PDA

View Full Version : Dielectric Material



Richard Hebert
11-16-2012, 09:23 PM
Hi Guys,

I'm using the dielectric material for glass but I'm not sure how to incorporate grunge maps to control reflections. Up until now I've been using the standard material editor and applying gradients. Any links to tuts for this node? I've attempted plugging into various attributes but I really don't have a clue as to what I'm doing even after researching on the web. Thanks for any pointers.

Richard

Paris
11-17-2012, 01:44 AM
This may help;

Glass Node by William Vaughan for NewTek (Part 1)
http://www.youtube.com/watch?v=MMEvA1mFqDI

Glass Node by William Vaughan for NewTek (Part 2)
http://www.youtube.com/watch?v=XgH04RMVIQw

RebelHill
11-17-2012, 03:36 AM
Just plug whatever into the nodes specular input. No need to use gradients to give fresnel though, that's built in to the mat.

Richard Hebert
11-17-2012, 06:00 AM
I'm using only radiosity to light the scene. Is there a way to affect the diffuse of the glass using this material node?

jasonwestmas
11-17-2012, 06:09 AM
Adjusting the absorption and roughness changes the appearance of the light diffusion. Some of these mat nodes are designed to balance diffusion and reflection automatically. The materials are designed to speed up setup time but at the cost of render time.

RebelHill
11-17-2012, 06:21 AM
I'm using only radiosity to light the scene. Is there a way to affect the diffuse of the glass using this material node?

Nope... the same is true of a few material nodes, which are designed to be energy conserving (spec goes up, diffuse goes down). Unless you're talking about absorption, in which case, as JW says.

Richard Hebert
11-18-2012, 02:50 AM
Thanks for the info. If I were to use the node editor to shade the glass which channel would I use to simulate 'Absorption' found in the dielectric material? Sorry guys, still wrapping my brain around the various channels in the node editor.

RebelHill
11-18-2012, 03:40 AM
the one labelled absorption..

jwiede
11-18-2012, 06:22 AM
The way other renderers with energy-conserving materials (Vray, Maxwell, etc.) get around this particular issue is by supporting the notion of "coatings", which allow for a representation of optically-differing material layered on the surface of the base material. In such a system, you'd create a map-driven opaque dirt "coating" layer as part of the dielectric material definition, and the material code sorts it all out at render time.

Implemented properly, coatings invoke much less render time cost versus outright material stacking, because they're typically handled as being very "thin" and thus limited in their ability to modify the optical behavior of the base material. They work great for uses like dirt layers, etching, or optical coatings. As I understand it, though, coating support must be implemented in the material nodes themselves to keep render costs low. Implemented separately, the render cost is no better than material stacking.

I'd really like to see something akin to coating support in LW, though I'm not quite sure how they would be represented in the existing nodal system if implemented in the material nodes themselves.

toby
11-18-2012, 05:39 PM
The way other renderers with energy-conserving materials (Vray, Maxwell, etc.) get around this particular issue is by supporting the notion of "coatings", which allow for a representation of optically-differing material layered on the surface of the base material. In such a system, you'd create a map-driven opaque dirt "coating" layer as part of the dielectric material definition, and the material code sorts it all out at render time.

Implemented properly, coatings invoke much less render time cost versus outright material stacking, because they're typically handled as being very "thin" and thus limited in their ability to modify the optical behavior of the base material. They work great for uses like dirt layers, etching, or optical coatings. As I understand it, though, coating support must be implemented in the material nodes themselves to keep render costs low. Implemented separately, the render cost is no better than material stacking.

I'd really like to see something akin to coating support in LW, though I'm not quite sure how they would be represented in the existing nodal system if implemented in the material nodes themselves.
Material switch node can do that! I just tried it. The renders do speed up when less of the Dialectric material is rendered.

Richard Hebert
11-19-2012, 01:11 AM
I'm sorry Rebel, I wasn't being clear. When using gradients and the Surface node in the Node Editor to create grungy glass which of the all of the input properties would I use to achieve a similar result to the dielectric's 'absorption' feature?

RebelHill
11-19-2012, 04:24 AM
oh, ok... WITHOUT the dielectric node... Simples, set thickness as the gradients input parameter, plugged into transparancy.

Richard Hebert
11-19-2012, 04:55 AM
Got it, Thanks.

jwiede
11-22-2012, 03:11 AM
Material switch node can do that! I just tried it. The renders do speed up when less of the Dialectric material is rendered.
I'd originally mentioned using the switch node, actually, but edited it out because in my experiences it rarely looks as expected/desired. The problem with the switch node is that it _replaces_ the dielectric material based on the map, but really you want the appearance that the dielectric is still there underneath the grunge. Switching out the dielectric material looks increasing visually incorrect the "thicker" the object surfaced, because internal reflections, etc. aren't properly respected when the dielectric is actually switched out for another material. Still, the material switch node might give visually acceptable results in some cases (I just am either too picky, or haven't found the right cases).

toby
11-22-2012, 03:12 PM
That would be the same as stacking Materials then. You won't save the render time of calculating both Materials if you don't replace one with the other. Material switch node would probably work great for labels, but not for something like grunge, where you expect to see some gradation between the materials; but you will always have to render both Materials in that case.

jwiede
11-25-2012, 05:01 AM
That would be the same as stacking Materials then. You won't save the render time of calculating both Materials if you don't replace one with the other. Material switch node would probably work great for labels, but not for something like grunge, where you expect to see some gradation between the materials; but you will always have to render both Materials in that case.
Right, that's why I brought up "coatings", they're the "usual" solution to support such things without material stacking cost. There are a lot of common needs (decals, paint, grunge, etc.) which don't really require full 3D optical evaluation, and can be handled as if a "2D" layer on the surface of the material.

Cageman
11-25-2012, 07:45 AM
Hmm... DB&W has a neato Material Blender and Booster node that should be able to achive what is suggested regarding "coatings". I havn't used it for a while, but I was able to add a regular Spec-hit ontop of a Dialectric node, to create a certain look. Since it supports both Materials and regular shaders, I'm quite sure that one can get the result he/she wants in such special cases.

Oh.. and it is free!!!

http://www.db-w.com/products/dbwtools/docs

jwiede
11-25-2012, 02:44 PM
Yeah, one of those two db&w nodes should be able to set up something close if not precisely correct, the inputs they offer should enable it, it's just a matter of figuring out the right node flow. I have no idea whether the render cost of doing it that way will actually be less than material stacking, but it's definitely worth a try. You'd also still be modifying the optical behavior of the dielectric material itself, which isn't quite what you really want, but I bet you can get much closer to acceptable results versus using material switch node.

The main problem, similar to using the material switch node, occurs if the camera can "see" (due to refl/refr/whatever) both the "top" of the modified surface, and is in a position to see "under" it as well (like when you can see the face and underside of a sticker on a mirror). Since you're still modifying the material itself, there is no "underside" to see in this approach, which a viewer's eye might catch. I can't easily think of how to work around that issue even using the db&w nodes, but if a surfacing "pro" here would like to demonstrate a node flow that solves the issue (without outright material stacking), I'm sure we'd all be grateful.