RADIANCE Digest Volume 2, Number 7





Dear Radiance Users,

It has been several months since the last digest was posted, and in that
time there have been a number of developments.

First, I would like to mention that a paper was published on Radiance
in the 1994 issue of Computer Graphics (the annual Siggraph
proceedings), which contains a lot of useful information on the history
and algorithms of this package.  Also, the CD-ROM version of the
proceedings has an HTML version of the paper, as well as the 2.4 source
code distribution.  The HTML version is also linked to on our new
Radiance Web page, which is:

	ftp://hobbes.lbl.gov/www/radiance/radiance.html

There you will find many other goodies, such as online documentation and
images.  If you have a Web site with Radiance-related pages, please write
to me so I can link it in to our page.

Another announcement I would like to make is the start of an unmoderated
discussion group for Radiance users, which will be accessible from the alias:

	radiance-discuss@hobbes.lbl.gov

If you wish to subscribe (or unsubscribe) to this group, please send your
name and most permanent e-mail address to:

	radiance-request@hobbes.lbl.gov

Please do not mail the general group with administrative requests!
This has happened in the past, and it is nothing but a pain to
everyone.  Mail sent to radiance-request currently gets forwarded to
me.  If you are on the moderated mailing list (which you must be
if you just received this digest), that does not automatically place
you on the discussion group list.  You must subscribe separately.
If you wish to unsubscribe from the moderated group, please send
a message to radiance-request to that effect.

One last thing about the discussion group list -- since it is a simple
mailing list and my mailer's not too bright, you may get some bounced
mail when you post to this group.  You should either ignore such
messages or forward them to radiance-request so that I may update the
list.  Please, do not complain to the discussion group, because the
problem members won't even get your complaints, and everyone else will
tire of reading them!  Bounced mail is usually caused by out-of-date
e-mail addresses, which is why it's important to give me the most
permanent address you can on your initial subscription, and inform
me (at radiance-request) whenever your address changes.

Finally, I may submit a Siggraph course proposal on Radiance,
and would like to know how many people would sign up for such a course,
and whether they'd prefer a half-day or full-day adventure.  Please
write back to me (at GJWard@lbl.gov) with your opinions.

Without further ado, here are the topics covered in this digest:

	NOISE_FUNCTIONS			- Fractal vs. Perlin noise functions
	DAYLIGHTING			- Various daylight-related topics
	PATTERNS			- Mapping patterns onto surfaces
	AMBIENT_FILES			- Using ambient files
	VIEW_ANGLES			- Computing -vh and -vv parameters
	INTERACTIVE_WALKTHROUGHS	- Generating interactive walkthroughs
	SPECTRAL_COLORS			- Multi-spectral sampling
	X11_ERROR			- Rview "cannot open command line..."
	TRANS_MATERIAL			- Setting parameters for "trans" type
	PARTICIPATING_MEDIA		- Smoke and fog simulation
	MATERIALS			- Several related materials questions

Enjoy!
-Greg

=========================================================================
NOISE_FUNCTIONS

To: greg@hobbes.lbl.gov
Subject: quick jiggle questions
Date: Wed, 27 Apr 94 11:06:47 EDT
From: Philip  Thompson <phils@MIT.EDU>

Hi Greg,

What is the difference between jigglepic_u and fjigglepic_u
in jigglepic.cal? or fnoise3() and noise3() for that matter?
(is it fast or floating point?)

Thanks,
Philip


Date: Wed, 27 Apr 94 09:31:54 PDT
From: greg (Gregory J. Ward)
To: phils@MIT.EDU
Subject: Re:  quick jiggle questions

Hi Philip,

The difference is that noise3(x,y,z) is the Perlin noise function and
fnoise3(x,y,z) is a fractal noise function with otherwise similar
characteristics.  Ideally, you would create such a function using
a summation of Perlin functions with a 1/f frequency spectrum, but
this is rather expensive in practice so I have written a special
version.  Fractal noise usually does a better job mimicking natural
phenomena.

-Greg

========================================================================
DAYLIGHTING

Date: 27 Apr 1994 12:03:12 -0400 (EDT)
From: "Richard G. Mistrick" <RGMARC@ENGR.PSU.EDU>
Subject: rtrace for daylighting
To: greg@hobbes.lbl.gov

Greg:
    We are attempting to use rtrace to analyze daylighting in empty rooms.  We
have four walls and a window as defined in the example included in the manual.
We are interested in illuminance on the work plane as well as at partially
shielded and unshielded photocells.  The photocells are modelled with a small
obstruction surrounding a point to block the view in a certain direction.  Our
main problem is one of computation time when we attempt to perform accurate
rtrace runs (as indicated in the "SETTING RENDERING OPTIONS" table that you
sent to me.  
    In a situation such as this, how can we best gain speed without losing
accuracy.  I assume that it is best to manipulate the -a? commands.  Room size
that we are planning to model are 15x15 ft and 30x60 ft.  I am also concerned
that we achieve an accurate analysis of light arriving at the shielded
photocell point.  Can you give us some advice on what parameters will most
affect speed and accuracy by describing how -aa, -ar, -ad and -as affect what
is occuring in the computation of the interreflected component.  A simple
description of what Radiance is doing and how these values affect the process
would be helpful.  A calculation at only one point is running for 20 hours on
our system (SGI Idigo/Iris workstation) using the "Accurate" input parameters
(-ab 3) and we are looking for a way to reduce this time.
    Thanks in advance for your help.

