Page 1 of 2

GENERAL: Nd in relation to opaque or transparent objects

Posted: Fri Jul 07, 2006 3:57 am
by Thomas An.
Hi all,

This seems to come up often.
Here are a couple of pointers in regards to this aspect.

In reference to some physics concepts illustrated here:
http://www.maxwellrender.com/forum/view ... hp?t=17142

INDEX OF REFRACTION IN GENERAL:
Index of refraction is used both on Snell's law and in Fresnel equations.
This means that index of refraction determines:
  1. The angle that the incident rays will bend to; when entering a dielectric (This is from Snell)
  2. The proportions of energy (the intensity) that goes into reflection versus refraction. It is this aspect of Nd that is responsible for the mirror-look (reflections) in objects. (This is from Fresnel)
DIELECTRIC IN MAXWELL MATERIAL EDITOR:
The setting that switches a material from opeque to dielectric is the transmittance color (not the Nd). If the transmittance color is other than zero (0) then the material is technically a dielectric; at which point the lightness of the transmittance color and attenuation distance will define how dark it appears.

Each dielectric in nature has an index of refraction (at 489nm) associated to it. Most dielectrics range from 1.0 to 3.0, but the Snell and Fresnel equations can be used to predict imaginary dielectrics with higher Nd values as can be seen here:
Image
These might not all exist in nature, but they might be constructed in a lab (material science). For example some chemical elements do not occur in nature but they are manufactured in a lab by means of breaking atomic nuclea and recombining them into new elements.

The above graph illustrates the fresnel behavior as the Nd is increased for a dielectric. In Maxwell we can turn off the transparency of the material (by switching transmittance to RGB[0,0,0] ) and then we are left to keep the fresnel reflection characteristics for that particular Nd value.
  • An Nd of 1.5 corresponds to a reflectance-0 of 4%
    An Nd of 3.0 corresponds to a reflectance-0 of 25%
    An Nd of 5.0 corresponds to a reflectance-0 of 44%
    An Nd of 10.0 corresponds to a reflectance-0 of 67%
    An Nd of 20.0 corresponds to a reflectance-0 of 82%
    An Nd of 50.0 corresponds to a reflectance-0 of 92%
    An Nd of 100.0 corresponds to a reflectance-0 of 96%
    An Nd of 200.0 corresponds to a reflectance-0 of 98%
    An Nd of 500.0 corresponds to a reflectance-0 of 99.2%
    An Nd of 1000.0 corresponds to a reflectance-0 of 99.6%
EMULATING REAL LIFE METALS WITHOUT USING COMPLEX IOR FILES
Real life materials have a specific reflectance which can be measured in a lab. Also, the complex IOR files (n,k data) contain lab measured reflectance values.

We can take the actual experimental data for reflection and then work backwords to find what Nd value of the material editor would give us approximately the same reflectance percentage.

When we do such corrolation then the result is something like this for some common metals:
Image

For example, to emulate silver (which has a reflectance of 97% - 98.6 accross the spectrum) without using complex IOR data we can set:
  • Reflectance-0 at RGB(248,248,248)
  • Reflectance-90 at RGB (255,255,255)
  • Transmittance RGB(0,0,0) (opeque object)
  • Nd=140 (or up to 200)
  • Roughness=3.0
So the use of high Nd values on opeque objects is actually a mathematical shortcut to emulate the reflectance look of shinny (or mirror) objects without having to deal with complex IOR files and at the same time being as close to those complex IOR files as possible (if not nearly identical). So, for the silver example above the process was to ask the question "what imaginary/theoretical dielectric material would have a reflectance of 97 percent ?". The answer was "a dielectric whose Nd = 140". Then we made that material opeque (by switching transmittance to 0) and piggyback on the reflectance characteristics inherited by that high Nd value via Fresnel laws.

