You need to know this about Specularity in Blender 4.0
ฝัง
- เผยแพร่เมื่อ 5 ก.ค. 2024
- NOTE: I've produced an updated video dealing with specularity in 4.0 that is newer. This video was produced with a beta of 4.0 that didn't contain updated user interface. Watch here: • What is the new 'IOR L...
The upcoming Blender 4.0 currently in development has made an important change to how the Specular function in the Principled BSDF operates. This video talks about that and how to use this important change.
00:00 Introduction
00:17 Basics of Specularity
00:43 Index of Refraction
01:33 The basic change made
03:11 Specular Neutral
03:18 Specular & bitmaps
05:07 Specular & Dialectrics
07:03 Creating diffuse-only surface (important!)
#specular #blendertutorial #fresnel #refraction #ior - ภาพยนตร์และแอนิเมชัน
This is the kind of channel I've been hoping for since forever
You deserve a lot more recognition. I wish you and your channel the best. The content you are making helps people like me alot. :)
Hey Christopher!
As always big thanks for sharing your knowledge!
Beautifully done. This is the kind of explanatory, authoritative, straightforward user manual /reference material that you would expect to be published by Blender themselves. If they do, you should link to it. If they don't, they should link to you. Or they should hire you, if they don't already....😊
I love your content, hope your channel blows up and get a ton of subscribers !
Your videos are so frickin' good. Thanks for making them.
Brilliantly informative video, thank you!
Very informative! Great video!👍🏻
A 3d purist? Exactly what I need as the Blender information source. Instant subscribe!
This is important info. Thank you!
Thank you for the update!
Excellent explanation, thank you.
Superb Technical Explanation.........♥♥♥
Excellent video
Thx for this video, good to know.
Feels like there should be a checkbox with a separate ior for fresnel, similar to how modo does it
THE Christopher Tyler of Strata 3D fame. Welcome to Blender!!!!
Thank you
Maybe they should call Specular Override and put it after the IOR, so you can have a clue its overlapping something.
Bingo.
No, I disagree as that will obfuscate the backward compatibility aspect of this. Leaving it as is makes it so users who are used to how 3.6 handles specular will be able to continue using it that provided the IOR hasn't been changed. If you changed this to some foreign sounding name it'll just cause confusion in my opinion.
@@christopher3d475 You're right! I didnt consider that!
@@christopher3d475 But wouldn't they notice the difference, if they used the same IOR Value (and the same specular value), since the IOR Value now directly influences the specular reflection of the Object? It could mean that the visual look of the specular reflection on the surface is now either higher or lower even though it uses the same specular value as used in Blender 3.6. So, If 4.0 users want to have the same stylistic look of the object as in Blender 3.6, they would either have to increase or decrease the specular value in 4.0 (meaning they would have to make a subjective adjustement - trusting their eyes, so that it looks the same as before)?!
The way I understand it (based on how you explained it): The visual look (glossiness of the surface) of the scene would only be the same (in 3.6. and 4.0), if the users have set the specular value to the neutral position of 0.5. In both cases the glossiness of the surface would look the same, because the IOR setting alone defines how glossy the surface looks like?!
Yeah Vray had this, then people complaint and it got a seperate IOR for reflection. Making own shaders and physical correction should be seperated.
For example: if you set transparency to 100% and IOR to 1, the reflections are gone. Makes it hard for thin glass. In Octane ans Redshift you have a thin glass option for that.
Yeah Cycles needs a proper thin glass/architectural glass function.
I don't think this will add anything to "thin glass" functionality. One sided geometry is a whole different beast to handle. But this is easy enough to setup manually:
Layer Weight/Facing -> Power 5 -> add 0.05, to mix between Transparency/Refraction and Glossy node.
If using Refraction to get access to roughness, you need Geometry/Incoming as its normal.
You *can* use fresnel, but you need to account for inverted normals in order to prevent Snells Window for the backfacing face. The above is simpler.
And then there is the question of how to best handle opaque shadows if using refraction, particularly if using roughness. Yeah, a different and complex beast.
Did I know I needed to know it? No. Do I now know I needed to know it? 100% yes!