Regards,

Rick Mistrick 


Date: Wed, 27 Apr 94 09:44:12 PDT
From: greg (Gregory J. Ward)
To: RGMARC@ENGR.PSU.EDU
Subject: Re:  rtrace for daylighting

Hi Rick,

In an empty room, many of the options may be relaxed without significant
loss in accuracy.  I recommend you do the following:

	1. Make sure your window is an illum, as described in the tutorial.

	2. Reduce the ambient divisions (-ad) to 256.

	3. Reduce the ambient super-samples (-as) to 0.

	4. Reduce the ambient resolution (-ar) to 16 if you have no
		groundplane, or the groundplane size divided by your
		maximum room dimension divided by 16 otherwise.

	5. Increase the ambient accuracy (-aa) to 0.2.

	6. Set the direct jitter (-dj) to 0.7.

	7. Set the direct substructuring (-ds) to 0.2.

	8. Set the ambient value (-av) to something sensible, probably
		around .3 .3 .3 in your situation.

Even with your current settings, I am a bit surprised that the calculation
is taking so long.  Do you have any external structures modeled?  Is your
window a light source?  Are you really just calculating a single point
using rtrace -I?  Perhaps you can send me your model.

-Greg


Date: Fri, 06 May 1994 12:55:21 -0400
To: greg@hobbes.lbl.gov (Gregory J. Ward)
From: stuart@archsun.arch.gatech.edu (Stuart Lewis)
Subject: Glazing Light Loss Factors

Hi Greg,

Thanks again for your help with my last set of questions;  I now have 
another one for you!

I am in the process of writing a little daylighting program for Architecture 
students based on the Lumen Method for Daylighting (IESNA RP-23) and am 
seriously considering using Radiance to develop some data that will let us 
add correlation factors for non-standard (ie not  0.50!) wall 
reflectivities.  We thought it would be very useful for the students to be 
able to visualize (and quantify!) the relationship between surface color and 
interior illumination.  I have gotten pretty good agreement between the 
manual results and Radiance, but ran into a question which seems to _maybe_  
have wider relevance:

How would you handle the Light Loss Factor ("accounting for dirt 
accumulation") of the glazing?  Initially, I simply reduced the workplane 
illumination levels by that amount (as the IES does), but it seems to me 
that dirt accumulation might actually affect the behavior of a 
view-preserving window (causing it to become somewhat diffuse.)  Or, do you 
think it would be adequate to simply reduce the transmittance of the glazing 
by that amount?  I haven't seen anything in the literature we have available 
that would clue me in. 

Since we began using Radiance for physical modelling, it seems this question 
is more relevant in that context than in reproducing the reults from RP-23 
(adding this feature to our program is an unanticipated bonus!)  I'm 
thinking for example about clerestory windows or toplighting, where dirt 
accumulation can become quite significant.
------

Also, we haven't received any of the requested information on WINDOW or DOE 
that you were going to pass along.  If you don't mind, would you please pass 
along that request again?  This fax number may be more reliable: 
404-458-4090  (Max Akridge's home!)

Thanks a lot,
Stuart

[Stuart Lewis, GRA
College of Architecture
Georgia Institute of Technology
Atlanta, GA 30332]


Date: Fri, 6 May 94 10:32:17 PDT
From: greg (Gregory J. Ward)
To: stuart@archsun.arch.gatech.edu
Subject: Re:  Glazing Light Loss Factors

Hi Stuart,

I have forwarded your request again as a reminder to those who manage DOE-2
and SUPERLITE.  I hope this time they will respond.

I must admit I haven't given much thought to dirt accumulation on windows.
Certainly, it is possible to model this with Radiance, using either the
BRTDfunc type, transdata, transfunc, mixfunc with trans and glass, or
trans alone (in order of increasing simplicity and decreasing generality).

Since I have no data on how dirt affects the transmission of glazing,
I am left to guess.  My guess is that it adds some diffuse component,
which may be a function of incident angle.  What are the common ranges
and how important is incident angle are two questions I could only
answer by taking measurements.  I have a device that could be used
to such a purpose, but I have not calibrated it yet for transmission
(only reflection).

