trans mat

Dear all,

I am trying to nail this one down, once and for all.
What is your experience/approach? Feedback welcome.

refman:
"The transmissivity is the fraction of penetrating light that travels all
the way through the material.
The transmitted specular component is the fraction of transmitted light
that is not diffusely scattered.
Transmitted and diffusely reflected light is modified by the material color.
Translucent objects are infinitely thin."

BIG QUESTION:
How do CAPITAL indeces (RADIANCE) relate to
lower case indices (PHYSICS)??? See below. Also for a lot more little
questions...

mod trans id
0
0
7 red green blue spec rough trans tspec
#7 A1 A2 A3 A4 A5 A6 A7

Let's talk about grey objects for now:
A=179*(.265*A1+.65*A2+.065*A3)

[hope you like ASCII art]
!!! View with monospaced font !!!
             _.
_| /|
  \ /
1 \ --- /
    \/ \/ b \
    /\ /\ > a
   > \ / | c /
   > V |
---------------- g
   > \ |
   > \ | f \
    \ \/ > d
     \ /\ e /
      --- \
            \|
            -.
Legend

···

------
Incident light = 1 (is that our assumption?)
Reflected light a
- specular b
- diffuse c
Transmitted light d
- specular e
- diffuse f
Absorpt light g

What I am quite sure about:
---------------------------
1 = a + d + g
a = b + c
d = e + f

Some Qs:
--------
The specular components (both diff and spec) are hard to measure.
However, this is what the calculation in the radbook is based on.
How would one make estimations?
Is this only possible through visual comparision?
What are the steps to A1..A7 if given:
- reflectance (luxmeter + luminance meter)
- transmittance (2x luxmeter)
(practical scenario, assuming grey colour)
if
- Lambertion properties or
- clearly see-through
What is the plastic equivalent to a trans without transmittance?
Would that be _wrong_ or just _stupid_ to use?

References
----------
- http://radsite.lbl.gov/radiance/digests_html/v2n7.html#TRANS_MATERIAL
- radbook 5.2.6, page 325 (new edition)

Cheers

Axel

Axel Jacobs wrote:

Dear all,

I am trying to nail this one down, once and for all.
What is your experience/approach? Feedback welcome.

refman:
"The transmissivity is the fraction of penetrating light that travels all
the way through the material.
The transmitted specular component is the fraction of transmitted light
that is not diffusely scattered.
Transmitted and diffusely reflected light is modified by the material color.
Translucent objects are infinitely thin."

BIG QUESTION:
How do CAPITAL indeces (RADIANCE) relate to
lower case indices (PHYSICS)??? See below. Also for a lot more little
questions...

  http://www.schorsch.com/rayfront/manual/transdef.html

Scroll down a bit for the diagram.

-schorsch

···

--
Georg Mischler -- simulations developer -- schorsch at schorsch com
+schorsch.com+ -- lighting design tools -- http://www.schorsch.com/

Wow, that's a great diagram, Schorsch. I think I understand the "trans" type for the first time, myself!

Be sure also to refer to section 5.2.6 (pp. 325-6) in "Rendering with Radiance," which translates between more sensible physical quanitites and the parameters of the "trans" type. I believe Rob G. has an Excel spreadsheet that implements these formulas nicely, if you'd like to request a copy.

By way of excuse, the reason the trans type is so baffling is my stubborn adherence to the principle that Radiance primitives have well-defined legal ranges. All the material parameters (except roughness) have legal physical ranges of [0,1). Since Radiance does not enforce these limits, you can specify values outside this range, but you should know that you are on shakey ground at that point.

That said, I freely admit that the obtuse derivation of the trans parameters in particular has caused a good deal more consternation than if I had used more usual values of Rd, Td, Rs, and Ts and simply noted that these coefficients must sum up to something less than 1. Oh, well.

-Greg

···

From: Georg Mischler <[email protected]>
Date: March 10, 2005 8:47:19 AM PST

Axel Jacobs wrote:

Dear all,

I am trying to nail this one down, once and for all.
What is your experience/approach? Feedback welcome.

refman:
"The transmissivity is the fraction of penetrating light that travels all
the way through the material.
The transmitted specular component is the fraction of transmitted light
that is not diffusely scattered.
Transmitted and diffusely reflected light is modified by the material color.
Translucent objects are infinitely thin."

BIG QUESTION:
How do CAPITAL indeces (RADIANCE) relate to
lower case indices (PHYSICS)??? See below. Also for a lot more little
questions...

  http://www.schorsch.com/rayfront/manual/transdef.html

Scroll down a bit for the diagram.

-schorsch

Wow, that's a great diagram, Schorsch. I think I understand the
"trans" type for the first time, myself!

Couldn't agree more. Super! I might just like to your site, then.

Be sure also to refer to section 5.2.6 (pp. 325-6) in "Rendering with
Radiance," which translates between more sensible physical quanitites
and the parameters of the "trans" type.

I did, but called it radbook.

How about real-world examples for those of us that don't have a
goniphotometer in their kitchen cupboard?

Cheers

Axel

7 red green blue spec rough trans tspec
#7 A1 A2 A3 A4 A5 A6 A7

What is the plastic equivalent to a trans without transmittance?
Would that be _wrong_ or just _stupid_ to use?

Is trans with A6=0 equal to plastic? And could it be used interchangeably?

Axel

Sorry, Axel -- I should have read through your original posting more carefully. I missed your questions and references at the end:

Some Qs:
--------
The specular components (both diff and spec) are hard to measure.
However, this is what the calculation in the radbook is based on.
How would one make estimations?

