Should i be working in PBR from now on?
polycounter lvl 3
I am making things for my portfolio and am wondering if PBR is going to completely take over. I'd like to take advantage of Quixel and Substance Designer/Painter as well and know they can get amazing results, but I'm wondering how thoroughly its worth knowing the normal spec/gloss route. I hear gloss isn't even used sometimes.
I've made many things using specular maps, but not gloss. I feel comfortable using PRB in Substance Painter though. Am i missing something not working with gloss much though?
#2 I can afford Quixel btw and already own Substance Painter, am i crazy not to own Quixel already?
I've made many things using specular maps, but not gloss. I feel comfortable using PRB in Substance Painter though. Am i missing something not working with gloss much though?
#2 I can afford Quixel btw and already own Substance Painter, am i crazy not to own Quixel already?
Replies
Also, I believe Substance is far superior to Quixel. If you already own Substance, there's little reason to get Quixel too.
PBR doesn't refer to a specific set of map types/texture inputs, it's more of a holistic rendering system. With a PBR system you may have albedo/metalness/roughness, or you may have diffuse/specular/gloss. The differences are not actually very big.
Give these a read:
http://www.marmoset.co/toolbag/learn/pbr-theory
http://www.marmoset.co/toolbag/learn/pbr-practice
http://www.marmoset.co/toolbag/learn/pbr-conversion
Yeah, I would imagine the changes within an engine are fairly small unless they add something like image based lighting. One thing i noticed though is that the diffuse can look very different than its final result. Im used to painting the diffuse map by painting what i think the final result will look like, or at least more so. A dielectric blue metal for example look very different compared to its final look in the renderer.
Nearly all PBR systems use IBL.
There is no such thing as a "dielectric metal", those two properties are mutually exclusive. Raw metals are never dialectic or insulators. It's important to remember that you're representing the top layer of a material. So a metal with that is coated with plastic/rubber, painted, or oxidized (rusted) would be an insulator and will generally reflect light in the same manner as non-metallics.
The diffuse map for pure metals will typically be black, because metals do not diffuse light, they get all of their color from reflections. This isn't really exclusive to PBR systems, even with old school shaders this principal applies, it's just that many artists would give metallic materials both a bright diffuse and specular pass, which never looks quite right.
But in Quixel, you cannot rotate the texture because Photoshop doesn't allow rotation of the overlay texture effect thingy thing..
You have to give up some of the control you are used to, but you get some powerful tools in exchange and the biggest pro: You work on all the maps at once instead of copying everything over to make the different maps. It's also pretty cool to use the different noises in the height input to get some quick normal detail.
I personally hate quixel only for the photoshop plugin part, it's quite annoying sometimes. I still use nDo2 and 3Do quite often if I want to preview something directly out of photoshop. So yeah I don't regret having both.
https://www.allegorithmic.com/products/substance-painter
It sounds more like you're using substance designer to me.
Its Painter im talking about when i refer to the sort of disconnect and lack of control i feel vs manipulating things with photoshop. For me at least. I would really like to be moving textures around to create my specific details in various areas. And I'd like to create text and have it be effected by the existing wear masks without going to another program.
edit: You mean trying different things in general?
I have no idea why you wouldn't use the texturing apps available they are jaw dropping. Also I have noticed you can use presets as a start point and endlessly tweak if you need to. Being able to see stuff realtime is an incredible way to work. Also press the gogo button and 4 or 5 maps get updated in one go. Try doing that manually.
Yes. I believe that he is also refering to the fact that the more locked you are into a certain workflow the less compatible you will be with certain studios with very specific workflows.
Don't try finding an alpha and omega tool, its not meant to replace photoshop and you will get best results combining those tools. Procedual is at its best to do the rough 80% and have a great base, then add the important layer of polish per hand. It is possible, but dont expect to trump a combination by just using a single tool out of convenience. Its just about convenience right ?
An experience artist doesn't need a system that corrects his/her errors.
PBR was created not as a tool for artist convenience to make his/her life easier but rather as an automated solution to eliminate a necessity of experienced artist at all for the price of having higher resolution/repetitive textures to cover edge artifacts + more real time calculations.
Literally the way I produce content has not changed at all. I still make my spec/gloss the same and my color map is basically the same minus AO.
The metalness workflow is really the only "new" thing end even then its not that hard of a concept to grasp, especially if youre using something like DDO.
The purpose of PBR is to unify the pipeline which makes art asset creation and art direction/lighting more consistent and efficient. It's not there to rob artists of freedom, if you're feeling hugely restricted by it that's probably because you were doing illogical stuff that tends looks bad to begin with, like making both your specular and diffuse inputs bright for metals.