All the windows experts seem to be gone at the moment -- maybe there's
a staff meeting or something.  Anyway, I'll ask them when they come
back and write again if I get any good research pointers for you.

-Greg


From: rcl@scs.leeds.ac.uk
Date: Thu, 18 Aug 94 13:56:49 BST
To: GJWard@lbl.gov
Subject: Radiance - daylight

I am fairly new to the radiance package so please forgive my
ignorance in such matters...

1) You state in digest v2n4 that EXPOSURE = K * T * S / f^2, where
T = exposure time (in seconds), S = film speed (ISO), f = f-stop,
K = 2.81 (conversion factor 179*PI/200) and that this is approximate for
35mm photography. Is the 35mm assumption important for this equation or is
the equation true for, say, CCD cameras (which I am trying to model)? This
assumes of course that I can equate the responsivity of a CCD array with
a certain ISO film speed.

2) I am trying to get an accurate representation of a scene lit purely by
skylight (generated either by gensky or gendaylit). I have defined my sky
and ground as follows (adapted from the tutorial); (Note I am only
viewing the outsides of buildings)

    !gensky 7 15 12 -a 54 -o 2 -m 0 -g .2

    skyfunc glow ground_glow
    0
    0
    4 1.6 .8 .25 0

    ground_glow source ground
    0
    0
    4 0 0 -1 180

    skyfunc glow sky_glow
    0
    0
    4 .8 .8 1 0

    sky_glow source sky
    0
    0
    4 0 0 1 180

	I then define a finite 'ground plane' on which my buildings stand and
    cast any shadows onto.

Does having the ground defined as a glowing hemisphere provide a realistic
representation? I have tried defining the ground as large disc but I always
end up gaving a black gap in the horizon. I am more interested in modelling
reality than what may 'look' correct to the viewer.

3) When I have the above as my only source of lighting rpict warns me that
I have no light sources. Is this correct? What am I doing wrong to generate
such an error?

Thanks in advance for any help you can provide, and I apologise again for
anything fundamental that I may have missed in the documentation.

Rob Love

School of Computer Studies
University of Leeds
Leeds
England


Date: Thu, 18 Aug 94 11:51:56 PDT
From: greg (Gregory J. Ward)
To: rcl@scs.leeds.ac.uk
Subject: Re:  Radiance - daylight

Hi Rob,

In answer to your questions:

1) The exposure suggested is only if you wish to reproduce a photographic
image.  I don't know how to correlate ISO to CCD sensitivity, and even if
I did, most cameras have such things as automatic irises and automatic gain
that make any efforts to pin down the exposure futile.  The main question
is, "why do you want to do this?"  Are you really trying to reproduce camera
output, or are you simply trying to set the right exposure?  If it's the
latter, then there are much better ways.

2) Your sky representation is fine.  If there are no significant shadows
in your exterior environment cast by large structures, then the distant
ground approximation is adequate.  If there are significant shadows, then
simply ADD a groundplane to this description.  Do not remove your distant
ground approximation, or you will get that dark horizon problem.

3) I do not know why the sky you showed me would give you the "no light
sources found" warning.  That should only happen if you use a cloudy sky
or other description w/o a sun.  Anyway, this is merely a warning, and
indicates that without an indirect calculation (i.e. -ab >= 1), you will
have no illumination at all.

Hope this helps.
-Greg


From: rcl@scs.leeds.ac.uk
Date: Thu, 18 Aug 94 20:46:28 BST
To: greg@hobbes.lbl.gov
Subject: Re: Radiance - daylight

I am actually working in computer vision and I'm using Radiance to 
generate some test sequences of the sun moving over some buildings during
a complete day. Eventually I aim to be using a CCD camera to capture these
sequences but I am using Radiance at the moment as I have complete control
over the environment (it's always overcast in Leeds), the images are noise
free, and it's very accurate. I would ideally like Radiance to give me 
images that would be the same/similar to what I expect to get from a CCD 
camera. 

Since I am dealing with varying lighting my CCD camera will have gain control,
gamma correction, iris control, etc turned off. This enables me to compare
images taken at different times of the day without the camera attenuating
for lack of light.

So, ideally I'd like to be able to set Radiance's exposure so that it
mimicks a certain setting on my camera (f-value, shutter speed, CCD
responsivity) and seeing your equation for film based photography gave me
some hope. Equating CCD sensitivity to film speed is my problem but I was
unsure if the 35mm asumption in the equation was necessary - would the
equation still hold if I was using a 120 size film (for example)?

I am certainly not concerned with setting the exposure so it 'looks' right.
It's what what the computer will see that is important.