SUMMARY:
-Index of refraction controls bending of rays as well as reflection percentage
-The Snell and Fresnel equations can predict high Nd dielectrics.
-We can turn off the transparency of a material but keep its reflection properties inherited by the Nd number
-We can use these high Nd numbers to emulate highly reflective materials such as metals.
-We do all this so that we do not have to always use complex IOR files which are hard to find.
-Therefore the material editor works in two modes
a) using complex IOR files
b) manual mode where we can emulate the look of real life materials

______________________________________________________
A word of caution:

Do not use the IOR values often listed on the web for *non* dielectrics in the "custom IOR" slot of the material editor (manual mode) ...
  • These values are incompatible with the manual mode of the material editor; since they relate to complex IOR values. Actually they are substantially simplified / incomplete versions of complex IOR data and cannot be used without a corresponding extinction coefficient(s).
    For example silver(Ag) is often listed as having IOR=0.18 which is only the 'n' value of the n,k file corresponding to about 400nm
    This can be checked here: http://www.luxpop.com/
    (Enter: Ag, 400nm, 25 C)

Posted: Fri Jul 07, 2006 4:21 am
by DrMerman
Thomas, you are on FIRE tonight :)

Posted: Fri Jul 07, 2006 4:23 am
by Thomas An.
DrMerman wrote:Thomas, you are on FIRE tonight :)
Yeah, I had a bunch of stuff but have been procrastinating.
(so I am just pushing them out of the "to DO" list :P )

Posted: Fri Jul 07, 2006 6:47 am
by superbad
Very helpful. Thanks for taking the time to put all that together, and for laying it out so nicely as well. I've printed off the Physics summary and am looking through it now.

Posted: Fri Jul 07, 2006 1:27 pm
by jomaga
Thanks a lot Thomas, very useful!

Posted: Sun Jul 09, 2006 6:15 am
by chrisvconley
I nominate Thomas for the A-Team.

Posted: Sun Jul 09, 2006 10:49 am
by tom
:D

Posted: Mon Jul 10, 2006 1:04 am
by JDHill

Posted: Mon Jul 10, 2006 2:09 am
by Thomas An.
A word of caution:

Do not use the IOR values listed on the web for *non* dielectrics ...

a) They are substantially incomplete versions of complex IOR data and cannot be used without a corresponding extinction coefficient(s) (as a matter of fact they are most likely complex IOR averages across the spectrum for the material).
For example silver is listed as having IOR=0.18 which corresponds to about 400nm
This can be checked here: http://www.luxpop.com/
(Enter: Ag, 400nm, 25 C)

b) Because these listed non-dielectric values are relating to complex IOR data they are not compatible with the Maxwell manual material creation method as laid out in the first post. (again this partains to the fact that the Maxwell system can work either in manual mode or complex IOR mode)

Posted: Mon Jul 10, 2006 2:47 am
by JDHill
...so what reference do you suggest for approximating Nd values in non-opaque, non-metallic materials? (polystyrene, urethanes, etc.)

~JD

Posted: Mon Jul 10, 2006 3:19 am
by Thomas An.
JDHill wrote:...so what reference do you suggest for approximating Nd values in non-opaque, non-metallic materials? (polystyrene, urethanes, etc.)

~JD
Hi JD,

Tha values you listed above are ok for dielectrics only (they can be dropped in the "custom IOR" slot).

The caution is in regards to non-dielectrics (that appear interspersed on these lists)

Posted: Mon Jul 10, 2006 3:34 am
by Robert Cervellione
holly dear lord i finally understand ND value. thanks Thomas.

Posted: Mon Jul 10, 2006 5:07 am
by JDHill
Sorry Thomas, I made a typo...I meant to say 'opaque, non-metallic'...plasics and such.

~JD

Posted: Mon Jul 10, 2006 5:39 am
by Thomas An.
JDHill wrote:Sorry Thomas, I made a typo...I meant to say 'opaque, non-metallic'...plasics and such.

~JD
oh, I see...

Complex IOR data for plastics are not easy to find. However ... there are some references that I have which can be worked to get some values for a few plastics.

For example PET (Polyethylene terephthalate) would have approx. a Maxwell Nd=24 (with Refl-0 V=216: where 'V' is for HSV color)

By the way ... if anyone has any references like this one ... please let me know
http://www.shef.ac.uk/physics/research/ ... tivity.pdf

Posted: Mon Jul 10, 2006 6:01 am
by JDHill
how about this one: http://chemistry.clemson.edu/ChemDocs/s ... 202003.pdf

...should come in handy for that trifluorovinyl ether-based perfluorocyclobutyl material I was working on. :lol: