Stars of Shareware: Raytrace & Morphing/SOS-RAYTRACE.ISO/programm/raytrace/radiance/doc/radiance.v13

Text File | 1993-09-03 | 44.6 KB | 1,077 lines

~s Radiance Digest, v1n3
Hello Radiance users,

I am trying to keep caught up with my correspondance better and
redistribute interesting tidbits before they turn to compost.
Here is the latest batch of letters on various topics that seemed
to me to be of general interest.  Once again, I have tried to make
it easier to find what you want by searching for the corresponding key.

    INST    Installation of 1.4 and associated problems
    MOLEC    Molecular modeling using Radiance
    PHONG    Phong surface normal interpolation
    INTERN    Radiance internals (image.c)
    PREVW    X11 previewer for Radiance
    SPEC    Spectral distributions curves
    DAYF    Daylight factors and unknown programs
    VISION    Vision-3D modeler for the MacIntosh

-Greg

=========================================================================
INST    Installation of 1.4 and associated problems

Date: Mon, 29 Apr 91 19:37:09 EDT
From: [e-mail] (Haydn Richardson)
Subject: Radiance 1.4

I was playing with the new cabin model and got hung with the following error.

I ran
    rview -av .01 .01 .01 -o sundev oct/cabin
after building oct/cabin according to the Makefile.

The error message is:
    rview: /images/local/lib/ray/rayinit.cal, line 61: syntax error:
    rview: and(a,b) : if( a, b, a );
    rview:          ^ '=' expected

Does this indicate a problem with the function and() in rayinit.cal?
I noticed that and() is defined in my old rayinit.cal as
    and(a,b) = if( a, b, a);

I assume there is a trivial fix for this problem but I don't want to risk any
possible interdependencies by substituting my old version.

-Haydn Richardson
Queen's University

Date: Tue, 30 Apr 91 08:01:22 +0200
From: greg (Greg Ward)
Subject: Re:  Radiance 1.4

The new ':' definitions go with the new 1.4 release of Radiance, and you
must recompile the programs for them to work.  If you just want to look
at the cabin model without recompiling, you can set the RAYPATH
environment variable so it searches the location of the old library first,
but you should include the new library location in it as well or it won't
be able to find some of the files needed for the cabin.  I recommend
running makeall to install the new programs.

-Greg

Date: Tue, 30 Apr 91 11:16:59 EDT
From: [e-mail] (Haydn Richardson)
Subject: Re:  Radiance 1.4

Actually I did run makeall install and it compiled with minimal errors (all
associated with missing XWindows files as we run in Sunview.) RAYPATH does
include the location of the new library. The Makefile in the cabin directory
complains that plasfunc and metfunc are unknown types. Is there something else
that needs to be done to set up the rayinit.cal definitions?

-Haydn

Date: Tue, 30 Apr 91 17:34:19 +0200
From: greg (Greg Ward)
Subject: Re:  Radiance 1.4
Status: RO

The new version of rview requires X11 to be installed in order to compile
properly.  You must still be running the previous version.  You must
manually modify the Makefile in the ray/src/rt directory to remove the
X11 dependency.  Make the following changes:

37c37
< DOBJS = devtable.o devcomm.o editline.o x11.o x11twind.o \
---
> DOBJS = devtable.o devcomm.o editline.o \
41c41
< DLIBS = -lX11
---
> DLIBS =

You must also remove all mentions of x11 from devtable.c (you might
as well take out x10 while you're at it) and reset dev_default[] to "sun".

I guess I should have had makeall complain a bit more on failure!

-Greg

Date: Tue, 30 Apr 91 17:12:06 EDT
From: [e-mail] (Haydn Richardson)
Subject: Continued problems installing Radiance 1.4

I removed the x10 and x11 dependencies from the rt, util, and px Makefiles.
However, I still have the following problems.
   glareval.c line 480 syntax error near (
which refers to the following statement.
#ifdef ALIGN
    scansize = scansize+(sizeof(ALIGN)-1)) & ~(sizeof(ALIGN)-1);
#endif

and my compiled version of rview can no longer find sundev. An old version of
sundev is in my bin and my bin is in RAYPATH.

I appreciate your effort in assisting us through our growing pains.