Many thanks for your prompt reply
Rob


Date: Thu, 18 Aug 94 13:07:05 PDT
From: greg (Gregory J. Ward)
To: rcl@scs.leeds.ac.uk
Subject: Re: Radiance - daylight

Hi Rob,

All you need do is calibrate your camera against your renderings somehow.
You can do this if you look at an evenly illuminated surface and adjust
the speed/f-stop to a good exposure setting.  Capture an image, then
the captured values (average) will tell you the relationship between
illuminance and image for these settings.  (I forgot to mention that
you must measure the light level with a luminance or lux meter.)

For Radiance, the exposure is simply a multiplier between radiance
(luminance/179) and pixel value.

The formula for photography applies to your camera, in the sense that
exposure is inversly related to the square of f-stop, and speed is
of course a linear relationship.

-Greg


=========================================================================
PATTERNS

From: "Mr. M.J. Lupton" <sk8@liverpool.ac.uk>
Subject: Patterns from tiff files
To: greg@hobbes.lbl.gov
Date: Fri, 29 Apr 1994 10:42:10 +0100 (BST)

Greg

I have just ftp'd some textures from ccu1.auckland.ac.nz in tiff format.

I then used ra_tiff with the -r option to convert them to pic format, but
when I come to use the pic file as a texture using this arrangement of
commands:
void colorpict blockwork_pat
9 red green blue block1.pic picture.cal tile_u tile_v -s .5
0
1 1

blockwork_pat plastic concrete_blocks
0
0
5 .55 .55 .5 0  0

!genbox concrete_blocks northwall 9 0.06 6 | xform -t 1 15 .02

the pattern only appears to be on one face of the box?

What am I doing wrong

Martin Lupton. Liverpool University.


Date: Fri, 29 Apr 94 09:13:42 PDT
From: greg (Gregory J. Ward)
To: sk8@liverpool.ac.uk
Subject: Re:  Patterns from tiff files

Hi Martin,

Radiance doesn't automatically rotate a pattern onto various faces of
a solid, at least not with the tile_u and tile_v variables defined in
picture.cal.  You either have to rotate the pattern to make 3 types for
three surface orientations (XY, YZ and XZ planes), or modify the U and V
variables defined in rayinit.cal to do it for you.  Something like the
following should work:

void colorpict blockwork_pat
9 red green blue block1.pic . frac(U) frac(V) -s .5
0
0

blockwork_pat plastic concrete_blocks
0
0
5 .55 .55 .5 0  0

!genbox concrete_blocks northwall 9 0.06 6 | xform -t 1 15 .02

-Greg


=========================================================================
AMBIENT_FILES

Date: Tue, 03 May 1994 14:48:25 +1200 (NZT)
From: mat@ccu1.auckland.ac.nz (M Carr)
Subject: ambient files
To: GJWard@lbl.gov

Hi Greg

Just a pair of questions

If I am rendering two views of a single scene at once, and they are
both using the same ambient file, will I have any problems?

What exactly are ambient files? Being view independant, they still
change quite dramatically between views (hence my concern above), ie
they grow.


Mat
______________________________________________________________________________
Matiu Carr

School of Architecture Property and Planning
University of Auckland
New Zealand

email: m.carr@auckland.ac.nz


Date: Thu, 5 May 94 22:48:08 PDT
From: greg (Gregory J. Ward)
To: mat@ccu1.auckland.ac.nz
Subject: Re:  ambient files

Hi Mat,

The answer is that you needn't worry.  The values will be shared to the
extent possible in the two renderings.  Especially if you are doing more
than one bounce (-ab 2 or greater), many values will be common and you
WILL save a lot of time.  It is true that the file will continue to grow,
but that's the price you pay for speed.

And, if you're using 2.3 or later, you can have as many rpict processes
sharing the same ambient file on as many machines as you like.  Provided
you have a working NFS lock manager (which is by no means certain), the
file will be updated in a consistent fashion and values will be shared
among processes.  Read the document in /pub/doc/parallel.txt on hobbes.lbl.gov
for more details.  If you have Mosaic, you can access these through our HTML
pages starting at "ftp://hobbes.lbl.gov/www/radiance/radiance.html".

-Greg


=========================================================================
VIEW_ANGLES

Date: Mon, 16 May 1994 16:23:28 +1200 (NZT)
From: mat@ccu1.auckland.ac.nz (M Carr)
Subject: rpict -vh -vv
To: GJWard@lbl.gov

Hi Greg

Is there a formula for working out what a particluar pair of -vh -vv
settings will produce in terms of the final image dimensions, there
does not seem to be a linear correspondence.

I am just using the vtv view setting.

It's not urgent.

