RADIANCE Digest Volume 3, Number 3


Radiance Digest Volume 3 Number 3        October 10, 1997

The latest version of Radiance is 3R1P5.

Archives can be found at: ftp://radsite.lbl.gov/pub/digest
and   ftp://radsite.lbl.gov/pub/discuss

Index of Topics
---------------------------------------------------------
LATEST PATCH FILES AVAILABLE
AUTOCAD AND DXF TRANSLATORS
RADIANCE FOR WINDOWS NT
LIGHTS INSIDE BOXES WITH SMALL HOLES
USING MKILLUM FOR SIDELIGHTED OFFICE
LARGE AREA SOURCE GLARE CALCULATIONS
PROCEDURAL MODELING WITH GENSURF
AQUARIUM ANALYSIS
MODELING WIRE MESH
SPECULAR LIGHT SHELF
'EXPOSING' IMAGES FOR INDOOR AND OUTDOOR BRIGHTNESSES
---------------------------------------------------------

LATEST PATCH FILES AVAILABLE

>From radiance  Tue Sep  2 18:21:04 1997
Date: Tue, 2 Sep 1997 18:14:37 -0700
From: radiance (Radiance Account)
To: raydisc
Subject: New Radiance patch now available


Radiance users,

A new patch is now available at the Radiance WWW site:

   http://radsite.lbl.gov/rad/patch/

The changes include:

3.1.4 patch files:
src/rt/ambient.c        - fixes problems with -aw option
src/cv/ies2rad.c        - improvement for direct/indirect fixtures
src/cv/source.cal       - improvement for direct/indirect fixtures
src/px/x11image.c       - enhancement of 'h' and 'a' commands

3.1.5 patch files:
src/cal/util/Rmakefile - fixes missing compatibility flags
src/rt/persist.c       - fixes problem with AIX select() differences
src/util/netproc.c     - fixes problem with AIX select() differences
src/common/color.h     - fixes problem with AIX macro definitions
src/common/lookup.h    - fixes problem with AIX macro definitions

Instructions for installation are in the README File.  

Thanks Greg!

-Chas
---------------------------------------------------------------------------
AUTOCAD AND DXF TRANSLATORS


>From akienyy@nus.sg  Fri Aug 29 11:01:15 1997
Date: Sat, 30 Aug 1997 01:56:39 +0800
To: radiance
From: akienyy@leonis.nus.sg (Edward Ng)

Hi

As a new user of radiance, I have heard people saying that there is a
autocad/dxf to radiance translator somewhere written by a Phil Thomson. Any
idea where I can locate that or are there other offerings somewhere?

Edward

________________________________ ________________________________
Dr Edward Ng   .........  reply to (akienyy@nus.sg)
School of Architecture, National University of Singapore
Kent Ridge Crescent, Singapore 119260 Tel: 65-7723567 Fax:7793078
________________________________ ________________________________


>From radiance  Tue Sep  2 13:49:44 1997
Date: Tue, 2 Sep 1997 13:49:43 -0700
From: radiance (Radiance Account)
To: akienyy@leonis.nus.sg (Edward Ng)
Subject: ACAD to Radiance
Cc: radiance


Dear Dr. Ng,

Although answering this questions always makes me a little nervous because
of my personal interest, I will venture to provide an answer that is as
complete as possible.  Any omissions are unintentional and I am eager
to hear about additional solutions to this problem from the Radiance
community.

There are a few options for exporting and/or translating geometry from
AutoCAD and DXF files into Radiance, none of them perfect.  The AutoLISP 
"torad" program is available from the Radiance ftp site:

   ftp://radsite.lbl.gov/rad/pub/translators/torad.tar.Z

Torad is slow because it is AutoLISP and will only run within versions of
AutoCAD that support AutoLISP (is AutoLISP still supported in R14?).

There is another "dxf2rad" lisp program in the same ftp directory, but
as far as I can remember, it does not support as many entity types.

The "radout" program is a close cousin of torad and because it is
written in "C", it is much faster.  If you have the Unix version of
AutoCAD on a Sun workstation, then you can get a version of radout 
which is part of the "DDRAD" package from Georg Mischler at:

   http://www.schorsch.com/autocad/radiance.html

{The last time I checked this site, it wasn't working.  Give it a try
in a few days before giving up.}

And there is a version of "radout" available for AutoCAD Release 12
DOS and Windows3.11.  It is mostly the same as torad except it does not
support the export of views (this never works well in torad anyway),
but, it includes support for CLOSED 3D PLINES.  This capability means
that any other entity type that degenerates to a 3D PLINE is also
supported.  This includes most SOLIDS.  Sometimes this means "exploding" 
the higher-order geometry, sometimes even this is not necessary.  Radout 
for DOS/Windows is available for $50.00 from:

   http://www.innernet.com/radiance/radout.htm

Lastly, the ADELINE software package contains an MS-DOS executable
program that can translate DXF files into Radiance with varrying degrees
of success.  It will aparently work with ACAD Ver 13, but it does
not support the new V13 3D entities.  For more information about
the ADELINE package, please see:

   http://radsite.lbl.gov/adeline

This is by no means an endorsement for any of these packages and your
success may vary.  If anyone else would like to be mentioned, please let
me know.

Happy exporting,
-Chas
+---------------------------------------------------------------------------+
| Charles Ehrlich                                     Phone: (510) 486-7916 |
| Principal Research Associate                        Fax:   (510) 486-4089 |
| UC Lawrence Berkeley National Laboratory            Contact person for:   |
| 1 Cyclotron Road MS: 90-3111, Berkeley, CA 94720    RADIANCE and ADELINE  |
+---------------------------------------------------------------------------+

-------------------------------------------------------------------
RADIANCE FOR WINDOWS NT

>From MREANEY@KUHUB.CC.UKANS.EDU  Wed Aug 27 13:56:05 1997
Date: Wed, 27 Aug 1997 15:53:42 -0500 (UTC -05:00)
From: Mark Reaney <mreaney@KUHUB.CC.UKANS.EDU>
Subject: Radiance on an Alpha box?
To: gregl@asd.sgi.com

Greg,

A novice question.

We would like to do some experiments with Radiance, but most of our
computers are pretty wimpy.  We may have the oportunity to use a very
speedy NT Workstation that uses Alpha processors.

Does the NT version of Radiance run on Alpha?  Is there some way to get
the unix version to run on it?

Thanks for any advice.
Mark Reaney

======================================================================
     Mark Reaney, i.e.VR         http://www.ukans.edu/~mreaney       
Institute for the Exploration         
   of Virtual Realities              ()    ___   \\  //|| \\
 Dept. of Theatre & Film             ||   //_\\   \\// || //
     Univ. of Kansas                 ||() \\__ ()  \/  || \\
    mreaney@ukans.edu                     
======================================================================


>From radiance  Tue Sep  2 14:06:34 1997
Date: Tue, 2 Sep 1997 14:06:33 -0700
From: radiance (Radiance Account)
To: mreaney@ukans.edu
Subject:  Radiance on an Alpha box?
Cc: radiance


Dear Mark,

I enjoyed my brief visit to your WWW site.  I can see the use of
Radiance in your work enhancing the realism you have already 
achieved with the tools you have at your disposal.  A whole audience
of people wearing head-mounted displays, eh?  Quite impressive!

There is a version of Radiance which will run on Wintel machines,
but it has not been compiled for Alpha NT.  The Wintel version
is part of the ADELINE package which is available from LBNL
at a cost of $450.00.  Please see our WWW pages for more info:

   http://radsite.lbl.gov/adeline

It is not optomized for NT, but does run well as a DOS shell 
application.  In fact, it runs 40 to 70% faster that under DOS 6.22!

I also suggest you contact your esteemed colleagues in the lighting
department.  Professor Moeck and his students did a fantastic job
of using Radiance to render a church as part of a student competition.
You can find their WWW site at:

   http://www.arce.ukans.edu/arce/681ld/daylight/dosdoc2.htm

Sincerely,
-Chas
+---------------------------------------------------------------------------+
| Charles Ehrlich                                     Phone: (510) 486-7916 |
| Principal Research Associate                        Fax:   (510) 486-4089 |
| UC Lawrence Berkeley National Laboratory            Contact person for:   |
| 1 Cyclotron Road MS: 90-3111, Berkeley, CA 94720    RADIANCE and ADELINE  |
+---------------------------------------------------------------------------+

---------------------------------------------------------------
LIGHTS INSIDE BOXES WITH SMALL HOLES

>From sunil@teil.soft.net  Mon Aug 25 22:34:34 1997
Date: Tue, 26 Aug 1997 10:56:18 -0530
From: Sunil S Hadap <sunil@teil.soft.net>
Organization: Tata Elxsi (India) Ltd.
To: Radiance Account <radiance>, raydisc, radiance
Subject: Set oferflow error!


Dear Radiance Users,

We are developing Radiance support for Alias and is almost over. We model a
sceen for the beta test and we get error "set overflow" somthing related
MAXSET which is set 127 currently. The error is at that part of the scene
where number of polygons (greater than 256) in small region. In perticular
the scene contains a cylender of small size order 10 cm when overall scene
is 12x8 meters. I increased MAXSET (sorry) to 1024 it works for some cases
but again fails in some other cases. Can any one suggest what is the error
and remedy for it.

Thanking You

Sunil Hadap

Tata Elxsi (India) Ltd.
fn:             Sunil Hadap
n:              Hadap;Sunil
org:            Tata Elxsi (India) Ltd.
adr:            Whitefield Road, Hoody,;;Tata Elxsi (India) Ltd.;Bangalore;Karnataka;560048;India
email;internet: sunil@teil.soft.net
title:          Specialist, Visual Computing
tel;work:       +91-80-8452016/17/18
tel;fax:        +91-80-8452019


------

>From tcvc@indiana.edu  Tue Aug 26 09:45:18 1997
Date: Tue, 26 Aug 1997 11:40:34 -0500 (EST)
From: "robert a. shakespeare" <tcvc@indiana.edu>
To: Jean-Louis Maltret <jlm@eiffel.univ-mrs.fr>
cc: tcvc@falstaff.ucs.indiana.edu, raydisc
Subject: Re: Set overflow error


       I suggest that you break the scene into components and create
a seperate octree file for each. These component octree files can then be
combined with either instance commands or using the -i flag in oconv.

       Using this method, I have been able to eliminate most set overflow
problems and have not had to eliminate small details from huge scenes.

-Rob

Rob Shakespeare, Director
Theatre Computer Visualization Center
Room 200, Theatre Building             V/Fax: 812-855-8827
Indiana University                     tcvc@indiana.edu
Bloomington, IN 47405                  shakespe@indiana.edu
http://appia.tcvc.indiana.edu/~tcvc

Announcement:
RENDERING WITH RADIANCE: THE ART AND SCIENCE OF LIGHTING VISUALIZATION
       by Greg Ward Larson and Rob Shakespeare
       Morgan and Kaufmann Publishers (December 1997)


--------------

Date: Tue, 26 Aug 1997 08:01:24 -0700
From: greg@pink (Gregory W. Larson)
To: Sunil S Hadap <sunil@teil.soft.net>
Subject: Re:  Set oferflow error!
Cc: radiance-discuss

We are developing Radiance support for Alias and is almost over. We model a
sceen for the beta test and we get error "set overflow" somthing related
>From Sunil Hadap of Tata Elxsi (India) Ltd.
>
> MAXSET which is set 127 currently. The error is at that part of the scene
> where number of polygons (greater than 256) in small region. In perticular
> the scene contains a cylender of small size order 10 cm when overall scene
> is 12x8 meters. I increased MAXSET (sorry) to 1024 it works for some cases
> but again fails in some other cases. Can any one suggest what is the error
> and remedy for it.

If you read the oconv manual page, it talks about this error, and the
solution is to increase the -r command-line parameter value by factors
of two.  If you take your overall scene dimension and divide by the
smallest region in which you expect to have a compound object, this
arrives at he appropriate setting for this parameter.

If you run into memory problems with this solution (I could have sworn
we just discussed this in the latest digest), you can put your small,
complex objects into separate octrees and populate the scene using
the instance primitive.  This is a really good idea if your small
geometry is highly repetitive, like identical pieces of furniture
scattered about.

-Greg
_____________________________________________________________________
Gregory Ward Larson (the computer artist formerly known as Greg Ward)
Silicon Graphics, Inc.                   Computer Science Department
2011 N. Shoreline Blvd., M/S 07U-553     537 Soda Hall, UC Berkeley
Mountain View, CA  94043-1389            Berkeley, CA  94720-1776
(415) 933-4878, -2663 fax                (510) 642-3631, -5775 fax
gregl@sgi.com                            on Tues., Thurs. and Fri.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~



>From sunil@teil.soft.net  Wed Aug 27 03:29:58 1997
Date: Wed, 27 Aug 1997 15:55:40 -0530
From: Sunil S Hadap <sunil@teil.soft.net>
Organization: Tata Elxsi (India) Ltd.
To: Radiance Account <radiance>, raydisc, radiance
CC: sunil@teil.soft.net
Subject: Slightly dirty image, help me to improve! (HTML)


Dear Radiance Users,

The image is modelled using LightGen tool. But we can't render it better.
The rendering is with high quility and spends hours. The rif file is
attached.

Following points can be of concern

    * Point lights used are very small (radius .05m) and intensity [254 253
229]
    * The light sources are spherical and are enclosed in box type
enclosure with one face open which throws light. There is no direct
lighting in the room except small portion of the room. All the light
received by the room is due to interreflections. Mind room color is not
green but almost gray. [0.7056      0.72     0.710823]
    * The rif file says room is with detail medium.
    * VAR is medium, we use logic: there is no sunlight hence it is not
high and we see lightsource directly hence can not be low.
    * PENUM is FALSE, we don't want jitter to further spoil the picture.

We don't understand the spotches, and we can not remove them even if we
increase -ad 1024 and -as 512. Also there is strange glow on the edges of
the walls and pot looks as if made of florescent material. Shadow below the
table is brown ting. We don't think it's a reality. Image in the mirror
looks speckled when roughness is very small (though not zero) 0.001

Sorry for HTML mail. I am including the image.


view= Camera -vf hallOfVision/hughes_Camera.vf
QUALITY= H
AMBFILE=hallOfVision/hughes.amb
RESOLUTION= 512
ZONE=I  -1e-06  4  -1e-06  4.00054  -1e-06  3
PENUMBRA=FALSE
render= -t 1 -ab 2
DETAIL=M
VAR=M
oconv=-n 3
scene=hallOfVision/hughes_l.rad hallOfVision/hughes_s.rad
hallOfVision/hughes_o.rad

Thank You

Sunil Hadap
Mehaboob Ali

Tata Elxsi (India) Ltd.

----------

>From tcvc@indiana.edu  Wed Aug 27 07:40:43 1997
Date: Wed, 27 Aug 1997 09:36:00 -0500 (EST)
From: "robert a. shakespeare" <tcvc@indiana.edu>
To: Sunil S Hadap <sunil@teil.soft.net>
Subject: Re: Slightly dirty image, help me to improve! (HTML)


       This is just a stab in the dark (no pun intended!) but by locating
the -ab 2 in the render option, rad does not take 2x interreflection into
account when generating the rendering process.
       I would suggest removing the -ab 2 from render and including:

INDIRECT= 2

       Then take a look at the rad variable settings and compare them
with what you were running previously....

               rad -n -e your.rif

By including INDIRECT=2 you will likely find that rad will recreate your
ambient file with data from a small image using a setting of -aw 0.
Then using this file, your final rendering is produced with a setting of
-aw xxx. I would think that you might  continue to overide the -ad and -as
automatic settings, replacing them with your higher values.

Though this is s subtle change, it might improve the outcome.

As per the greenish ambient light and glowing quality:

... I assume that you are using a white light source. As the wall is
slightly weighted to green and blue, the greenish hue  might be due
to overexposure. I would expect that the blue vase, though insignificant
to a broadly lighted scene,  will effect the scene in your case.
 Try a red vase and see what changes!!

 I would not expect the undertable to appear so "bright"
so again, you might reduce the exposure a stop, which in turn, might help
to elliminate the "glowing" quality.

I would be interesting to see how pcond would modify the appearance of
the image... after rerendering you might try:

pcond -h- test.pic > testpd.pic


Rob Shakespeare, Director
Theatre Computer Visualization Center
Room 200, Theatre Building             V/Fax: 812-855-8827
Indiana University                     tcvc@indiana.edu
Bloomington, IN 47405                  shakespe@indiana.edu
http://appia.tcvc.indiana.edu/~tcvc

Announcement:
RENDERING WITH RADIANCE: THE ART AND SCIENCE OF LIGHTING VISUALIZATION
       by Greg Ward Larson and Rob Shakespeare
       Morgan and Kaufmann Publishers (December 1997)

---------

Date: Wed, 27 Aug 1997 09:05:43 -0700
From: greg@pink (Gregory W. Larson)
To: Sunil S Hadap <sunil@teil.soft.net>
Subject: Re:  Slightly dirty image, help me to improve! (HTML)
Cc: radiance-discuss

The main problem is sounds like your light sources.  You must instead of
placing your spheres in boxes and expecting the interreflection calculation 
determine their output, use small polygons at their openings and eliminate
the spherical sources.

This will cure most of your problems.  You may have to manually set the
-aw parameter to 0 in the "render" variable of rad if you still see strange
glowing edges.

-Greg

----------------------------

USING MKILLUM FOR SIDELIGHTED OFFICE
 

>From Markku.Norvasuo@vtt.fi  Tue Aug 12 04:39:46 1997
Date: Tue, 12 Aug 1997 14:37:50 +0300 (EET DST)
To: ckehrlich@lbl.gov
From: Markku Norvasuo <Markku.Norvasuo@vtt.fi>
Subject: Adeline/Radiance


Hi Charles,

The Adeline package has been delightful and well working. However, getting
acquainted requires some time as you said.

Concerning Radiance I have a few questions: 

1) I have been modeling a sidelit office space. Is it reasonable to define
window surfaces as illum sources if blinds (from genblind) are at the same
time used beside them? I.e. does this arrangement behave correctly?

2) Should I use illum sources, anyway, to include sky glow in the interior
illumination, or is this glow sufficiently included if the option -ab > 0?

3) I put the office scene in the mkillum files box in the Adeline/Rad user
interface. When the original octree was named R013.oct, rview generated two
additional ones, R0130.oct and R0131.oct. If I use rpict instead from a
command line, is it correct to give the last one (R0131.oct) as the input
octree? (I did not find a clear advice in the manual but this seems to work). 

Do you have a mailing list similar to Radiance Digest or should I subscribe
the latter one?

Thank you for advice

Markku 
________________________________________________________________
Markku Norvasuo, Senior Research Scientist
Technical Research Centre of Finland, Building Technology
Postal address: 
VTT Building Technology, P.O.Box 1801, FIN-02044 VTT, Finland
Tel. +358 9 456 6269 (office), +358 40 515 1100 (mobile)
Fax  +358 9 456 6251, Internet: markku.norvasuo@vtt.fi


------------

>From chas@pink  Tue Aug 12 09:38:27 1997
Date: Tue, 12 Aug 1997 09:35:35 -0700
From: chas@pink (Charles Ehrlich)
To: Markku Norvasuo <Markku.Norvasuo@vtt.fi>
Subject: Re:  Adeline/Radiance
Cc: radiance@pink

Markku,

I'm glad to hear that you are not having as much problems with
ADELINE as some of my other users.

1. The main purpose of the mkillum feature is to provide a means
to define an average distribution for complex fenestration.
Therefore, the most efficient way of modeling a sidelit office
with blinds is to put the blinds on the "outside" of the illum
surface.  This may require that you create an "imposter" polygon
with material type "void" which serves this purpose.  It can
sometimes be challenging to find ways to create the imposter
geometry in a way that minimizes the potential inaccuracies
which can be introduced with the use of illum.  In particular,
if direct sunlight strikes only a portion of your "imposter"
or "illum" surface, the distribution for the whole window will
be the average, but this may not produce satisfactory results.

2. In most cases you will want to use mkillum to define the
diffuse component of the sky when you have complex glazing (blinds,
light shelf, etc.). If you have a more simple glazing (no blinds), 
then ab > 0 works fine.  In effect, using mkillum reduces the 
number of bounces necessary to acheive adequate results by one.
Unfortunately, mkillum can sometimes cause "banding" when very
close to the illum.

3. The "rad" command handles the two-step octree generation in
the way you discovered.  The latter octree will contain the
scene geometry plus the glazing or imposter surfaces which have
been sampled by mkillum.  

4. I have been thinking about the fact that the promised user
support e-mail list for ADELINE never coagulated, mostly because
I was doubtful that there were enough users of the software to
make it work.  Perhaps now that I am in charge of the Radiance
e-mail lists, I could combine ADELINE with the rest of the
Radiance users.  Perhaps I could also create a separate list
which is for ADELINE-specific needs like Superlite, scribe,
plink, etc, and keep the discussion of Radiance issues open
to ADELINE users in the Radiance forum.  Hmmmm.

Good luck.  Please share with me some images when you have 
something satisfactory to show.

-Chas
Charles Ehrlich
Principal Research Associate

-------------------------------------

LARGE AREA SOURCE GLARE CALCULATIONS


>From haico.schepers@arup.com  Mon Jul 21 22:49:28 1997
Date: Tue, 22 Jul 1997 05:41:00 +0000
From: Haico Schepers <haico.schepers@arup.com>
To: greg
Subject: Re:  Can you help with this one? -Reply

Greg

Thank you for your last reply, I've been reading the journals you
referenced in the manual and it seems to imply that the program does
take into account large sources.  I tried to email LESO in Switserland for
confirmation at  lesoweb@lesosun1.epfl.ch, but I have had no success
getting through.  Do you know of a person or email account I can contact
to ask these questions to?
<<How does radiance performs glare calculations for day lighting
analysis and how this is best modelled?  For a simple room with large
windows how does Radiance calculate glare?  Does it take into account
large sources as indicated by Guth (and referenced in the Radiance
documentation)?  What are the algorithns thatit uses?  Has any work
been done to validate this?>>
>From a technical stand point it would be of great benifit to know what
algorithns radiance uses to calculate glare indices.

 Also with reference to your reply;
> 1/ I can't find xglaresrc to run the visualization program for Glare

Doesn't exist in ADELINE, since it's tied to X11.

Is there any way to get a copy of the script used for this program and
can I update Adeline2 at all to run this script?

 I recently read your paper on A Visiblity Matching Tone Reproduction
Operator for High Dynamic Range Scence.  Our company is largely
using radiance to model natural daylight for passive solar design.  We
are greatly concerned with in occupant comfort (with regard to glare)
and their perception of the work place environment thus any
developments in how a scene is veiwed is of great interest.  Can you
inform me of any developments in this feild and whether it has been or
will be incorporated into adeline2.

Any information on this topic would be greatly appreciated.

With kind Regards
Haico Schepers

-----------

>From greg  Tue Jul 22 09:51:44 1997
Date: Tue, 22 Jul 1997 09:51:43 -0700
From: greg (Gregory W. Larson)
To: Haico Schepers <haico.schepers@arup.com>
Subject: Re:  Can you help with this one? -Reply

Hi Haico,

I dug up an old document I'd written in 1992 after I wrote the glare script
and programs in Radiance, and I put it on our mirror web site (still under
development).  You may access it directly over the web at:

       http://radsite.lbl.gov/mirror/radiance/man_html/Notes/glare.html

This should get you partway there, but for the full scoop on daylight glare
calculations using Radiance, you should talk to Raphael Compagnon, who is
on sabbatical at Cambridge University, and may be reached at "rjrc2@cam.ac.uk".


There really is no easy way to incorporate the xglaresrc program into Adeline,
since it relies on X11 to circle glare sources in a displayed image.  It's
not as simple as translating a script, and I doubt anyone is putting the
work into porting this particular utility, though you can write to Charles
Ehrlich at "radiance@floyd.lbl.gov" if you want to ask about it.  While
you're at it, you can also ask him about the porting of pcond (the tone
reproduction program I wrote for 3.1) to Adeline.  There is no reason in
principle why this should pose a problem, but the development plans on
the PC-compatible platform are still in flux and I have nothing further
to do with them as I no longer work at LBNL.

-Greg
_____________________________________________________________________
Gregory Ward Larson (the computer artist formerly known as Greg Ward)
Silicon Graphics, Inc.                   Computer Science Department
2011 N. Shoreline Blvd., M/S 07U-553     537 Soda Hall, UC Berkeley
Mountain View, CA  94043-1389            Berkeley, CA  94720-1776
(415) 933-4878, -2663 fax                (510) 642-3631, -5775 fax
gregl@sgi.com                            on Tues., Thurs. and Fri.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

--------------------------------------
PROCEDURAL MODELING WITH GENSURF


>From panop@xarch.tu-graz.ac.at  Tue Jul  8 05:07:27 1997
From: "Christos Panopoulos" <panop@xarch.tu-graz.ac.at>
To: greg (Gregory J. Ward)
Date: Tue, 8 Jul 1997 14:05:42 +0000
Subject: create a quarter cycle

Hi Greg,

how can I create surface of cycle segment (angle = PI/2) ?
And then extrude this.Thank you in advance.


Christos

Christos Panopoulos         E-Mail:panop@xarch.tu-graz.ac.at 
http://xarch.tu-graz.ac.at/~panop

>From greg  Tue Jul  8 09:29:42 1997
Date: Tue, 8 Jul 1997 09:29:41 -0700
From: greg (Gregory W. Larson)
To: panop@xarch.tu-graz.ac.at
Subject: Re:  create a quarter cycle

Use the gensurf command.  The x(s,t) y(s,t) and z(s,t) are independent
parametric functions for surface indices (s,t) each running between 0 and 1.
Define them however you like.

