PDA

View Full Version : Dielectric Material



Richard Hebert
07-20-2014, 05:09 PM
The Dielectric material has a parameter called 'partial internal reflections'. Does anybody know what this refers to and how this can be simulated in the standard material editor? Just wondering. Thanks everyone for all the support!

Richard

spherical
07-20-2014, 10:15 PM
It, well, does what it says on the tin. There is a function of a dielectric that is what is known as Total Internal Reflection. TIRs are created on internal surfaces that are adjacent to the surface where the ray enters and the surface where the ray leaves. IOW, in a cube, the side wall will always be in TIR because the refraction of the ray entering will place the ray outside the Critical Angle. No rays that entered from the front or enter from the rear can escape through that side surface. The reflection is absolute and, thus, Total.

Partial Internal Reflection is the property of a surface that interfaces with a medium having a different Index Of Refraction. You get dim reflections on the inside, the same as you get them on the outside; according to the Fresnel Equations. The surface reflects, no matter which vector it may be viewed from, but at different values of intensity depending upon the angle. This is why when looking up toward the surface when submerged in water, you see objects below you reflected in it.

I asked for PIRs, as there were none in the first iterations of Dielectric. Once we got them, I asked for a switch on PIRs, so they can be turned off and dramatically speed up render times for situations where absolute realism isn't necessary. Objects in the distance would consume many cycles computing these reflections and you wouldn't see them in the render. So, they're useless to compute. Also, if you don't care about pure realism, you can turn them off on foreground objects.

For more on this, see these threads:
http://forums.newtek.com/showthread.php?85685-Want-Faster-Dielectric
In the first post, you can see my Dielectric test scene. The first image is Dielectric Material alone. The right wall shows that it is in TIR, as you can see only the large black sphere on the vertical rod that is to the left of the cube in the distance. The faint reflections on the back and front surfaces of the cube are PIRs; showing the internal spheres and external rods. That particular post deals with a method of defeating PIRs until the switch was coded in to turn them off, so other than educational reference on how this was done, not really pertinent to your question, here.

Detailed explanation (mid-thread of the link following) on TIRs and how they work:
http://forums.newtek.com/showthread.php?90456-Dielectric-Materials-More-Complexity-3D-Color-amp-Depth&p=772148&viewfull=1#post772148

Near complete treatise on Dielectric and how to use it, especially with color:
http://forums.newtek.com/showthread.php?90456-Dielectric-Materials-More-Complexity-3D-Color-amp-Depth&highlight=dielectric

As for simulating in the Standard Material Editor, I presume that you mean in the Surface Editor. AFAIK, it can't be done without standing on your head to create extra geometry upon which you could change the reflective properties to be different and, if it could, why? You're regressing back into copy/paste flipped polys and/or double-sided polys and that is just not recommended.

Richard Hebert
07-21-2014, 06:09 AM
Wow, thanks for the detailed response. Gonna take a while to digest everything you just shared! Sorry, I was referring to the surface editor, my bad.

spherical
07-21-2014, 03:39 PM
Happy to help. Dielectric gets a lot of queries as to how it works and whether it is accurate. If you want to move away from accuracy, get LightWolf's db&w Tools (http://www.db-w.com/). In there, are Material Booster and Material Tweaker. If the aspects of Dielectric are not to your liking, even though they are correct—just not what you think—perception is a funny thing if you don't have a real-world example to compare, you can adjust the computation. I use them sometimes for specific effect but don't move too far away from zero. If it's subtle, you can get a way with it and it still looks real.

Yes, there was a bit of a question as to whether you wanted to do something in the Surface Editor or in the Standard Material Node. In the latter, you could assemble a node network that could replicate PIRs, probably using Spot Info for poly side and a Fresnel or Gradient, but unless it is a very specific use, probably redundant.