Thanks

Mat
______________________________________________________________________________
Matiu Carr


Date: Mon, 16 May 94 12:28:22 PDT
From: greg (Gregory J. Ward)
To: mat@ccu1.auckland.ac.nz
Subject: Re:  rpict -vh -vv

Hi Mat,

The relationship between perspective view angles and image size is
determined by tangents, i.e.:

	tan(vh/2)/tan(vv/2) == hres/vres

Note that the angles must be divided in half (and expressed in radians
if you use the standard library functions).  If you know what horizontal
and vertical resolution you want, and you know what horizontal view angle
you want (and your pixels are square), you can compute the corresponding
vertical view angle like so:

	% calc
	hres = 1024
	vres = 676
	vh = 40
	vv = 180/PI*2 * atan(tan(vh*PI/180/2)*vres/hres)
	vv
(resp)	$1=27.0215022

Thus, -vh 40 -vv 27 -x 1024 -y 676 should result in (almost) no
adjustment of the horizontal or vertical resolutions by rpict.
If you had just taken a rough guess of the vertical view
angle, rpict would shrink the horizontal or vertical image size
to insure that the pixels were square.

Hope this helps.
-Greg


=========================================================================
INTERACTIVE_WALKTHROUGHS

From: lpostner@cs.clemson.edu
Date: Thu, 19 May 94 13:19:55 EDT
To: gjward@lbl.gov

I have recently installed Radiance2R4 on a system of SGI Indys 5.2.
I was interested in rendering 3D models and walking around them 
interactively for use with VR helmets.  Is there a way of doing this
with Radiance?  If not how easy would it be to write an interactive
viewer?  I am not interested in the raytracing aspect, but rather 
the radiosity part of Radiance.  Anny and all suggestions would be
greatly appreciated.
Thanks.

Lori Postner 
Dept. of Computer Science
Clemson University
email: lpostner@cs.clemson.edu


Date: Thu, 19 May 94 10:26:51 PDT
From: greg (Gregory J. Ward)
To: lpostner@cs.clemson.edu
Subject: VR and Radiance

Hi Lori,

There is no program that comes with Radiance for interactive walk-throughs.
In fact, I think this problem is too difficult to solve interactively in
the general case.  The best you can do is a diffuse approximation, as used
in radiosity-type programs.

The one way to get what you want with Radiance is on a SGI Reality Engine
or some such that does real-time textures, then use Radiance to compute
illumination maps.  I haven't done this myself, but I know some folks in
Zurich who have if you need some more pointers.

-Greg


From: lpostner@cs.clemson.edu
Subject: Re: VR and Radiance
To: greg@hobbes.lbl.gov (Gregory J. Ward)
Date: Thu, 19 May 94 13:53:02 EDT

How do I get Radiance to dump me a 3D data file?  Also, please send me
the addresses of the people in Zurich, perhaps they will have some good
suggestions.  Thanks

Lori Postner
Department of Computer Science
Clemson University
email lpostner@cs.clemson.edu


Date: Thu, 19 May 94 11:05:05 PDT
From: greg (Gregory J. Ward)
To: lpostner@cs.clemson.edu
Subject: Re: VR and Radiance

Hi Lori,

The fellow who has done this is Daniel Lucius, and his address is
<lucius@arch.ethz.ch>

You cannot dump full 3-d illumination data from Radiance.  It is necessary
instead to compute illumination maps individually for each surface.  From
my limited experiments, it seems most efficient to compute the irradiance
using a parallel view of each wall, floor and ceiling, then combine this
with the appropriate colors and textures during rendering.

For example, let's say you have the following wall:

wall_mat polygon east_wall
0
0
12
	10	0	0
	10	0	3
	10	7	3
	10	7	0

You would run rpict like so to compute an illumination map with a resolution
of 0.1 units (meters?):

	rpict -i -vtl -vp 9.99 3.5 1.5 -vd 1 0 0 -vh 7 -vv 3 -vu 0 0 1 \
		-ab 1 [etc...] -x 70 -y 30 octree > east_wall.pic

The -i option tells rpict to compute irradiance instead of radiance.

You will have to adjust the brightness and convert the output picture
into the appropriate image format (using fromrad or something), then
write a GL program to use it, along with the other maps.

-Greg


=========================================================================
SPECTRAL_COLORS

From: Carlos F. Borges <borges@waylon.math.nps.navy.mil>
Subject: Radiance question
To: GJWard@lbl.gov
Date: Mon, 8 Aug 94 14:59:49 PDT

Dear Greg,

  I enjoyed your talk in Orlando very much.  I am interested in using
the Radiance package to try different color representation approximation
methods and was] wondering what is the color modelling methodology of
Radiance.  Do you use, or allow, full spectral description of lights and
reflectances (transmittances) or is some kind of RGB method used.  Is it
possible to change the manner in which color computations (independent
of scene geometry) are done?  Where can I find out more on this subject
before I start trying to use the package.  I am trying to find a system
that allows me to change the underlying color modelling to see how well
different simplification schemes work (like the one in my 1991 SIGGRAPH
paper, or different spectral sampling approaches). My email address is:

  borges@waylon.math.nps.navy.mil

--
Cheers,

  Carlos


Date: Mon, 8 Aug 94 15:23:34 PDT
From: greg (Gregory J. Ward)
To: borges@waylon.math.nps.navy.mil
Subject: Radiance color rep.