Thanks for this. I feel like this is at a broken state with the current alpha version. Eevee Reflections do not react the same as Cycles Reflections. With time it'll get fixed. I guess this is a pretty recent change.
Hey do you share sample scenes by any chance? great video!
a weird penomenon I have noticed is that even if you get a nice specular, as soon as you zoom out the camera it can easily disappear, leaving just the diffuse look
How would it influent on already made materials? Do we have to change the ior/specular valuse to make it work as natural? Will the new update change the look of old materials?
It'll depend on how you've set up your old materials. For transmissive materials, nothing changes. You plug in the IOR and call it a day. But for opaque materials, it depends what you've got plugged into the specular slider. If you've left it at the default 0.5, then nothing much changes either. You can go find the Index of Reflection/Refraction (same thing, all surfaces have one) if you want and plug that into the IOR, or just leave it at the default 1.52. But if you've adjusted the specular slider, then yes, it'll probably look different and your best bet would be to just put it back to 0.5 and find the correct IOR. If you've used the formula given for the old specular in the docs, you should just delete that and input the IOR value instead further down.
If you open up an existing material, it should look the same. The one exception to look out for is diffuse only materials. Currently if you open a file that has a material with a Specular value of zero, the new system will apply specular reflections on that surface because the default IOR is 1.52. So you'll need to change that to 1.0. I don't know if they plan on putting in some sort of intelligent adjustment for this situation, but it's something to watch out for.
I just wanted to know how to apply a specular map to a texture. The specular node is gone.
Please make a new video about this topic but with the final principled node in 4.0 with some examples too. By the way great content
I'm planning on doing just that. I actually recorded some material that I was going to include in my latest video on SSS, but didn't include it because the video would have been too long. So I'll do a separate video on it soon.
I was aware of this already from brief testing, but this breaks *ALL* my projects. I use specular 0 for shadowgaps all the time (no specular whatsoever at any angle), now I have to edit those materials to also modulate IOR? And to achieve the same look (not that's it's *that* important to me) from "modulating specular", incl all the way down to 0), how would we even go about doing it? The specular has a certain function to it to linearize it for artistic purposes and a reasonable range, whereas the IOR is a direct control with a nonlinear response (1.01->1.11 is a hell of a lot more sensitive than 2.01->2.11 which is barely a visible change).
You don't have to modulate the IOR if you don't want to. You can continue to use Specular the way you were using before. Just leave the IOR alone. And yes, I agree the Spec function linearizes the low range values for the fresnel curve's F0 end. This is why it's still available if you like using it that way. But the inverse situation is also addressed by this new approach, people coming from other renderers that only use IOR for specular control can now do this in a consistent way with those other apps.
glad i watched this, as someone who’s a generalist, it’s the kind of thing i would never have figured out on my own in a hundred years. Thanks for the helpful video! But honestly, sometimes I wonder what they’re thinking… “0.5 is the unmodulated default”… who on earth would ever find that intuitive? Blender is the most difficult software I’ve ever tried to learn (and I’ve learned a bunch of them). sorry just venting a little I guess, sometimes I wonder if switching was a mistake.
Yeah, it's to maintain backward compatibility with older files. you can open an 3.6 or earlier file and have it match. But you're right 0.5 doesn't correspond to anything physical which is the problem. But at least in 4.0 you'll be able to just leave it as is and use the IOR.
💣
Have they fixed the issues with specularity when using roughness and bump maps? Or the bump map issues in general?
There is a note about improvements to normal mapping and I've been meaning to investigate this. 3.6 and before definitely have some weirdness bump maps.
@@christopher3d475 recenently someone posted on twitter way to fix bump with strength larger than 1. It seems to me that in newest 4.0 its still broken
Tell me please
what PC configuration is needed to complete this tutorial?
I'm sorry I'm new to blender.
My English is not very good and I can't fully understand what you're saying.
Does the parameter "IOR" in this node mean index of refraction or reflection?
I don't know.
But I use the fresnel node to control the reflection and it works fine.
The parameter "specular tint" also produces a Fresnel effect, but it's very weak.
I used the parameter "IOR" to control the refractive index of the transparent material and it worked as expected.
Yes, IOR = Index of Refraction. It works the same in other nodes too. In fact it works the same in all 3D applications. It's a physical parameter.