PBR has nothing to do with resolution, repetitive textures, edge artifacts or anything like that. Those are all technique specific issues that are not mutually exclusive or inclusive to PBR.
I've said this countless times, but the metalness workflow does not = PBR. You can do PBR without metalness, and you can do metalness without PBR. The metalness workflow is simply one method of mapping reflectivity, it has various pros/cons.
[ame]
Perhaps I am still missing something but specular veriant of PBR is not exactly physically based since it allows to make what you think is right rather than what the energy conservation dictates. To a some extent at least.
I don't feel hugely restricted actually, I just see no real advantages + extra head ache.
The only advantage is GGX highlight spot shape being slightly more anisotropic but I guess it's nothing related to locked nature of the shader.
ps. What makes PBR materials looking better is a bit of environment reflecting/color influence. Looks like evrybody think it's something only possible with PBR letters while in fact you could do the same in unlocked shader too and have a lot more control over it, especially when you couldn't allow true dynamic reflections and huge amount of probes.
For instance, you can use a bright albedo value and a partial (gray) metallic value and break the metalness workflow in the same way you can break the specular workflow. Again, these are simply texture input methods, and using specular/metalness does not mean your workflow is physically accurate or is not physically accurate.
Edge artifacts are specific to the metalness workflow and have nothing to do with PBR. Complaining about issues with the metalness workflow and attributing them to PBR is like saying "I hate beer" because Miller Lite is awful or "I hate crayons" because you don't like the color blue. These things are not intrinsically tied together.
The substance suite can absolutely replace Photoshop completely - I've been using it for a few months now both at work and at home and you can most definitely texture things from start to finish within the suite. This is especially true if you're using an engine that can read SBS files directly (UE4 and Unity5 both do this).
There's few major studios who have been using it extensively for texturing.
Here is a great GDC talk that shows the kind of pipeline that integrates substance designer into it:
http://www.gamespot.com/videos/remaking-the-art-of-halo-2-for-xbox-one-gdc-2015/2300-6423667/
It's what I am doing for dirty dust covered metals to avoid artifacts . But every time I wonder why I have to invent ways to deceive that inflexible PBR system and can't just do what I think is right the way I think is most suitable ? Why I should fight with things reflecting something from too distant and too wrong probe and so on.
And regarding specular PBR. Any too noisy and contrast roughness values produce halos too, in our specular PBR variant at least so I also have to waste my time finding right input combination.
Things like the wrong probe being used is really an engine dependent problem. It's more of a failing of the system being used to implement PBR rather than a sign that PBR itself is bad.
Yep, that's the crux of the matter. All of these issues are implementation specific and have nothing to do with PBR per say.
And we still have a lot of such limitations in often very faked environment things.
I don't understand how anything inflexible could be a good thing while every 3d soft tends to be as much flexible as possible.
For assets unification purposes a few material samples are quite enough. Wy we extra need some hard codded limitations?
http://readyatdawn.com/presentations/
http://blog.selfshadow.com/publications/s2015-shading-course/
gnoop, the point of PBR is to represent reality better than previous system and allow assets to behave well in different real-world light conditions (proper HDR...). It's widely adopted in film and games as it helps to drive costs down (less per-scenery tweaking) and improve quality as EarthQuake has been mentioning.
I cannot follow your flexible/inflexible reasoning, given all real-time shading systems have certain limitations.
It is not a "trend", the term has been around for a while, just look into the popular "Physically Based Rendering" book, which was published in first edition 2004 http://www.pbrt.org/ We simply can afford doing more of it today.
I wasn't aware it destroyed all NPR or otherwise artistic possibilities in engines? Surely, it's still possible to stick any given custom shader into the currently availeble engines despite them having PBR as the default?
I wouldn't be worried until that is the case?
You don't have to tell me that more realism in art/graphics is not going to make things better all by itself.
This is an art content/art direction issue and has nothing to do with PBR or any technical aspect of rendering. Whether you make boring or interesting content is entirely up to you.
Yeah, it's what I heard many times. Represent reality better, less tweaking and improve quality. But in reality you may tweak them to death, fighting with rigid system not allowing to compensate some tiny aspect not exactly 100% physically correct in your engine/ PBR implementation. I understand such approach for unbiased offline renderer but in games it just ties your hands even more.
Having a bit of environment reflecting in every material is basically what makes PBR looking more real. But you could do the same easily with old shaders without all those collateral edge/halo artifacts.