Hi Carlos,

Unfortunately, I was rather stupid in my initial use of color in Radiance,
and now I'm more or less stuck with a 3-sample representation.  These
normally correspond to RGB, though you can define them to be whatever
you like (e.g. XYZ).

Nevertheless, certain desperate people have used Radiance to compute
images with more spectral samples by rendering the same image several
times, changing only the materials.  This is of course not the most
efficient route to take, but it does work if nothing else is available.

It would be possible to alter the source code to handle more spectral
samples, but there are several places where 3 samples are assumed.
(Most of the code is in the form of macros, which can be easily changed
in contrast.)  The main reason I have still not generalized the code
is because it would mess up the input format rather badly.  I may
yet think of a way around this problem, but until I have a greater
need and a little time to do it...

If you want to use a scene description language that provides full
color flexibility, why don't you investigate the MGF format described
on our anonymous ftp site (hobbes.lbl.gov) in the /www/mgf directory?
There you will also find HTML documents for Mosaic.  (i.e. try the URL:
"ftp://hobbes.lbl.gov/www/mgf/HOME.html".)

-Greg


=========================================================================
X11_ERROR

Date: Tue, 16 Aug 1994 10:30:19 -0700 (MST)
From: AGMXR@acvax.inre.asu.edu
Subject: Help with rview - radiance utility
To: GJWard@lbl.gov

Hi,
I have i9nstalled Linux on one of the Pentium PC and compiled radiance on
it.  When I tried to display the 'oct' using rview, it gives me a message
"rview: cannot open command line window".  Can you help to resolve this 
problem.  
Thankyou very much.
Muthu
AGMXR@ACVAX.INRE.ASU.EDU


Date: Tue, 16 Aug 94 10:46:34 PDT
From: greg (Gregory J. Ward)
To: AGMXR@acvax.inre.asu.edu

Hi Muthu,

The most probable cause of the command line window not opening is that
the X11 driver is not finding the default text font, "8x13".  You
can modify the x11.c file in the ray/src/rt directory to use a font
you DO have, or figure out some way of making this font available.
One way to do this is in your X11 fonts directory, there should be
a file called "fonts.alias", in which you may add an entry for 8x13,
and alias it to another font that you do have.

Hope this helps.
-Greg


=========================================================================
TRANS_MATERIAL

Date: Thu, 1 Sep 1994 05:08:22 -0700
From: COURRET Gilles <courret@divsun.unige.ch>
To: greg@hobbes.lbl.gov
Subject: translucent material

Hi greg,
I am working on zenithal opening Radiance simulation, and espacially
on translucent plastic glazing. Can you confirm that :
"
void trans opale
0
0
7 .3 .3 .3 0 0 .6 .1
"
is effectively a grey translucent material which has
a transmission fraction of 60% with a specular component of 10%.
Thanks in advance,
Yours,

Gilles

Date: Thu, 1 Sep 94 13:33:55 PDT
From: greg (Gregory J. Ward)
To: courret@divsun.unige.ch
Subject: Re:  translucent material

Hi Gilles,

So I see that you get confused by the trans type, just as I do!

If what you want is no reflection from the surface whatsoever, and
60% total transmission, 10% of which is specular (leaving 50% diffuse),
you should use:

void trans opale
0
0
7 .6 .6 .6 0 0 1 .1666

I hope this is what you're after.  What you gave me was:

void trans opale
0
0
7 .3 .3 .3 0 0 .6 .1

which would have 12% diffuse reflectance ((1-.6)*.3), 16.2% diffuse
transmittance (.3*.6*(1-.1)), and 1.8% specular transmittance (.3*.6*.1).

I'm sorry that this type is so confusing.  I get horribly confused by
it myself...

-Greg


Date: Fri, 2 Sep 1994 04:55:26 -0700
From: COURRET Gilles <courret@divsun.unige.ch>
To: "(Gregory J. Ward)" <greg@hobbes.lbl.gov>
Subject: Re: translucent material

Hi greg

Thanks for your fast reply.
With the new material definition you gave me the results are much more realistic!
The daylight factor is much higher.
But i have the feeling that it is a little bit to high.
Are you sur about trans=1 for a transmission of 60% ?
Yours,

Gilles


Date: Fri, 2 Sep 94 09:09:31 PDT
From: greg (Gregory J. Ward)
To: courret@divsun.unige.ch
Subject: Re: translucent material

Hi Gilles,

Yes, I am sure about trans=1 for transmission of 60%, because the color
was set to .6 .6 .6, which both transmitted components (specular and diffuse)
are multiplied by.

-Greg


=========================================================================
PARTICIPATING_MEDIA

Date: Tue, 6 Sep 94 13:59:57 -0400
From: randal.sims@srs.gov (Randy Sims)
To: GJWard@lbl.gov
Subject: Radiance vs. Participating Medium

I am pursuing computations of global illumination in the presence of
participating media.  The assumption is that absorption, emission and
scattering events in the environment result in a radiance along any
direction that changes with the position along that direction. Such
media might include soot, dust, smoke, fog, etc.  Past treatments of
such phenomena include Rushmeier and Torrance's extended radiosity
methods (Siggraph'87) and direct Monte Carlo simulations.

But, other than my interests, I'm sure you are well aware of these
issues, phenomena and treatments. I have browsed the Radiance digests
and find little discussion of such effects.  Is there any work and/or
research with Radiance in this area?  Can Radiance be extended to
address these phenomena?  Do you (individually and/or collectively --
the Radiance community) have an interest in addressing participating
media?

Randal N. Sims (Randy)
Westinghouse Savannah River Co.
Savannah River Site
773-42A, 129
P.O. Box 616
Aiken, SC 29802
USA

TEL: (803)725-8347
FAX:   (803)725-8829
Email  randal.sims@srs.gov


Date: Tue, 6 Sep 94 11:24:03 PDT
From: greg (Gregory J. Ward)
To: randal.sims@srs.gov
Subject: Re:  Radiance vs. Participating Medium

Hi Randy,

I have only recently begun to dabble in participating media, following
some advice and articles from Holly Rushmeier on the topic.

I'm afraid that Radiance is not very well equipped to deal with this
problem in its current state.  I had toyed with the idea of modeling
the particles themselves, since Radiance can handle even ridiculous
scene complexity, but I think the resulting simulation would be so
slow that it wouldn't be worth the bother.

Instead, I have been using a shortcut that accounts only for some
interactions, such as absorption of direct light and along eye rays,
as well as approximate scattering from light sources to the eye.
I do this by combining a change in the direct lighting calculation
with a post-process using pcomb.  If you are interested in seeing
the results, I have just one picture of the lower deck of a ship
with a uniform distribution of soot.  (Non-uniform distributions
might be modeled with this method, but the relation would have
to be fairly simple.)  Just let me know what format you want the
image in, and I'll drop it in the /xfer directory on hobbes.lbl.gov.
(It's there right now as a Radiance picture if that will do.)

-Greg


=========================================================================
MATERIALS

Date: Mon, 12 Sep 1994 23:08:48 -0500
From: Dana Peters <peters@ERC.MsState.Edu>
To: greg@hobbes.lbl.gov
Subject: newbie

Greetings,
	As the subject says I am a new user and have some very basic questions.
I am currently an architecture student and have been using Wavefront for three
years to create architectural walkthroughs and various other animations.  I am
familiar with very general rendering principles, but have very little 
experience with the graphics programing principles and the actual physical
calculations.  I do, however, have a basic understanding of C.  I have 
installed Radiance, read the manual, worked the tutorial, and read most of the
digests.  Ok, now for the questions

1.  I have no problem with the modeling, but I do need some help with
	materials.  I understand the basic material types (plastic, metal, 
	glow, light) but have problems thereafter, specifically those
	that require func files.  What types of real materials have 
	ansiotropic roughness used in plastic2, metal2, etc?  What is the 
	benefit of using this material type?    What, in layman's terms, does
	a bidirectional reflectance distrobution function do to a surface?
	Can you give me some expamples of real materials that this is used 
	to represent?  I also have some problems with the difference between
	dielectrics and glass, but I think I can figure that out on my own.

2.  I would like to start using Radiance extensively because of its wonderful
	and accurate renderings.  Will it be necessary for me to learn how to
	write function definitions myself, or can I get by using those supplied
	in the software and those written by others?  Learning how to write 
	these would be helpful, but I currently have no idea how to define
	procedural textures, etc.  

3.  In the digests someone mentioned the possibility of having a radiance
	training class.  Has this actually happened in the past?  will it 
	happen in the future?  I am sure that a class would be the best way
	for me to learn radiance. (aside from hiring a personal tutor)

4.  Concerning the files at your ftp site... are most of these included in the
	distrobution?  for example, I noticed several libraries and some 
	objects.  Are these included somewhere or should I pick them up?
	Also what exactly are all those things in the tests/empty/...
	directory?  just curious.

