 2024 02:38:05 UTC No. 985125

>>980127
>>985110
I think you guys mean diuretic