-Haydn

Date: Wed, 1 May 91 08:31:59 +0200
From: greg (Greg Ward)
Subject: Re:  Continued problems installing Radiance 1.4

Hi Haydn,

The syntax error is an extra parenthesis.  I never tested this particular
segment with ALIGN defined, so just s/-1))/-1)/ in the problem statement.
Sorry about that.  Guess I should update the distribution (already).

The name sundev has been changed to just plain old "sun", so you just
need to use -o sun instead of -o sundev, although in your case you could
make sun the default (dev_default in devtable.c) and not specify a -o
option at all.

-Greg

Date: Sat, 4 May 91 18:15:21 NZT
From: [e-mail]
Subject: Radiance and TAR message

Loaded RADIANCE this evening but when I tar it I get the following:

/dept/arc/pdbourke >tar xfo Radiance1R4.tar
tar: ray/src/util/glareval.c - cannot create

I thought I may have a corrupt copy from the FTP so redid it, same result!

Any suggestions??

Date: Mon, 6 May 91 08:31:43 +0200
From: greg (Greg Ward)
Subject: Re:  Radiance and TAR message

Oops!  Sorry about that!  Glareval.c is a duplicate that was added at the
end of the tar tape.  Apparrently, this won't work unless you change the
mode of the first extracted file during the extraction and before it gets
to the second one.  You can ignore the error for now, but you may get
a syntax error during compilation of this file later which you can also
ignore.

I will repair the distribution and you can try again later if it concerns
you.

-Greg

=======================================================================
MOLEC    Molecular modeling using Radiance

Date: Fri, 26 Apr 91 18:15 EST
From: [e-mail]
Subject: Radiance

I have been looking for a renderer that can do a good job modeling the
area of intra-penetration of two nearly transparent intersecting
spheres.  I am hoping Radiance may solve this problem.  I will be
experimenting with materials and Radiance parameters hoping I can
finally get a good rendering.  If you have any suggestions that would
nudge me in the right direction with Radiance, I would be grateful.

I just started with Radiance and I have been having trouble getting a
correct view so I can look at the window directly for the model
described in the tutorial.  The default view as specified in the
tutorial looks at the wall opposite of the window wall.  I tried
different view points and view directions without luck.  One thing that
confuses me is in rview the point retains its original unnormalized
form: 2.25 0.375 1, however, the view direction appears to be normalized
(-0.25 0.125 -0.125 is reported as -0.8 0.408 -0.408).  Is this the
correct behavior?  Do you know the correct view point and view
direction so that the wall with the window appears in the image?

Thanks in advance.

[e-mail]

Date: Mon, 29 Apr 91 11:41:03 +0200
From: greg (Greg Ward)
Subject: Re:  Radiance

Dear Jonn,

Unfortunately, I do not have a copy of the files used in the tutorial so
I will have to guess at the viewpoint from the input description.  You
can try the following parameters for rview to get a view of the window:

    -vp .1 .1 .75 -vd 1 .8 0 -vh 60 -vv 45

This puts the viewer in the corner looking towards the opposite walls.  From
there, you should be able to adjust the view to your liking using the "aim"
command from within rview.  You are right that the view direction gets
normalized by the program.  Since it is a vector indicating only the direction
to look, it's magnitude is irrelevant.  If you give a vector of 1 2 .5 it
is the same to Radiance as if you had given 2 4 1 or any positive scaling
thereof.

-Greg

P.S.  Regarding your question about transparent spheres.  I think I
need some more specifics.  Are the spheres of a solid material, or like
soap bubbles?  What problems have you had with other renderers?

Date: Mon, 29 Apr 91 12:45 EST
From: Cristy <[e-mail]>
Subject: Re: Radiance

Sorry about not supplying enough information about transparent
spheres.  The two problems I am trying to solve has to do with
molecular modeling.  Many times a chemist will come to me with a
molecule he/she is interested in rendering for publication.  Typically
they want a cluster of atoms in the center of the molecule to be opaque
and the surrounding atoms to be almost totally transparent.  That way
you can see the structure of the molecule and still emphasize the area
of interest (typically the center atoms).

I have tried many raytracers (RAYSHADE, VORT, DBW, TRACER, etc.) and
they produce a nasty artifact at the area of intersection of two
transparent spheres.  For example, assume two atoms represented by
spheres that intersect about 20% of the total area.  The area of
intersection normally appears to be black in all the raytracers I used.
This black area interferes with the transparent effect I am looking for
and distracts from viewing the area of interest in the center of these
outer transparent spheres.  Of course this effect gets worse the more
transparent spheres you have-- to the point where you cannot see the
opaque atoms in the center.