-Greg
_____________________________________________________________________
Gregory Ward Larson (the computer artist formerly known as Greg Ward)
Silicon Graphics, Inc.                   Computer Science Department
2011 N. Shoreline Blvd., M/S 07U-553     537 Soda Hall, UC Berkeley
Mountain View, CA  94043-1389            Berkeley, CA  94720-1776
(415) 933-4878, -2663 fax                (510) 642-3631, -5775 fax
gregl@sgi.com                            on Tues., Thurs. and Fri.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

----------------------------------------------------
AQUARIUM ANALYSIS


>From chrisg@GLASS.lplizard.com  Fri Sep  5 10:03:27 1997
Date: Fri, 5 Sep 1997 12:58:24 -0400
From: Chris Green <chrisg@lplizard.com>
To: radiance
Organization: Leaping Lizard Software, Inc. (301)-963-8230 FAX (301)-963-9016

       Do you know if anyone has ever used radiance to render an
aquarium environment? I am designing a very large salt-water reef
aquarium aquarium for my home, and lighting is a big issue. I wish to
be able to try out different combinations of halide bulbs of differing
wattages and color temperature with different reflector types, and see
the lighting effect on the tank and the room viewing it. The surface
ripples of the water also have a big effect on the way the light
looks. It's kind of a worst-case caustic generator.

       Things I'd like to be able to do:

       See some pictures that would show what it might look like.

       Try different bulbs in different combos and placements. I'm
not sure if the data needed by the simulation is available for the
higher temperature halide bulbs used for SW aquariums.

       Know the amount of PAR (photosyntehtically active radiation)
hitting a given point inside the tank. This would be used to figure
out what areas were adequate for which kinds of corals. PAR is the
number of photons hitting an area that are in the photsynthetically
active wavelengths. Because of the ripples, I'd need it averaged over
time.

       Try different types of glass for the front panel to see how
whether I need to use special fancy glass to not have too much of a
greenish tint. I could also see the effect of different coverings for
the back and sides.

       Plan the layout of the reef structure in the tank and be able
to see the shadow effects.

       I would appreciate anything that you could tell me about the
feasability of this particular application, or if you know of anyone
who has done anything similar.

       Thanks,


-- 
Chris Green

-------------------------------------------------

Dear Chris,

Radiance is probably as close as you'll come to a tool for
the simulations you need.  Short of a full-blown forward ray 
tracer like Breault Research's ASAP software (which is no 
easier to use than Radiance and doesn't produce nice images), 
the one limitation of Radiance that you'll have to live with 
is the caustics.  Radiance can do a wavy surface of
water, but you won't know if you have a problem with high
concentrations of light where light rays are converging.
This seems like a minor effect to me because the surface of
the water in aquaria are more or less flat, right?  Of more
interest might be the bubbles.

Greg's "podlife" sculpture is a good example of a model that
uses water and bubbles.  It can be found at:

  ftp://radsite.lbl.gov/pub/models/podlife.tar.Z  

The issue of caustics came up a year or two ago and is documented 
either in the Radiance Digests or Discussion group archives
which can be found at:

  ftp://radsite.lbl.gov/pub/digest   ../discuss

Search for caustic and you should find more information.

A great improvement in Radiance since the digest postings is
that you can now model single-scatter "participating media"
which in the case of an aquarium would be particles of dirt
and/or haziness in the water.  This material entity is
called "mist."

What you'll need to do for the photon counting business
is develop a correlation between the photosyn-whatever
sensitivity you're after and Radiance's Red, Green and
Blue spectral samples.  It will be easiest to represent
this value as an illuminance, so you'll have to correlate
numbers of rays to illuminance as well.  Take a look at
the file called rlux inside the Radiance distribution.  
It is a C-shell script which calculates photopic illuminance.
You'll have to tweak the multipliers for R, G and B to
get what you want.

Your other tasks all seem very reasonable.  The most difficult
task will be modeling the geometry of the coral.

And, once you've gone through all of this trouble to develop
an aquarium analysis tool, the great thing about Radiance
is that it can be licensed for worldwide distribution at a
cost of $10,000.00 so that you could sell it to other aquarium
afficionados, if you're into that sort of thing.

Good luck,
-Chas
+---------------------------------------------------------------------------+
| Charles Ehrlich                                  Phone: (510) 486-7916    |
| Principal Research Associate                     Fax:   (510) 486-4089    |
| UC Lawrence Berkeley National Laboratory         email: ckehrlich@lbl.gov |
| 1 Cyclotron Road MS:90-3111, Berkeley, CA 94720  Contact person for:      |
| http://radsite.lbl.gov/radiance  ../adeline      RADIANCE and ADELINE     |
+---------------------------------------------------------------------------+

Date: Fri, 5 Sep 1997 11:09:45 -0700
From: greg (Gregory W. Larson)
To: Chris Green <chrisg@lplizard.com>
Subject: Re: radiance and aquarium design

The sad answer is no, Radiance would not do the right thing, no matter how
you modeled the surface of the water.  Its backwards ray-tracing algorithm
depends on knowing the locations of the light sources in order to find them,
especially if they're nearly point sources!  There's very little hope with
the current version of simulating this environment correctly.

Every system has it's Achille's heel, and this is it for Radiance.  You would
need some kind of bidirectional ray-tracing system to do what you want.  You
might investigate Arris Inspirer as a possible solution to your problem, or
perhaps Specter, both from Integra corporation (http://www.integra.co.jp/eng/).

Good luck!
-Greg

---------------------------------------------------------------------
MODELING WIRE MESH

Note: this message is being forwarded to the entire Radiance
discussion group instead of the Radiance Modeling group because 
in the transition from Greg to Chas, the "radiance-model" list 
has been lost.  My appologies.

