[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k] [s4s] [vip] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Home]
Board
Settings Home
/3/ - 3DCG



Thread archived.
You cannot reply anymore.



So there's these charts floating around that claim to have the correct values for physically realistic results when using PBR workflow for something like Unreal Engine 4. But all the artist work or included samples that come with the engine don't adhere to any fucking standards at all. Their textures appear to have completely random values for luminosity etc. that don't follow these charts at all.

For example this chart claims the mean luminosity of the albedo map for painted metal is around 94, but it's fucking impossible to make any red or yellow painted metal while keeping the medium at 94 because it looks ultra dark and de-saturated. Am I a retard and doing this shit incorrectly? It's confusing as hell. I have no idea how bright to make any of my textures, I thought PBR was supposed to remove guess work but all I've been doing is guessing how bright or dark to make my fucking albedo maps.
>>
>marmoset
>>
>>561739
>Their textures appear to have completely random values for luminosity
This
>>
Do the values you've entered look like the material you're trying to emulate? If yes, job done. If no, tweak until satisfied.
make your albedo around 50% luminosity and multiply it with a float in your shader for easy tweaking.
>>
>that don't follow these charts at all.

that's why texture artists create the base shaders and some of the environment artists tweak them to create variation.
a base material still have the same values but often different patterns
>>
nah
>>
>>561739
>any game engine
>physically based
I want this meme to die, along with the metalness map part, it just makes everything look the same, not to mention those nasty artifacts on metal/non-metal border. Also these reflectivity values are bullshit, most non-metal dont have colored reflection
>>
>>561751
it might not be accurate but that's thing we have so far.
>>
>>561752
you can always use specular
>>
>>561746
this

do you think anyone is actually going to notice that your copper is reflecting 12.4281479% of blue light instead of 12.4147819% of blue light?
fuck no.

if it looks good, then its fine. use references if you want it to be realistic.

this idea of "one value fits all" is retarded
>>
>>561739

No these materials were scanned by quixel and these were the colors the scanned data gave in the environment the data was captured.

In marmoset high gloss is white and in unreal it's black. So anything you create in Substance Painter or DDO looks completely fucking different in every other engine. Pure annoying.
>>
>>561739
>painted metal
u wot mate
it there is paint on the metal, the material is no longer metal. and it will depend on the paint, there are all kinds of paint, some glossy, some matte. smdh.
>>
>>561739
> Metalness values neither 0 or 1 but in between for every material

I seriously hope Marmoset doesn't do this
>>
>>561799
>using metalness map
I seriously hope you dont do this, unless you have no other options
>>
>>561809
I don't work with shaders a whole lot, but It was my understanding that a metalness map influences the reflective scatter, does it not? i.e how rough vs smooth the surface looks. Why wouldn't you use that?
>>
File: image.png (743 KB, 1063x354)
743 KB
743 KB PNG
I came here to moan about this as a separate thing but this thread seems as somewhat relevant to the incoming whine-fest from me....

What's the fucking point? See attached image - this is part of a floor texture I've been working on. What you see here is literally just the diffuse map I painted in 'shop, plugged into a VRay material that I just ran out as a test before creating all the other maps in SP...

(Continues...)
>>
File: image.png (771 KB, 1069x365)
771 KB
771 KB PNG
>>561823

And here it is again after painting in a load of height detail and glossiness to represent wetness on the floor and yadda yadda yadda... (Ignore the lower two steps - I hadn't gotten to them yet)

There's an AO map multiplied onto the diffuse, a reflect map, a reflect gloss, an IOR map, a normal map blended with a height map, and a roughness map....

For what? It takes 10 times longer to render and it looks absolutely no better.

Where am I going wrong? I just don't know where half the effort I put in actually goes sometimes...
>>
>>561824

And for the record - yeah it looked pretty good in SP. I just cannot get this shit I do in SP to work back in VRay....
>>
>>561826
If you want to compare you render in Vray with your render in Substance it would help tremendously if you use the same HDRI image as illumination/reflection sphere.

Reflect map, reflect gloss IOR map and a roughness map? You only need the roughness map. For such a flat surface you use height AND normal?
Doesn't seem overkill to you?

So far your scene doesn't provide anything which would make these effects stand out. IS there something to reflect?
If you don't have a light at grazing angles you won't see fresnel effect. There is no direct light reflection so you don't see the roughness variation.