So I have been looking for a renderer that correctly models the area of
intra-penetration of two transparent spheres.  I looked at radiosity
programs, talked with experts in the field (Pat Hanarhan for instance)
but have not found a program that handles this problem well.  I do not
have the expertise to write my own algorithm so...  I am hoping
Radiance may work.  I am looking at the problem now in Radiance, but I
thought that you would have a better feel of Radiance's capability to
handle the problem correctly.  Also perhaps you know of the correct
material properties and other Radiance parameters to solve the
intersecting transparent sphere problem.

Thank you in advance

[e-mail]

Date: Tue, 30 Apr 91 08:44:22 +0200
From: greg (Greg Ward)
Subject: Re: Radiance

I'm guessing that the problems you've encountered with other ray tracers
has to do with thier handling of dielectrics.  If you were modeling the
spheres as solid glass objects, there would definitely be some confusion
as the two objects interpenetrated.  You would be best off modeling the
spheres as an outer surface and an inner surface with a small difference
in radius.  Thus, you would be intersecting two "bubbles" rather than two
solids.  Fortunately, Radiance has a material type that allows you to
model an infinitely thin glass object with a single surface.  The material
type is called "glass" and the parameters are the transmission in red,
green and blue, which you will probably want to set to 1:

    void glass clear_glass
    0
    0
    3 1 1 1

    clear_glass sphere bubble1
    0
    0
    4 x1 y1 z1 r1

    clear_glass sphere bubble2
    0
    0
    4 x2 y2 z2 r2
    ...

This should give you the desired effect.  Again, you should be able to
get similar results by using two slightly different concentric spheres
(with the inside radius negative, or however they specify an inward
surface normal).

-Greg

Date: Fri, 31 May 91 12:04 EST
From: [e-mail]

A fellow scientist wants to model atomic orbitals with Radiance.
Unfortunately there is no easy way (that he knows of) to express orbitals
in terms of cartesian space.  The equations are always in the form of a
spherical harmonic.  Is it possible to express surfaces in spherical
coordinates with Radiance?

Thanks in advance.

Date: Mon, 3 Jun 91 16:20:01 +0200
From: greg (Greg Ward)
Subject: orbitals

Cartesian and sphereical coordinates are of course convertible using
a simple transformation:

    x = rho sin(theta) cos(phi)
    y = rho sin(theta) sin(phi)
    z = rho cos(theta)

I suppose what you are really asking is if there is an easy way to
represent surfaces defined by some arbitrary (in this case spherical
harmonic) function.  The answer is a qualified yes.

The generator program gensurf can be given a parametric description of
the surface in terms of two independent variables (for a spherical
harmonic these would probably be theta and phi) from which it produces a
tesselation of the desired surface into quadrilaterals and triangles.
Thus, the Radiance rendering programs themselves do not model arbitrary
parametric or implicit surfaces, but gensurf can be used to approximate
most parametric surfaces as a collection of polygons.

Unfortunately, I have not done anything about modeling implicit surfaces
(ie. surfaces described by a function of the form F(x,y,z) = 0).
Tesselating such surfaces is not an easily solved problem, and I
have not yet had a strong enough need for them.

-Greg

=======================================================================
PHONG    Phong surface normal interpolation

Date: Mon, 6 May 91 12:55:44 EST
From: Eric Ost <[e-mail]>
Subject: smooth surfaces using normals

If I have a surface defined using discrete polygons, actually triangles,
is there a way that I can use surface normal vectors to derive a smoothly
shaded appearance?  For example, the human form geometry we have access to
consists of approximately 3500 triangles.  When we render instances of
this geometry the edges between triangles are painfully apparent.
Is there some way to smooth these edges without resorting to interpolating
the triangles themselves and creating a geometry database with an increased
number of polygons?  Thanks.

eric

Date: Tue, 7 May 91 08:57:35 +0200
From: greg (Greg Ward)
Subject: Re:  smooth surfaces using normals

Hi Eric,

Yes, it is quite possible to interpolate the surface normals, using a
procedural texture (texfunc) applied to the elements individually.  Of
course, this is not very convenient if you already have the database,
but it can be done using rcalc.  Unfortunately, the math for this
procedure is not very straightforward, and the only place I've done it
is in the Phong shading procedures of gensurf (ray/src/gen/gensurf.c
and see also ray/lib/surf.cal).