Well, I guess that is enough for now.  Sorry to bother you with such basic 
questions.  Perhaps you could refer me to a good book or paper that might 
explain some of the basic stuff that I am lacking.  There is no need to hurry
with answers to these questions.  I am not in any rush.  

thanks in advance,
-Dana Peters
peters@erc.msstate.edu


Date: Tue, 13 Sep 94 10:37:11 PDT
From: greg (Gregory J. Ward)
To: peters@ERC.MsState.Edu
Subject: Re:  newbie

Hi Dana,

Good questions, all.  It sounds like you have already investigated the
Radiance Digest archives in the /pub/digest/ on hobbes.lbl.gov.  Did you
know that these have also been collected and indexed by Veronika Summerauer
in HTML format, and are available on the Web from the Radiance page at:

	ftp://hobbes.lbl.gov/www/radiance/radiance.html

Anyway, I don't think I've answered your questions before, so I'll attempt
to do justice to them here.

> 1.  I have no problem with the modeling, but I do need some help with
> 	materials.  I understand the basic material types (plastic, metal, 
> 	glow, light) but have problems thereafter, specifically those
> 	that require func files.  What types of real materials have 
> 	ansiotropic roughness used in plastic2, metal2, etc?  What is the 
> 	benefit of using this material type?    What, in layman's terms, does
> 	a bidirectional reflectance distrobution function do to a surface?
> 	Can you give me some expamples of real materials that this is used 
> 	to represent?  I also have some problems with the difference between
> 	dielectrics and glass, but I think I can figure that out on my own.

Yes, materials are difficult, aren't they?  Plastic2 and metal2
are appropriate for surfaces like varnished wood and brushed or rolled metal
-- anything that has elongated highlights.  Note that you don't necessarily
have to use a function file for these types.  If the brushed direction is
aligned with a vector and the surface is relatively flat, you can simply
use a constant vector in the place of the "ux uy uz" variables, e.g:

	void metal2 brushed_aluminum
	4 1 0 0 .
	0
	6 .7 .7 .7 .85 .02 .08

Since the .02 value corresponds to the roughness along the [1 0 0] vector,
and the .08 value is the roughness in the perpendicular direction, the
above material has a highlight that is narrower in the X direction.

A bidirectional reflectance-transmittance distribution function (BRTDF,
often called BSDF for bidirectional scattering distribution function),
is a general function describing how light interacts with a surface material.
You should use it only when all the other material types fail, as it is the
most difficult to apply and the least efficient type in Radiance.  One
example where it might be needed is velvet, which has very peculiar
reflectance properties.  Retroreflective materials are another example.

The difference between dielectric and glass is simply that glass simulates
two close, parallel, dielectric surfaces.  Glass is more efficient since
it approximates the internal reflections in closed form rather than
computing all the many rays that two dielectric surfaces would require.

> 
> 2.  I would like to start using Radiance extensively because of its wonderful
> 	and accurate renderings.  Will it be necessary for me to learn how to
> 	write function definitions myself, or can I get by using those supplied
> 	in the software and those written by others?  Learning how to write 
> 	these would be helpful, but I currently have no idea how to define
> 	procedural textures, etc.  

Unfortunately, this is something I have never documented.  Suprisingly, a
number of users have figured it out on their own and written procedural
patterns and textures and contributed their work to the /pub/libraries/
directory on hobbes.lbl.gov.  I suggest that you look at these as well
as the .cal files distributed with Radiance and try to learn by
example if you need to roll your own.

> 
> 3.  In the digests someone mentioned the possibility of having a radiance
> 	training class.  Has this actually happened in the past?  will it 
> 	happen in the future?  I am sure that a class would be the best way
> 	for me to learn radiance. (aside from hiring a personal tutor)

Yes, there was a workshop held for about 20 people in Spring of 1991, and
another this Spring for only 5 people in Germany (disappointing turnout,
but the tuition was rather high to pay for my travel out there).  I am
thinking of holding a Siggraph course on Radiance either next year or
the year after.

> 
> 4.  Concerning the files at your ftp site... are most of these included in the
> 	distrobution?  for example, I noticed several libraries and some 
> 	objects.  Are these included somewhere or should I pick them up?
> 	Also what exactly are all those things in the tests/empty/...
> 	directory?  just curious.

The main distribution contains source code, documentation and examples not
found elsewhere on the ftp server.  The /pub/* directories contain mostly
Radiance-related contributions, and are distributed by tape to people who
don't have ftp access.  In most cases, there are README files describing
the contents of each directory.

The /pub/tests/ directory was designed to contain comparisons between
Radiance and other global illumination calculations, but contributions
over the years have been rather disappointing.

-Greg

Back to Top of Digest Volume 2, Number 7
Return to RADIANCE Home Page
Return to RADIANCE Digests Overview


All trademarks and product names mentioned herein are the property of their registered owners.