---------------

Date: Wed, 10 Sep 1997 07:59:31 -0500
From: "T. Prater" <squid@ksu.edu>
To: "'radiance-model@radsite.lbl.gov'" <radiance-model@radsite>
Subject: Wire Mesh


Hello Radiance users,

Recently we have been trying to use Radiance to simulate the
lighting conditions inside a small 'chamber'; this chamber
will eventually be used to grow a plant. Two sides of the
box will be covered in stainless steel wire mesh.

My question is: how do I model this wire mesh? Are there any
preexisting material definitions that would be useful in this
situation? Really, the most important properties to us would
be how much light (overall) is transmitted through and
reflected back from the surface of the mesh (in other words,
we are not that concerned with the subtle effects of indivi-
dual wires in the mesh).

We have conjectured that the BRTDfunc material would be the
best option. Does this sound right? If so, I need a bit of
help with that material. Are the 'rfdif', 'gfdif', ...'btdif'
parameters used internally by Radiance? Or, are you supposed
to use them in your function file? What, intuitively,
do these quantities correspond to? How about physically?

Thanks in advance for any help you can give...

Todd Prater
Dept. of Agronomy
Kansas State University
Manhattan, KS 66506

----------------
A reply from Chas:
Hello Todd,

With regard to your questions about a growing chamber and
the modeling of wire mesh.  While BRTDfunc would work, you
would have to subject your wire mesh sample to some quite
rigorous bi-directional reflectance and transmittance tests
in order to adequately utilize this material.  I would 
suggest instead, that you try the trans or trans2 materials.
The benefit of the trans_ materials is that a light-source
ray is traced from both the reflected and transmitted ray
directions (that is, a direct ray is used to calculate both
the transmitted specular and reflected specular component.)
While neither BRTDfunc or trans_ are simple to understand,
you can find examples of the use of each in the Radiance
digests and/or discussion group archives on our WWW site.

Have you spoken with your distant cousins at KU?  The
Prof. Martin Moek in the lighting department has used 
Radiance extensively.  He might be able to offer you some
additional advise.

Another posting to the discussion group which I will be
forwarding deals with the modeling of an aquarium.  The
individual is interested in calculating the amount of 
photosynthetic light reaching a particular surface.  While
I don't have the exact equation for this, you might also
be concerned with this quantity.  I suggest that you contact
him and possibly collaborate and let me know what you
discover.

Good luck,
-Chas
--------------------------


Date: Wed, 17 Sep 1997 17:45:32 -0700
From: greg@pink (Gregory W. Larson)
To: radiance (Radiance Account)
Subject: Re:  Wire Mesh
Cc: radiance-discuss, radiance, squid@ksu.edu

I would go with Chas' recommendation to use the "trans" type as the
most efficient solution.  It may not reproduce the individual mesh
elements, but neither would BRTDfunc.

If you know the basic reflectance of your mesh material and the percentage
of mesh to air in it, you can compute the parameters for the trans type.
See the URL:

http://radsite.lbl.gov/mirror/radiance/digests_html/v2n10.htm#TRANS_PARAMETERS

for some more hints.
-Greg


---------------------------------

SPECULAR LIGHT SHELF

Date: Tue, 16 Sep 1997 18:13:16 +0200 (MET DST)
To: radiance-discuss
From: Krzysztof Wandachowicz <wandach@sol.put.poznan.pl>
Subject: light shelf

Dear Radiance users,

I'm working on calculation of daylight in office room with internal
reflective shelf. My shelf looks like long (it is on the whole length wall
with windows) and wide (1 m) window sill (I don't know if the "window still"
is the right definition, I have found in dictionary two different
definitions: "window stool" and "parapet wall"). This shelf is only first
easy model of my system of daylight. I have more complicated idea to direct
the light from sun and sky to the ceiling, but I have used this shelf to
test the material. 

I need two different reflective materials: diffuse (Lambertian
characteristic) and specular (or/and semispecular) reflection. I have tested
the Greg's material (e-mail 12 Sep 95 from Greg to manuel@ise.fhg.de)
BRTDfunc for partially transparent shelf 55% reflection 25% transmission
(below). 

void BRTDfunc blind_grey
10 if(Rdot,.55,0) if(Rdot,.55,0) if(Rdot,.55,0)
       .25     .25     .25
       0       0       0
       .
0
9      0       0       0
       0       0       0
       0       0       0

void mirror blind_illum
1 blind_grey
0
3 .55 .55 .55

This example is good for specular reflection and direct transmission, but
how to change this description in order to obtain diffuse and semidiffuse
characteristic. I try many times without success.     


So, for diffuse reflection I chose "plastic" and "mirror":

void plastic blind_grey
0
0
5 .7 .7 .7 0 0

void mirror blind_illum
1 blind_grey
0
3 .7 .7 .7

Plastic type in this example describes diffuse reflection, mirror type
supports secondary light sources. I have made calculation for sunny day at
noon with windows on the south side. The light from sun and sky reflects in
my shelf and it is directed to the ceiling. This is diffuse reflection and
the ceiling should be illuminated almost uniformly but picture looks
differently.I see on the ceiling the virtual image of my windows with high
luminance. This is the light from the sun, the light from sky reflects in
shelf by directional way too. It strange because if I look at the ceiling I
think that shelf reflects daylight by directional way however if I look
directly at the shelf I'm sure that shelf reflects daylight by diffuse way.
I don't know if "mirror" is the right type of material but it is only type
for supports secondary light source. I have tried with "illum" type but
without success. The other difference is that the change of material type
and turn off or on sun doesn't influence the distribution of luminance.  