Yeah, this is why I never really liked rendering in blender. Even tho the workflow is amazing. to preview a noise in the viewport directly can be super useful. Maya has something almost as good, but unfortunately I've yet to see a way to do it in 3dsMax.
Any software is somehow better or worse than others. I switched to Blender from Cinema 4D just for the modeling and real-time rendering tools, because. Blender does it better. But creating materials in Blender and working with them is much worse than in C4D. And your video is one of many that confirm my words. Blender is very difficult for me, but the transition back to C4D will be even more difficult. I just have to be patient and do what I love. Thanks for the video!
diffuse and reflections look like missing energy for me! If i put a little bit of roughness, no depht in front and with the angle there are so reflexions and white coat! In others engines with the same setting i have a "correct" render.
Watch my video called The Furnace Test where I talk about energy conservation and preservation in Blender 4.0.
now we can set high ior (+-50) values and get a metallic surface
While you can do that, Blender isn't really set up to properly do metals with this approach. Still better to use Metal mode.
If anything, I'd hope a flexibility to work on 2 different PC, one for Blender, the other for Unreal Engine 5
Whatever you do in Blender, that same object also changes in Unreal Engine 5.
And please don't ask me to: Why not just build a PC dedicated for that? Well, limited budget and affordability
whihh version does this as the 4.0 alpha version seems no different?
If you get the most recent alpha version, it should have this. I'm testing out the IOR and spec while also typing this, lol.
@@Ironpants57 yeah I got it now. that's really useful, same with the improvement on the sheen node with sheen roughness
That was not a blender thing, but a specific design goal of the disney shader, to make it more intuitive instead of making you look up tables for IOR values. The disney (what the principled shader is) shader was meant for cartoons and never was intended to be perfectly realistic.
Perhaps, but the world is moving on towards more physically correct and the way the Blender devs have made is that you can do both in 4.0, 'artist easy' or more technically correct.
@@christopher3d475 what an idiotic thing to say
sir idon't agreed with new bsdf it is not backward compatible
Sometimes you have to move forward.
Why not just increase the roughness?
Because then you're rendering two scattering layers when you only need one.
In a time where we are going to Physical Materials where most of this stuff is done under the hood and material/render looks realistic and natural "by default", blender is entering an era of an unnecessary complexity and confusion that Vray just left a few years ago...
This change is driven by needing to retain a kind of backward compatibility with older materials while also moving towards doing things more correctly. So it can seem unnecessarily complex.
@@christopher3d475 I wish they'd implement an equivalent to V-Ray materials and V-Ray Physical Camera. I understand how it'll suck to break existing libraries so, keep the current system for a limited time to allow transition. To me, a more technical approach will be best for everyone eventually. Better results and knowledge.
@@christopher3d475 I just checked in Lukas' presentation at the Blender Conference (The new Principled BSDF model in Cycles) and as far as I can see, the issue at the moment seems to be that the UI and names haven't been updated and this is quite confusing of course. Not sure if I can post a link to the video here, but you find a reasonably good overview image at 52:07. To me, the structuring alone makes a lot of sense and would make it less confusing.
th-cam.com/video/DQeP363Xmn4/w-d-xo.htmlsi=3IVJeoST3XEx6YIj&t=3127
Well, that video was done nearly a year ago. At the time they were considering putting in a new Principled V2 shader and keeping the existing one in place for backward compatibility, but it was decided that would be more of a pain to do and to just update the existing principled shader which is a direction I agree with. So I don't know that we can look back at that video as an exact roadmap for the current implementation. During one of the builds of 3.5 that had early versions of PV2, Specular was gone altogether, relying only on the IOR. I didn't think that was a good idea and voiced that concern. I'm glad they put it back in and I think this new behavior is a good balance between old and new behavior. I do think that we might see some UI updates to the Principled shader which might include some names changes, or perhaps a new panels UI function. But I think how they have it now makes sense as far as forward looking and backward compatibility. @@deepblender
Do I really? Do I really need to know this? I'm still struggling to do bevels.
I just really don't get it. Not much of any visual results explained.
Does driving up the IOR real high turn the object metallic now like in Redshift?..