Richard Hebert
07-21-2014, 10:06 PM
Dielectric is great but as far as I know, no way to 'dirty' it up with grime and fingerprints etc.

spherical
07-22-2014, 12:01 AM
Like this?
123163
MultiFractal procedural Color and Alpha applied to Color and Absorption inputs.

djwaterman
07-22-2014, 12:21 AM
Grime on a dielectric material? Check this thread (http://forums.newtek.com/showthread.php?136307-Hans-Pieck-Lounge-Chair-modeling-and-renders&highlight=Hans+Pic+armchair), scan down to a scene file I uploaded for putting a decal on glass, it's the same principal I used to put grime onto a clear plastic. I'm no node scientist but it does work, and can possibly be improved upon.

spherical
07-22-2014, 12:25 AM
There ya go! Gotta take that scene apart. Nice old Radio, BTW. A friend of mine has quite a collection of them.

<Rodney Dielectric>"I don't get no respect... no respect at all!"</Rodney Dielectric>

grabiller
08-06-2014, 08:28 AM
Hello,

For your information, I discovered a bug in LW with dielectric material and interpolated animated cached radiosity, while using "Directional Rays".

Normally you want to enable "Directional Rays" with the dielectric material as it has the most visible (and correct) effects on it.

If you render an animated scene with these element without caching (but still using interpolation) you'll of course get flickering, but no other special issues.

However, if you bake your radiosity cache and then render the same scene, you'll get black squares poping up here and there quickly, mostly in the reflection but also on other surfaces influenced by the light rays reflected by the dielectric material.

The workaround is to disable "Ray Cutoff" by setting it to 0. Issue gone. But longer render times. I suspect there is a precision issue related to radiosity cache files.

I'm in the process of producing elements to show the issue and report it.

Meanwhile, be aware of this potential issue if you use dielectric material and interpolated animated cached radiosity, with "Directional Rays" on.

spherical
08-06-2014, 05:23 PM
Unable to reproduce this. Thought that perhaps it may be Use Transparency setting. Tried both On and Off, clearing and re-baking the cache each time. This was with both Interpolated FG and Interpolated MC. Other than render times and additional quality, no difference. What is your RRL set to and does it cover the number of recursions required by the GI settings and a complex set of bounces that a transparent can generate?

grabiller
08-06-2014, 06:38 PM
Unable to reproduce this. Thought that perhaps it may be Use Transparency setting. Tried both On and Off, clearing and re-baking the cache each time. Other than render times and additional quality, no difference. This was with Interpolated FG. Trying Interpolated MC as I type. What is your RRL set to and does it cover the number of recursions required by the GI settings and a complex set of bounces that a transparent can generate?
I'll post 2 movies for you to see the issue, with the corresponding settings. The issue is not in RRL as when rendering the same sequence without cache it works.

spherical
08-06-2014, 07:23 PM
Will be interesting to see. So you figure it's a Radiosity Locked Cache issue.

grabiller
08-07-2014, 12:59 AM
Here is the video:

http://youtu.be/-P5gR3wGJBE
First half has Ray Cutoff to default value (0.01), second half has Ray Cutoff to 0.0.

Here are two original images:
123548
Ray Cutoff to default value (0.01), notice the black squares.
123549
Ray Cutoff to 0.0

Other common settings are:

Render
Ray Recursion Limit: 12
Ray Precision: 6.0
Ray Cutoff: 0.01 / 0.0
Shading Samples: 4
Light Samples: 8 (but no light used, only illuminated polys).

Global Illumination
ON: Interpolated, Use Transparency, Blur Background, Directional Rays, Use Behind Test, Use Bumps.
OFF: Ambient Occlusion, Use Gradients, Volumetric Radiosity.
Intensity: 100%
Indirect Bounces: 2
Rays Per Evaluation: 512
Secondary Bounce Rays: 256
Angular Tolerance: 0.0
Minimum Pixel Spacing: 1
Maximum Pixel Spacing: 128
Multiplier: 50%
Cache/Animation ON, Frame Step: 1

Cameras
AA: 1/16/0.015/Mitchell/Low-Discrepancy

Render Times
(i7 4 Cores 2.7 Ghz)
Ray Cutoff Default: 10 min
Ray Cutoff Zero: 18 min

spherical
08-07-2014, 04:28 AM
They seem to be limited to only that one front quadrant. I notice in the still that the large square above the 287 reflection has a corner cut out of its lower left. Looks to be a "ray shadow" of the round rail that surrounds the material ball at the base that is blocking rays; otherwise the square would be whole. You can see that the reflection of the rail itself doesn't come near the black square but it lines up with the missing corner. Makes me wonder what is in the distance behind the camera that could be influencing the interpolated calculation.