Time for the question. How describe diffuse reflection of such shelf? 

Now more complicated and additional question how to change the Greg's
description of BRTDfunc (above) to diffuse and semi diffuse reflection and
transmission. 

I will be grateful for any suggestions. 

The desperate Radiance user,

Krzysztof Wandachowicz.
############################################
# Krzysztof Wandachowicz                   #
# Poznan University of Technology          #
# Division of Lighting Engineering         #
# PL-60-965 Poznan, ul.Piotrowo 3a, Poland #
# fax +48 61 8782389                       #
############################################


--------------

Date: Tue, 16 Sep 1997 17:24:15 -0700
From: martin@color.arce.ukans.edu (Martin Moeck)
Subject: reflections
To: radiance-discuss

In response to Krzysztof Wandachowicz's questions, his modeling approach works fine for me. 

The diffuse component shows up on the ceiling when considering 

void plastic blind_grey
0
0
5 .7 .7 .7 0 0


and the specular component shows up when modeling a shelf of 

void mirror blind_illum
0 
0
3 .7 .7 .7 

without mixing the materials. If you mix them by defining 

void plastic blind_grey
0
0
5 .7 .7 .7 0 0

void mirror blind_illum
1 blind_grey
0
3 .7 .7 .7


both the diffuse component from the plastic and the specular component from the mirror are there. These components from the sun only work just fine. 

If you model the diffuse sky, the situation is a little bit more complicated. 
The ambient bounces from the diffuse plastic material are ok, but about the 
specular mirror and its effect on reflecting skylight I'm not so shure. 
It seems that the sky would have to be mimicked by creating an array of lights 
having the sky luminances of the corresponding patches. If you do 

void light skyglow
0  0 3
.96 .96 1.2

skyglow source sky
0 0 4
0 0 1 180

or something like that, a sharp image on the ceiling of that source is created. 
That does not do the job, because one light source with a center is assumed. 

Martin Moeck
mmoeck@ukans.edu

----------------------------------------------------------------
'EXPOSING' IMAGES FOR INDOOR AND OUTDOOR BRIGHTNESSES

>From el2gasae@uco.es  Wed Oct  8 05:52:35 1997
Date: Wed, 8 Oct 1997 14:49:10 +0200
From: Enrique Garcia Salcines <el2gasae@uco.es>
To: greg
Subject: A question

Hi Greg,

I'm working with Radiance in illumination design. I have a problem 
with the sun, the sky and the ground description that appear below. 
The model is a simple office with a great glass window. I have used 
the rad program with a view point inside the room. The problem is 
when I see through the window, the sky is not blue and the ground is 
not brown that is the RGB of these glow sources, all appear in white 
color like a very strong sun penetrating inside the room, the gensky 
options were "gensky 10 8 13".

void light solar
0
0
3 6.76e+006 6.76e+006 6.76e+006

solar source sun
0
0
4 -0.032656 -0.711654 0.701771 0.5

void brightfunc skyfunc
2 skybr skybright.cal
0
7 1 1.15e+001 2.17e+001 5.64e-001 -0.032656 -0.711654 0.701771

skyfunc glow sky_glow
0
0
4 0.90 0.90 1.15 0.00

sky_glow source sky
0
0
4 0 0 1 180

skyfunc glow ground_glow
0
0
4 1.40 0.90 0.60 0.00

ground_glow source ground
0
0
4 0 0 -1 180

void glass Glxxx0
0
0
3      0.950000        0.950000        0.950000

skyfunc brightfunc window_dist
2 winxmit winxmit.cal
0
0

window_dist illum Glass01
1 Glxxx0
0
3 0.88 0.88 0.88

Glass01 polygon cristal.0
0
0
12    ......the coordinates
----------------------------------------------------

Thanks in advance,


From: greg (Gregory W. Larson)
Message-Id: <199710081618.JAA03709@pink.lbl.gov>
To: Enrique Garcia Salcines <el2gasae@uco.es>
Subject: Re:  A question
Status: R

The reason the ground and sky appear white is due to exposure.  Just as
in a photograph, it is very difficult to obtain an exposure that correctly
presents both interior and exterior surfaces in their apparent colors.

Try using the pcond program on your rendered picture.  This program is
designed to compensate for these dynamic range limitations in the display,
and the result should be better in terms of perception.

-Greg

+---------------------------------------------------------------------------+
| PLEASE AVOID THE EVIL REPLY COMMAND, AND ADDRESS YOUR E-MAIL EXPLICITYLY! |
+---------------------------------------------------------------------------+
| Radiance Discussion list (unmoderated):  radiance-discuss@radsite.lbl.gov |
|*Radiance Digest mailing list (moderated):        radiance@radsite.lbl.gov |
| Radiance Modeling list (unmoderated):      radiance-model@radsite.lbl.gov |
| Requests to subscribe/cancel:            radiance-request@radsite.lbl.gov |
| Archives available from:             ftp://radsite.lbl.gov/rad/pub/digest |
| Radiance world-wide web site:            http://radsite.lbl.gov/radiance/ |
+---------------------------------------------------------------------------+

Back to Top of Digest Volume 3, Number 3
Return to RADIANCE Home Page
Return to RADIANCE Digests Overview


All trademarks and product names mentioned herein are the property of their registered owners.
All written material is the property of its respective contributor.
Neither the contributors nor their employers are responsible for consequences arising from any use or misuse of this information.
We make no guarantee that any of this information is correct.