I don't know of any *good* ways to measure these without some kind of photometer. A crude way to measure total transmittance would be to measure illuminance near a large white card, then again, covering the illuminance probe completely with your material. The same trick will work with a camera and Photosphere or hdrgen if you don't own a lux meter. Use your camera in the same way as the probe. Divide the average of the second image by the first using Photosphere (or hdrgen) and running "pvalue -h -H -d -b -o capture.hdr | total -m" (or "Edit -> Select All" in the image viewer of Photosphere).

For reflectance, you can try a similar trick if you know the reflectance of your white card, and this is described in RwR, I think.

Specular is more difficult to measure, but you can usually guess with reasonable accuracy. If a surface is a single dielectric interface (plastic or glass), it's specular reflectance will be around 4%. If it is a double interface, as in a pane of glass, it will be about twice that, unless the glass is very dark, where it will be somewhere between 4% and 8% depending on the amount of absorption.

As for roughness, I have a nice little device with LEDs and a viewport that allow me to estimate this value, but I don't know how to offer it to anyone as it's a piece of hardware, and requires a lot of experience to use effectively.

Is this only possible through visual comparision?
What are the steps to A1..A7 if given:
- reflectance (luxmeter + luminance meter)
- transmittance (2x luxmeter)
(practical scenario, assuming grey colour)
if
- Lambertion properties or
- clearly see-through
What is the plastic equivalent to a trans without transmittance?

I'm not sure I understand the question. If you set A6 to 0, trans is the same as plastic.

Would that be _wrong_ or just _stupid_ to use?

If you mean substituting trans for plastic to be general, there's no harm.

-Greg

···

From: "Axel Jacobs" <[email protected]>
Date: March 10, 2005 10:50:30 AM PST

Wow, that's a great diagram, Schorsch. I think I understand the
"trans" type for the first time, myself!

Couldn't agree more. Super! I might just like to your site, then.

Be sure also to refer to section 5.2.6 (pp. 325-6) in "Rendering with
Radiance," which translates between more sensible physical quanitites
and the parameters of the "trans" type.

I did, but called it radbook.

How about real-world examples for those of us that don't have a
goniphotometer in their kitchen cupboard?

Cheers
Axel

A P.S. to my last message -- the reasonable value ranges for the basic material types are given in the document called "materials" under ray/doc/notes, or from the website at:

  http://radsite.lbl.gov/radiance/refer/Notes/materials.html

-G

Greg,

Specular is more difficult to measure, but you can usually guess with
reasonable accuracy. If a surface is a single dielectric interface
(plastic or glass), it's specular reflectance will be around 4%. If it
is a double interface, as in a pane of glass, it will be about twice
that, unless the glass is very dark, where it will be somewhere between
4% and 8% depending on the amount of absorption.

Is this the so-called Fresnel reflection you're talking about? It's 4% at
right angles. I'm with you...

As for roughness, I have a nice little device with LEDs and a viewport
that allow me to estimate this value, but I don't know how to offer it
to anyone as it's a piece of hardware, and requires a lot of experience
to use effectively.

I wonder if it was possible to just measure the size of the hilight and
derive the roughness from there, given a known light source diametre and
distance, e.g. sun? Well, it's not going to be have sharp edges, but one
could take an HDR image and a luminance profile through it that finds the
50% intensity...

Is this only possible through visual comparision?
What are the steps to A1..A7 if given:
- reflectance (luxmeter + luminance meter)
- transmittance (2x luxmeter)
(practical scenario, assuming grey colour)
if
- Lambertion properties or
- clearly see-through
What is the plastic equivalent to a trans without transmittance?

I'm not sure I understand the question. If you set A6 to 0, trans is
the same as plastic.

What I means is almost exactly what you described above:
To measure the reflectance: refl = L * PI / E. Measuring the luminance and
the illuminance for the same spot (even illuminance assumed) gives us a
figure for the reflectance
To measure the transmittance: trans = E_behind_material /
E_in_front_of_material, ideally taken with 2 lux meters, but under
relatively constant illumination and with some averaging over several
readings, it will work.

Question is: how far does this improvisation get us with the trans material?

Axel

Hi Axel,

I wonder if it was possible to just measure the size of the hilight and
derive the roughness from there, given a known light source diametre and
distance, e.g. sun? Well, it's not going to be have sharp edges, but one
could take an HDR image and a luminance profile through it that finds the
50% intensity...

Well, you can take an HDR photo and use a known source and geometry to fit the distribution, but it's a fair amount of work.

What are the steps to A1..A7 if given:
- reflectance (luxmeter + luminance meter)
- transmittance (2x luxmeter)
(practical scenario, assuming grey colour)
if
- Lambertion properties or
- clearly see-through
What is the plastic equivalent to a trans without transmittance?

I'm not sure I understand the question. If you set A6 to 0, trans is
the same as plastic.

What I means is almost exactly what you described above:
To measure the reflectance: refl = L * PI / E. Measuring the luminance and
the illuminance for the same spot (even illuminance assumed) gives us a
figure for the reflectance
To measure the transmittance: trans = E_behind_material /
E_in_front_of_material, ideally taken with 2 lux meters, but under
relatively constant illumination and with some averaging over several
readings, it will work.

Question is: how far does this improvisation get us with the trans material?

Well, it's OK, but you still don't know how much of the transmitted light is specular and how much is diffuse. You could try putting a tube on your lux meter (or using a luminance probe) and couple with a collimated source (a good quality flashlight will do in a pinch) and back the source and probe away from the material. Subtracting the measured specular (direct) transmittance from the previous integrated measurement will give you the approximate diffuse value.

This is all very, very rough.

-Greg