What type of GI?

grabiller
08-07-2014, 04:53 AM
They seem to be limited to only that one front quadrant. I notice in the still that the large square above the 287 reflection has a corner cut out of its lower left. Looks to be a "ray shadow" of the round rail that surrounds the material ball at the base that is blocking rays; otherwise the square would be whole. You can see that the reflection of the rail itself doesn't come near the black square but it lines up with the missing corner. Makes me wonder what is in the distance behind the camera that could be influencing the interpolated calculation.

What type of GI?

Monte Carlo.

Mr. Wilde
04-03-2015, 05:52 AM
It, well, does what it says on the tin. There is a function of a dielectric that is what is known as Total Internal Reflection. TIRs are created on internal surfaces that are adjacent to the surface where the ray enters and the surface where the ray leaves. IOW, in a cube, the side wall will always be in TIR because the refraction of the ray entering will place the ray outside the Critical Angle. No rays that entered from the front or enter from the rear can escape through that side surface. The reflection is absolute and, thus, Total.

Partial Internal Reflection is the property of a surface that interfaces with a medium having a different Index Of Refraction. You get dim reflections on the inside, the same as you get them on the outside; according to the Fresnel Equations. The surface reflects, no matter which vector it may be viewed from, but at different values of intensity depending upon the angle. This is why when looking up toward the surface when submerged in water, you see objects below you reflected in it.

I asked for PIRs, as there were none in the first iterations of Dielectric. Once we got them, I asked for a switch on PIRs, so they can be turned off and dramatically speed up render times for situations where absolute realism isn't necessary. Objects in the distance would consume many cycles computing these reflections and you wouldn't see them in the render. So, they're useless to compute. Also, if you don't care about pure realism, you can turn them off on foreground objects.

For more on this, see these threads:
http://forums.newtek.com/showthread.php?85685-Want-Faster-Dielectric
In the first post, you can see my Dielectric test scene. The first image is Dielectric Material alone. The right wall shows that it is in TIR, as you can see only the large black sphere on the vertical rod that is to the left of the cube in the distance. The faint reflections on the back and front surfaces of the cube are PIRs; showing the internal spheres and external rods. That particular post deals with a method of defeating PIRs until the switch was coded in to turn them off, so other than educational reference on how this was done, not really pertinent to your question, here.

Detailed explanation (mid-thread of the link following) on TIRs and how they work:
http://forums.newtek.com/showthread.php?90456-Dielectric-Materials-More-Complexity-3D-Color-amp-Depth&p=772148&viewfull=1#post772148

Near complete treatise on Dielectric and how to use it, especially with color:
http://forums.newtek.com/showthread.php?90456-Dielectric-Materials-More-Complexity-3D-Color-amp-Depth&highlight=dielectric

As for simulating in the Standard Material Editor, I presume that you mean in the Surface Editor. AFAIK, it can't be done without standing on your head to create extra geometry upon which you could change the reflective properties to be different and, if it could, why? You're regressing back into copy/paste flipped polys and/or double-sided polys and that is just not recommended.

I'm researching the Dielectric node, and I can't see ANY of these links.


Mr. Wilde, you do not have permission to access this page. This could be due to one of several reasons:

Your user account may not have sufficient privileges to access this page. Are you trying to edit someone else's post, access administrative features or some other privileged system?
If you are trying to post, the administrator may have disabled your account, or it may be awaiting activation.

jwiede
04-03-2015, 12:08 PM
I'm researching the Dielectric node, and I can't see ANY of these links.

