MIA Transparency Issue ?

Discussions regarding Materials, Material-Compounds or Shaders, etc.
Post Reply
Falam

MIA Transparency Issue ?

Post by Falam » 04 Mar 2013, 20:08

Image
Image

The centre of the object is suppose to be transparent, as I have a transparent image applied (or gradient) :) There is an object within the object, I placed a light directly above the object, hoping that was the problem, same result I can't see the object beneath / within ?

User avatar
Draise
Posts: 891
Joined: 09 Oct 2012, 20:48
Skype: ondraise
Location: Colombia

Re: MIA Transparency Issue ?

Post by Draise » 04 Mar 2013, 21:36

Try invert the transparency map/mask, see if that works, or disconnect it and see if it's working without the mask.

Hope that works. ;)

Falam

Re: MIA Transparency Issue ?

Post by Falam » 04 Mar 2013, 23:05

Draise wrote:Try invert the transparency map/mask, see if that works, or disconnect it and see if it's working without the mask.

Hope that works. ;)
I tried all that :(

User avatar
mattmos
Posts: 445
Joined: 02 Dec 2009, 16:59

Re: MIA Transparency Issue ?

Post by mattmos » 05 Mar 2013, 03:37

try changing the index of refraction to 1 and see if that helps?

Falam

Re: MIA Transparency Issue ?

Post by Falam » 05 Mar 2013, 04:04

mattmos wrote:try changing the index of refraction to 1 and see if that helps?
It works, not without increasing the sampling higher then I thought X( I'm going to tinker with US.

Pancho
Posts: 659
Joined: 19 Sep 2010, 11:28

Re: MIA Transparency Issue ?

Post by Pancho » 05 Mar 2013, 09:05

Why these strange settings?

Ior >=1 and gloss >=.5 should give you predictable results. Gloss goes with about 16 to 32 samples for clean results with 0.5.

User avatar
Maximus
Posts: 1105
Joined: 09 Jun 2009, 15:45

Re: MIA Transparency Issue ?

Post by Maximus » 05 Mar 2013, 12:55

What material are you trying to achieve? You have totally weird settings, there isnt anything in real world that has 0.01 index of refraction.
If you are going for metal turn the "metal material" feature on, check the brdf "Default (IOF)" on and set the index of refraction at 30.

Falam

Re: MIA Transparency Issue ?

Post by Falam » 05 Mar 2013, 15:38

Maximus wrote:What material are you trying to achieve? You have totally weird settings, there isnt anything in real world that has 0.01 index of refraction.
If you are going for metal turn the "metal material" feature on, check the brdf "Default (IOF)" on and set the index of refraction at 30.
:) I changed the settings since then, I wasn't going to leave those settings :) Why a IOF to 30 for metal ?

User avatar
Maximus
Posts: 1105
Joined: 09 Jun 2009, 15:45

Re: MIA Transparency Issue ?

Post by Maximus » 05 Mar 2013, 20:25

Falam wrote:
Maximus wrote:What material are you trying to achieve? You have totally weird settings, there isnt anything in real world that has 0.01 index of refraction.
If you are going for metal turn the "metal material" feature on, check the brdf "Default (IOF)" on and set the index of refraction at 30.
:) I changed the settings since then, I wasn't going to leave those settings :) Why a IOF to 30 for metal ?
Because metals arent like other elements, and they have a "Complex Index of Refraction" and "Extinction Coefficient", this cannot be reproduced with Mia_material, you can go really near to it splitting the "Reflection output" (using mia_maerial_x) x 3 times for each channel, R G B. Its a complex setup i'll post a screen in a few, give me time to build.
Back to topic, thats how metal works with Mia_Material, you need to check metal material and use high IOR, metals are so heavily in reflections as opposed to actual lighting.
For example in maxwell render which is the most physically accurate one, you set the IOR for metals at 20+ because it has to simulate a higher surface dispersion of the light itself.

The IOR 20+ also makes the "Fresnel" effect more realistic to achieve metals. In the setup with reflection output using 3 mia_x plugged into a mia_material and using just a single mia_material you can tell the difference in quality of metals.

I'll get back to you once i've setup the scene.

User avatar
Maximus
Posts: 1105
Joined: 09 Jun 2009, 15:45

Re: MIA Transparency Issue ?