That beeing said, the reflections in an realtime engine will always look different than the reflections in an traditional render. At first sight those in engine might look sharper and stronger, but the reflections in Vray will look more realistic if done right.
>>
>>561833

Hey thanks for reply....

i'm not using an HDRI in the Max scene - just a JPG in a dome to add a little colour to the GI. Majority of illumination comes from IES lights nearby.

I have an approximately similar HDRI in SP though but trouble is - SP dumps your model in the middle of the sphere - this model, for obvious reasons, needs to be way down low - I can't figure out a way round that. Suggestions?

All those maps is crazy - sure, but those are the maps it spits out, so after a good bit of tinkering with different combinations - that was the best result I could get.
Trust me - roughness alone did not work. It looked shit. In fact - taking out any map *but* the roughness map made it look terrible. The roughness actually seems to be doing very little.

Yeah - I'll concede the heightmap + normal is overkill, but still - it is giving a more detailed result than either one of those maps by itself (not by much, I admit...)

There are a number of ies lights around the scene which I was hoping the puddles would pick up at least a tiny amount. There's also a lot of wetness over the sidewalk in general - again I was hoping this would pick up at least *some* direct reflections of the neon tubes around the scene....

Nada.
>>
>>561833

Also - you are talking straight knowledge about grazing angles and fresnel effect etc. etc. That hasn't gone unnoticed.

It's a rare pleasure to talk to someone on here who actually knows what the merry fuck he's talking about for once.

Cheers.
>>
>>561812

Hey I'm not the guy you're talking to, but I can hopefully lend a little bit of knowledge on this...

'Metalness' is distinct from reflectiveness / roughness. Real world electricity conducting materials (ie. metals) have very very low amounts of diffusion - practically 0% in most cases (gold for example).
So this means that nearly 100% of the light that we see when we look at metals is direct reflection. This raises the question - how does metal appear to have any colour to it? Why does gold appear 'gold' and not just a mirror of the colours in space around it?

This is because metals, unlike almost all other substances in the known world, actually colourize the direct reflections they send back.

So metalness is not simply influencing the amount of reflected light, but also it's colour, which - in PBR engines - is derived from the diffuse input.

('diffuse' in this instance is something of a misnomer, as there is *no* diffusion - this is why, I guess, it is referred to as 'albedo' in PBR engines).

A lot of what I've said here is based on things i have literally only just read about, and have only half digested - so I may have some stuff wrong. So don't bother flaming me if I have.

There are two really essential articles on the Marmoset site that anyone interested in PBR shading should read:

https://www.marmoset.co/posts/basic-theory-of-physically-based-rendering/

https://www.marmoset.co/posts/physically-based-rendering-and-you-can-too/
>>
>>561812
metalness maps are only used to tell the engine which part of a surface is metallic and which isn't. the value is either 0 (non-metallic a.k.a dielectric) or 1 (metal). in some cases you can use values in between 0 and 1 for example when you have a piece of dirt covering a metal surface.
if you made a sphere of gold, the metalness map would be pure white. if you made a sphere of gold sitting on a plastic cup, you would make the UV's for the sphere 100% white (metal) and for the cup 100% black (non-metal, it's plastic)
it's important to remember that only the upper most surface is whats counted. if you made an object that has painted metal, your metallic value will actually be ZERO because the paint is the upper most layer of the material, NOT the metal. it's the same with something like a beverage can. it's technically made out of aluminium, however the outer most layer is covered in a non-metallic surface which means it is NOT metallic. only bare exposed metallic surfaces should be configured as a metallic surface.

roughness / smoothness are completely different and handled on a different texture map (roughness map if you're in UE4, some programs call it a gloss map or smoothness)

here's an example of a sphere where on the left it is 100% non-metal and on the right it's 100% metal. other than that they use the exact same texture information:
https://cdn2.unrealengine.com/blog/metalness01-600x300-694626521.png

here's an example of a non-metal sphere with very low roughness on the left, average roughness in the middle and very high roughness on the right:
https://cdn2.unrealengine.com/blog/physicshadingheader-770x250-996585692.png
>>
>>561947

Anon is 100% Correct.

People are randomly using metalness values as an addition to roughness values, on all sorts of objects which are not supposed to be metallic - presumably thinking that metalness is the opposite of roughness.

It isn't. It's a parameter very specifically intended for electricity-conducting materials. If you are making a shiny plastic object, it's metalness value should be set to 0, regardless of how shiny it is - it isn't metal.

The clue's in the name kids.




Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.