Using a very high IOR for metallic mode isn't really supported by Blender/Cycles. You can do it, but Blender isn't quite set up to control a metal appearance correctly when doing this. For instance, if you set the IOR to say 70, you'll get a metallic appearance of a basic metal, but you can't change the tint and edge reflectance isn't correct. So switch over to Metallic mode for metals. This means that the usable range for IOR values is going to be from 1.2 to about 2.5, with most uses for non-dialectrics between 1.2 and 1.7.
There are many non-intuitive things in the Blender program. You are looking for the transparency parameter in the materials, you can't find it. Instead you find transmission and alpha. Which one to choose?
Or when you work with opaque materials, what is the meaning of the IOR parameter because opaque materials do not have such a thing. IOR is the ratio between the speed of light in vacuum and the speed of light through the material. If the material is opaque, then IOR becomes infinite.
I agree transparency is more intuitive than transmission. However, IOR is well known in physics as also applying to non-refractive objects and all modern 3D renderers use this convention. So what Blender/Cycles is now doing in 4.0 is consistent with everyone else. The IOR not only specifies how light behaves, or refracts, through an object, but how part of the light reflects off the surface. That's what I touch on in the video. That reflect part is the same on non-refractive opaque surfaces.
What Cycles isn't using is a parameter called an Extinction Coefficient that is used for non-refractive objects. Some renderers use this in conjunction with non-refractive materials, but it just complicates things.
I can't understand why index of REFRACTION should have anything to do with specular REFLECTION.
It's a physics thing.
@@christopher3d475 A physics thing you can explain, or point to an explanation?
When an glass object encounters light, most of the light refracts/transmits through the glass. But not all of it. A percentage reflects off the object. The IOR not only determines how light slows down and thus bends through the dialectric material, but also how much of it reflects off that object. It turns out that the term 'index of refraction' behaves the same way on non-transparent objects. Meaning it describes the exact same reflective behavior. Let's say you have a black marble with a perfectly polished exterior. MOST of the light is absorbed into the material, but a percentage reflects off it. That reflection behavior for the glass object and the black shinny marble is identical. For some objects physics defines two terms, a complex ior and an extinction coefficient that defines the termination of light as its absorbed into the non-dialectric object (this is mostly used for metals). Some renderers actually support this feature. But for what we need to know, is that the IOR defines the reflective behavior for both dialectric (glass for example) and non-metallic non-dialectrics (think plastics, or the black shinny marble). Blender now uses a separate IOR value for coatings for instance because the coat reflects a certain amount of light and the rest is refracted through to the material component below. We just don't see a lot of refraction because the coating is very very thin and it's handled procedurally.
So, moral of the story, don't let the term refraction confuse you, it functions for non-transparent objects also as far as glossy reflections go.@@mdoerkse
@@christopher3d475 Thanks for the detailed response. So you are saying that if light shines on a polished black stone, or black glass, or black plastic, it will have different specular characteristics due to IOR even though the light doesn't travel through the material? The light somehow knows what the IOR would be if it did pass through the material? I thought that specular reflections only care about the smoothness of the surface.
The IOR is a description of physical properties. However, when it comes to non-refractive object we have a bit more creative control to adjust it to create the look we want as far as reflection intensity goes. For dialectrics, they were defined a bit more rigidly because the IOR is also describing the behavior of the light through the object which is very precisely measured. For instance, diamond is much denser than common glass so it has a higher IOR value of 2.41, whereas glass typically is around 1.52. That 2.41 value means a diamond is quite hard, and slowing down light more than glass, but it's also actually going to refract less light because more of it is going to reflect off it. the higher the IOR is the more light ALSO reflect off it. So that's what an IOR is doing, it's defining how dense a dialectric is, AND as part of that property how much is therefore reflecting off it. It's one of the reasons diamonds are perceived as being really sparkly.
So, if you have 3 objects that use the same IOR, say the most common 1.5, even if they aren't refractive, they're going to have the same reflective behavior. Because the IOR also defines reflection intensity.
For instance, if we use common glass and its IOR of 1.52, that means that at a face on angle (looking right at the middle of a glass sphere for instance), it is going to reflection 4.25% of the light at that point and transmitt/refract 95.75 of the light at that point.