Post by Maximus » 05 Mar 2013, 21:50

Ok here is the render:

[rimg=400]http://img585.imageshack.us/img585/1748/metals.jpg[/rimg]

and here is the render tree

[rimg=400]http://img832.imageshack.us/img832/1020/mtlr.jpg[/rimg]

This setup is based on a tutorial i saw time ago which i recommend you if you are interested in rendering good metals you can find it here:

https://www.fxphd.com/fxphd/courseDetai ... Course=292

Anyway in the render, left sphere is this triple mia_material_x technique, which basically split the reflction color based on 3 different tipes of BRDF curve fresnel, to match as close as possible the real life behaviour of metals reflections. I shifted the colors a bit to have you notice the difference in shading and the fresnel effect that happens with some metals towards the edge, you can see a blending color change in that area while in the normal architectural material (right sphere) this doesnt quite happen, if i would have to add some color it would color the whole reflection, and even if you would split the color and try to match the other technique you wouldnt success, because mia_material doesnt offer you the "K" extinction coefficient function which is needed to render correct metals.

That said, the most correct method is the left sphere, it also adds a lot to render times, but everyone uses right sphere with just architectural material with ior 30+ because its faster and unless you want to achieve realism, no one would ever notice.
That said both are still quite far from Maxwell quality, that is because Maxwell have exposed the "K" value which is the extinction coefficient.
Hopefully with new MiLa shaders Mental Images will implement the K value and give us the opportunity to build correct looking metals.

Hope that helps

edit: actually i added the the compound you can just plug and use it, if you want to change color you need to change all the rgb values for each reflection output, but its quite hard to explain, its not "random" you need to know the correct values of real life metals which you can get from here http://refractiveindex.info/
Also you need to do some math for each metal because it requires you to calculate the brdf fresnel curve so to put correct numbers in each rgb channel. Quite a headache :)
Attachments
Titanium.xsirtcompound
(35.39 KiB) Downloaded 133 times

Falam

Re: MIA Transparency Issue ?

Post by Falam » 05 Mar 2013, 23:41

Maximus wrote:Ok here is the render:

[rimg=400]http://img585.imageshack.us/img585/1748/metals.jpg[/rimg]

and here is the render tree

[rimg=400]http://img832.imageshack.us/img832/1020/mtlr.jpg[/rimg]

This setup is based on a tutorial i saw time ago which i recommend you if you are interested in rendering good metals you can find it here:

https://www.fxphd.com/fxphd/courseDetai ... Course=292

Anyway in the render, left sphere is this triple mia_material_x technique, which basically split the reflction color based on 3 different tipes of BRDF curve fresnel, to match as close as possible the real life behaviour of metals reflections. I shifted the colors a bit to have you notice the difference in shading and the fresnel effect that happens with some metals towards the edge, you can see a blending color change in that area while in the normal architectural material (right sphere) this doesnt quite happen, if i would have to add some color it would color the whole reflection, and even if you would split the color and try to match the other technique you wouldnt success, because mia_material doesnt offer you the "K" extinction coefficient function which is needed to render correct metals.

That said, the most correct method is the left sphere, it also adds a lot to render times, but everyone uses right sphere with just architectural material with ior 30+ because its faster and unless you want to achieve realism, no one would ever notice.
That said both are still quite far from Maxwell quality, that is because Maxwell have exposed the "K" value which is the extinction coefficient.
Hopefully with new MiLa shaders Mental Images will implement the K value and give us the opportunity to build correct looking metals.

Hope that helps

edit: actually i added the the compound you can just plug and use it, if you want to change color you need to change all the rgb values for each reflection output, but its quite hard to explain, its not "random" you need to know the correct values of real life metals which you can get from here http://refractiveindex.info/
Also you need to do some math for each metal because it requires you to calculate the brdf fresnel curve so to put correct numbers in each rgb channel. Quite a headache :)
My scene has many dirty metal materials :) I applied the IOR to a object that is part of the scene. I don't understand how IOR which is associated with transparency relates to reflections ? :) My original IOR was 0.01, or 'off' as I refer to it as, I cranked it up to 30. I noticed my reflection area on the object was no longer white, rather took on the material color which is red, from what I take from it this is normal for realistic reflections ?

The FXpd.com link, links to a course, nothing about rendering good metals ? :)

How do you calculate brdf fresnel curve to RGB ? One thing I don't know and can't seem to find the direct answer in my searches is, under the periodic table of elements, most of the elements are metal, since there are various types of metals, finding the one I want on the refractive.info page link is difficult ~x(

User avatar
Maximus
Posts: 1105
Joined: 09 Jun 2009, 15:45

Re: MIA Transparency Issue ?

Post by Maximus » 06 Mar 2013, 07:30

In metals you use IOR to drive the fresnel curve, as i said you dont have exposed the parameter to render correct metals which in real life happens with IOR + K (Extinction coefficent).
That said, theoretically if you want a mirror chrome you would uncheck the "Default IOF" and set the facing reflectivity at 1.
Basically in the architectural material, all those options are used to control the BRDF. which also control the reflections curve.
In metals what matters a lot is the "Fresnel" effect.
Metals dont have the color reflection separated, Metals also influence the color of their reflections. So the color of the reflections in metals depends on the color of the metal (which depends on the diffuse color you put). When you turn on "Metal Material" checkbox, Architectural shader does exactly this, makes the reflection of your shader matches the color of the metal, because that is how they work.