I'm getting the same message. Might be because LW3DG removed the owning topic? Definitely odd, though, never ran into that before.

Matt
04-03-2015, 12:58 PM
Those links work for me.

tommyc
04-03-2015, 06:53 PM
I get the same message also.

Snosrap
04-03-2015, 07:44 PM
Those links work for me.

Internal folks can always get to everything. :)

spherical
04-03-2015, 08:42 PM
All three work for me, so they are definitely available from outside.

Just out of curiosity, what "message" was returned when the links were followed?

Mr. Wilde
04-04-2015, 12:08 PM
All three work for me, so they are definitely available from outside.

Just out of curiosity, what "message" was returned when the links were followed?

The message I get:

Mr. Wilde, you do not have permission to access this page. This could be due to one of several reasons:

Your user account may not have sufficient privileges to access this page. Are you trying to edit someone else's post, access administrative features or some other privileged system?
If you are trying to post, the administrator may have disabled your account, or it may be awaiting activation.
Why should posts like that be in a restricted sub forum?

Ztreem
04-04-2015, 03:32 PM
I think it's quite easy, the links go to the old open beta forums and likely it is that if you never was part of that beta you don't have access to that part of the forum. This is just a guess from my side.

spherical
04-04-2015, 03:41 PM
Ahhh... yes, that could be it. Ummm. now that 9.x is out of beta, I think it's time to open that up. There is a lot of high quality stuff in there. I just pinged Steve to see if that's possible.

Mr. Wilde
04-12-2015, 06:54 AM
Ahhh... yes, that could be it. Ummm. now that 9.x is out of beta, I think it's time to open that up. There is a lot of high quality stuff in there. I just pinged Steve to see if that's possible.

Is there any news on this?

jwiede
04-12-2015, 12:05 PM
I think it's quite easy, the links go to the old open beta forums and likely it is that if you never was part of that beta you don't have access to that part of the forum. This is just a guess from my side.
Problem is, I _was_ part of those v9 OB forums, so if that's the reason I should still be able to see them (and cannot, instead I still get the same msg as of today).

spherical
04-12-2015, 04:13 PM
Is there any news on this?

Steve replied back that it's up to LW3DG mods (forum is split between them and NT, even though it is all one thing in appearance) and the message to Lino bounced, due to his PM folder exceeding the limit. (He's a mod... why does he have a limit?) Asked Steve to forward to him somehow.


Problem is, I _was_ part of those v9 OB forums, so if that's the reason I should still be able to see them (and cannot, instead I still get the same msg as of today).

Can you get into the forum section itself?

jwiede
04-13-2015, 12:41 PM
Nope, I have the HC-era "Membership Zone", but at some pt the v9-era OB forums went missing.

hazmat777
04-13-2015, 02:00 PM
Nope, I have the HC-era "Membership Zone", but at some pt the v9-era OB forums went missing.

Hmm. That seems like an oversight on NT's part. I was never part of HC, but was part of the 9.6 beta days and the links work fine for me. It seems to me that the HC members should have access to everything even if they backed out of the program just as a simple and free gesture.

spherical
04-13-2015, 03:01 PM
I have:
Archives
LightWave Membership Zone
LightWave Open Beta

At this point, all should have access. There's nothing to be secretive about, because the application technology has marched significantly farther. Well..... except for the negative press that was going on at the time. Perhaps my request to open them up won't be granted.

hazmat777
04-13-2015, 03:36 PM
If this is against the rules, please delete this and accept my apology. It's been so long that I don't see the harm in it though. 7 pages on Dielectric thread posts.

spherical
04-14-2015, 03:11 AM
That's many, but not all. Thanks for bringing them to the public area. I was thinking about doing so, but got forked off onto other things today.

Mr. Wilde
04-14-2015, 06:35 AM
These posts are six years old. I think whatever has been secret back then could be revealed now.