When I say inflexible I also mean the fact that now we have to ask coder every time we need some small fix, wait a week and then adjust our work to new version of physical correctness.
These issues don't occur in Unity or Ue4. It's not PBR that's inflexible. It's the engine you're using. I've never seen the halo effect you're talking about and I've used all kinds of PBR enabled engines.
I have never tried Unity but in U4 I see all the same problems , not as severe as in our engine although.
I've done a bunch of different tasks with PBR. That sounds like a problem with the texture filtering or anti-aliasing. It shouldn't matter what texel size your texture is. It's not related to PBR unless they've poorly implemented the shader somehow.
You won't get the problem with the specular workflow, however, you will the opposite problem (dark edges) but much less pronounced (to the point that it's irrelevant). This again isn't related to PBR, you'll get it with any shader that uses IBL blurring for gloss/roughness levels.
Both problems have little to do with PBR but rather are caused by texture filtering. The problem is that texture filtering blurs the texture so that it won't look pixelated, but this means you're averaging the metalness and gloss/roughness values along edges.
Obviously turning off filtering isn't a solution. The only real way to solve this would be to go all the way back to an IBL shader that didn't blur the reflectivity content to match the gloss, which looked awful back in the day when we did that.
At this point if you're having some consistent issue with your PBR pipeline it would be more productive to post images of the problem so someone can give suggestions, rather than continue to blame it all on PBR or perceived (misconstrued) restrictions.
I have no huge problems with PBR workflow actually I just wonder why everybody seem so exited when I personally find it more restricting than helping.
What you call "specular workflow" is very similar to our old shader system. We started to use both specular and glossiness + Fresnel reflections decade ago and nobody called it PBR.
So yes, I call PBR typical metalness approach. not specular one. Our current specular workflow is basically metalness one with added extra speclevel input.
The metalness workflow in practice means severe quality decrease imo and thus more repetitive textures.
In theory you should save on textures with metalness PBR, in practice it turns pretty opposite.
I agree to some extent about issues with the metalness workflow, I don't think it's a severe quality decrease but I do personally prefer the specular workflow with a full color reflectivity input, which is no more or less physically accurate than other methods.
you do? you dont need a full rgb texture for specular. therefore you can have a multi mask and an albedo map for one texture set.
I meant that once I have to use metalness I know I either have to twice the texture size or make the texture twice more repetitive.
As of calling specular approach another PBR variant, I don't understand. Within that approach you can easily modulate the intensity of highlight by hand, breaking forced energy conservation and choosing what should be on roughness/gloss channel and what details go to speclevel.
It's a flexible approach and extremely helpful for a huge amount of materials where using only roughness is impossible. Basically every material with too strong pixel to pixel roughness contrast.
As a general rule this is simply not true or necessary. Countless artists and games use the metalness workflow and do not need to double the texture size to do so. There may be some very small range of specific situations where having extra texture res helps, but again, as a general statement this makes no sense.
If this is such a widespread problem, it would be very easy to provide visual evidence so we can understand the context of the problem. Even if you can't show work due to NDA, if such a glaring deficiency exists, surely you could make an example with 15 minutes of work.
You do not understand, that much is clear. What isn't clear is why, this has been explained in great detail in this thread and others and you've been provided with lots of reference material. I can only assume you haven't bothered to read any of it.
In addition to Christoph's (crazybutcher) links, I'm sure you've seen these already but I will link them again:
http://www.marmoset.co/toolbag/learn/pbr-theory
http://www.marmoset.co/toolbag/learn/pbr-practice
http://www.marmoset.co/toolbag/learn/pbr-conversion
Here's an article by S
When you have rendering engineers and technical artists posting here telling you that what you're saying is wrong, and you can't post a reference to a single article that agrees with your conclusions, yes, the burden of proof is on you. I do not consider "because I said so" to be a compelling argument.
This has already been covered, but I will say it again:
With both specular and metalness PBR workflows you can create results that are not physically accurate. Potential to create inaccurate results does not mean the entire workflow is inaccurate, it's only inaccurate if you don't know what you're doing or purposely break conventions, crucially, with both workflows.
From the same article.
I don't say PBR approach inaccurate.
I say it's inconvenient at first and creating false impression that once you measure and calibrate everything you couldn't do any better . Even if your result looks somehow not that great and artifact filled .
I can do same quality materials. PBR or not. I just don't see where it is helpful
And sometimes I have to invent illogical tricks like doing asphalt cracks
metallic to kill all possible highlight there or make secondary Fresnel to do some diffuse texture variation not exactly Schlick based.