Setting the IOR to 30+, is the only way to go as near as you can (without using the other method which is really complex and requires a big setup which is explained in that course) for rendering quite "correct" metals Fresnel effect.
To calculate brdf fresnel curve based on IOR+K from that website you need an excel sheet and some math which is explained in that course, thats why i linked you. And i should take some times to study it again since i quite never had the need of using this complex system. Most renders dont need this extreme correct setup because you wont notice.
With Architectural material, check "Default IOF", check "Metal MAterial" and set IOR to 30+ you can almost create all metals. (I'd say you can "Fake" all metals).

In the post before i've attached you the compound with the correct setup if you want to play with it. Each R G B Channel uses a mia_material_x that drives the reflection curves to produce "coloured" metal in the most possible correct way available to us with Architectural material (which still isnt remotely near what happens in real life).
Dont look at the Architectural node in which those are plugged because its IOR is driven by the other 3 shaders plugged in the "Refl_color".

Now each of those mia_material_x are controlled with the BRDF settings, 0 degree / 90 degree / curve. Those settings are the ones that makes the fresnel effect correct for each color channel. I've just set you the result of that tutorial in this compound, i know its hard to understand because you would have to watch the tutorial since they explain all the math you have to do and give you the tools to do that.

P.s. in the link i gave you its listed all the course, and the chapter where he explains this are chapter 7 and chapter 8.

Falam

Re: MIA Transparency Issue ?

Post by Falam » 06 Mar 2013, 16:59

Maximus wrote:In metals you use IOR to drive the fresnel curve, as i said you dont have exposed the parameter to render correct metals which in real life happens with IOR + K (Extinction coefficent).
That said, theoretically if you want a mirror chrome you would uncheck the "Default IOF" and set the facing reflectivity at 1.
Basically in the architectural material, all those options are used to control the BRDF. which also control the reflections curve.
In metals what matters a lot is the "Fresnel" effect.
Metals dont have the color reflection separated, Metals also influence the color of their reflections. So the color of the reflections in metals depends on the color of the metal (which depends on the diffuse color you put). When you turn on "Metal Material" checkbox, Architectural shader does exactly this, makes the reflection of your shader matches the color of the metal, because that is how they work.

Setting the IOR to 30+, is the only way to go as near as you can (without using the other method which is really complex and requires a big setup which is explained in that course) for rendering quite "correct" metals Fresnel effect.
To calculate brdf fresnel curve based on IOR+K from that website you need an excel sheet and some math which is explained in that course, thats why i linked you. And i should take some times to study it again since i quite never had the need of using this complex system. Most renders dont need this extreme correct setup because you wont notice.
With Architectural material, check "Default IOF", check "Metal MAterial" and set IOR to 30+ you can almost create all metals. (I'd say you can "Fake" all metals).

In the post before i've attached you the compound with the correct setup if you want to play with it. Each R G B Channel uses a mia_material_x that drives the reflection curves to produce "coloured" metal in the most possible correct way available to us with Architectural material (which still isnt remotely near what happens in real life).
Dont look at the Architectural node in which those are plugged because its IOR is driven by the other 3 shaders plugged in the "Refl_color".

Now each of those mia_material_x are controlled with the BRDF settings, 0 degree / 90 degree / curve. Those settings are the ones that makes the fresnel effect correct for each color channel. I've just set you the result of that tutorial in this compound, i know its hard to understand because you would have to watch the tutorial since they explain all the math you have to do and give you the tools to do that.

P.s. in the link i gave you its listed all the course, and the chapter where he explains this are chapter 7 and chapter 8.
Thank you, I understand now :) I'm going to probably get that course, regardless if future Arch shaders are fixed, never know when I need the info, although it's a big course :)

User avatar
Maximus
Posts: 1105
Joined: 09 Jun 2009, 15:45

Re: MIA Transparency Issue ?

Post by Maximus » 06 Mar 2013, 18:24

My advice is just to forget that system because is x2 times slower and you will barely notice the difference unless as i said you want to achieve a really good result. But its a gigantic mess and headhache to do that setup for every time you need a metal :)

Falam

Re: MIA Transparency Issue ?

Post by Falam » 06 Mar 2013, 18:46

Maximus wrote:My advice is just to forget that system because is x2 times slower and you will barely notice the difference unless as i said you want to achieve a really good result. But its a gigantic mess and headhache to do that setup for every time you need a metal :)
HaHa, you suggest I don't watch the tutorial :)
As soon as you add a mix2 node to the Arch/Mia Material you break the physically correctness anyhow, arghhh. ~x(

Post Reply

Who is online

Users browsing this forum: No registered users and 28 guests