But if we take water, which has an IOR of 1.33, and we use that imaginary ball, the face on point in the middle would reflect just 2%, and transmit/refract 98%.
For a diamond with its IOR of 2.41, its F0 (face on) will be just over 17%, with 83% being refracted. That's why it looks sparkly to us.
So if we just take away the refraction part and use the IOR, you can see that we can raise or lower the intensity of reflections on an object. Many types of materials have had the IOR values determined in the laboratory, but for the most part you can just use a stock 1.5 value and adjust it up or down a bit to increase or decrease reflections on non-dialectric surfaces. @@mdoerkse
The upcoming Blender 4.0 IS CURRENTLY SILL IN DEVELOPMENT. Why should I care about it NOW?!
A lot of folks are early adopters, want to know what's changing, preparing yourself for future releases, education about how Blender currently works. Any number of reasons. But if you're not interested, then it's not something you need to know.
This is the problem with having non-physical parameters lying around because of feature rotting. The "Specular" controller should be removed as specularity is handled by the conjunct action of IOR and Roughness
While technically it's correct to only control reflections/specularity via IOR, the Specular channel in Blender gives additional control via bitmaps as an example. It also linearizes the low end range of the Fresnel curve giving more fine tuned control over F0 reflections. But if you don't want to touch those, you can use IOR entirely and ignore specular. Just leave it at its default value.
@@christopher3d475 while I agree that even in a PBR shader it's good to give artist ways to break physical laws for the sake of artistic liberty, i think that at the very least the parameters should be arranged in two distinct groups, physical and non-physical. the way it is now is just confusing
10 years ago we forgot about this fakie pipeline. All now use IOR and reflection is ON always 100%. Blender 4 - specular 😅😂
Where white balance?
Where usefull light mix?
Distance maps, layered shaders? Common shader nod editor with instant?
'where' is the cost of Blender compared to those other applications? If those other applications with such features are working for you, keep paying to use them.
@@christopher3d475 Dude this is the worst excuse ever. what I described does not require any budgets. It's a matter of logic. There is where visualization is heading. This is Pathracing a mixture of biased and unbiased. Also, everyone has long since switched to electric-dielectric logic. as well as control of reflections using IOR.
For example, in its recent updates, the Corona has completely removed such a parameter as reflection. The world has come to this. Because everything has reflections and "everything has Fresnel". as the famous meme says. But obviously Blender doesn’t know about it.
It is these problems that prevent great professionals from abandoning outdated programs and finally switching to blender. And if they move there, it will be much better for all of you because... They will bring you 20 years of experience each.
Blender IS moving in the direction of greater physically correct surfaces. If you've been paying attention, you'll have noticed that 4.0 now has proper energy management at the material level. That's a WAY bigger problem to have been solved than removing specular.
As 4.0 stands, if you want to completely ignore the specular channel, you can do that. Ignore it, put tape over the monitor so you don't see it. Now you can work with it just like you do Corona, just IOR and roughness, you couldn't do this with IOR in 3.6.
The developers also had to consider backward compatibility and the balancing point here in what they've done in 4.0 was to leave Spec there for the time being.
I'm just the messenger. Go over to devtalk.blender.org and lecture them, they're the ones who actually code and make development decisions. You'll find that there are many interesting discussion the devs have with people about future directions. @@speltospel
@@christopher3d475 The Specular parameter is fake. from that era of 3D when no one thought about physical correctness. I remember the times of Brazil and early Vray renders. There was a Fresnel option, but no one used it. Everyone adjusted the level of reflection by eye, simply controlling the reflection strength from 0-100%. That’s when the meme “everything has a Fresnel” appeared, to show artist that this option should always be enabled, even if it is metal or paper, everything has a Fresnel. Then the Electric-Dielectric approach appeared. Which is already even used in the gaming industry and a specially developed PBR standard.
Now the Corona has gone even further, there is no more Fresnel option, no more free control of the IOR. Now everyone has come to the conclusion that a material electric or dielectric. What determines its properties. Just like in real life. It's been a long journey. Why is Blender throwing 15 years of experience of the world community down the drain?
Thank you