I suppose this would be a nice feature to have built into Radiance
directly, but I didn't feel that the method worked well enough to
warrant it.  Specifically, Phong (bilinear) surface normal
interpolation doesn't work for some degenerate cases or for concave
polygons, and I have no idea how to apply it to polygons with more than
four vertices.

I think there are some more general surface normal interpolation schemes
floating around the literature, but I couldn't really recommend one to
you because I haven't tried any of them.

Even gensurf may not help you that much, since I wrote the code to
handle only paired triangles, and never got it to work right for lone
triangles.

I hate to be so discouraging.  This is really something I would like
to see happen, so if you can find a method that you think will do the
trick, I will help you create an rcalc procedure or even a C program
that implements it.

-Greg

=======================================================================
INTERN    Radiance internals

Date: Thu, 9 May 91 16:45:32 EDT
From: [e-mail] (Haydn Richardson)
Subject: View Plane Transformation

Hi Greg,
    I'm trying to write a function which will read the 3-d world coordinates
of an object, as specified in a .rad file and the view parameters from a view
file to generate the corresponding image plane coordinates. My major problem
is defining the relationship between the distance between the view origin vp
and the view plane in terms of the -vh and -vv parameters specified in the
view file. It seems to me that this information is embedded in the function
viewpixel() which is in the file image.c. However, I have been unable to find
where this function is called from if at all. I would be most appreciative if
you could elaborate on the comments for this function, or address my problem
more directly. For example is double d in viewpixel() the distance between the
viewpoint and the view plane? What are the expected input and output
interpretations of the parameters xp,yp,zp? What is the expected input p?

As always thanks for the assistance

-Haydn Richardson
Queen's University at Kingston

Date: Fri, 10 May 91 11:57:28 +0200
From: greg (Greg Ward)
Subject: Re:  View Plane Transformation

Hi Haydn,

So, we've been digging in my code, have we?  I don't get many questions
about internals, so please forgive me for the confusion that follows.

The distance between the view point and the image plane is undefined,
since the image plane is an imaginary entity.  Viewpixel() takes a
point in world coordinates (ie. from the Radiance scene description)
and computes the corresponding image position in normalized coordinates
for a particular view.  The input point, p, is in world (x,y,z), and the
return values *xp and *yp are in normalized view coordinates.  These
coordinates run from (0.,0.) at the lower left corner of the image to
(1.,0.) at the lower right and (1.,1.) at the upper right.  The returned
value *zp is distance along the view direction from the viewpoint to the
world plane containing the point p.  Perhaps this is the grail you seek.

I apologize for the unreadability of my code.  It even gives me trouble
sometimes.  You should ignore anything called i, j, k, d, etc.  These
are almost always temporary variables whose meaning changes depending
on where you are in the procedure.  Such is the case in viewpixel().

The viewpixel() routine is used by certain picture processors such as
px/pinterp.c and util/glareval.c.  I don't recommend learning how to use
it by reading these modules, however.  They are rather nasty.

Good luck!  I'm gone all next week, so any further questions will have
to wait until I get back for a response.

-Greg

======================================================================
PREVW    X11 previewer for Radiance

From: [e-mail]
Subject: Radiance input Previewer.

Dear Greg,
    I am planning to use the Radiance input data format
for my experimentation in Image Synthesis. I've just written
its line drawing previewer for X. I've successfully tested
it on 3 UNIX Platforms. If U want it to be included in
your PD distribution I'll be very happy to send it to U.

If U make any scene data public, pl let me know.
I'll be interested in using them.
----
sumant
(email : [e-mail])
------------------------------------------------------------------
Sumant Narayan Pattanaik
N.C.S.T.
Juhu, Bombay 400 049

Date: Tue, 21 May 91 08:59:24 +0200
From: greg (Greg Ward)
Subject: Re:  Radiance input Previewer.

Hello Sumant,

That's great!  I'd love to try your previewer out, and include it for
distribution.  I am in the process of setting up scene models and programs
for public redistribution via anonymous ftp, and I'll let you know when
it's ready.  In the meantime, you can deposit your previewer in the public
ftp directory on hobbes.lbl.gov (128.3.12.38) where I can retrieve it.
If it's small enough to go by e-mail, you can send it to me directly at
"[e-mail]" instead.  Thanks a lot!

-Greg

Date: Fri, 24 May 91 13:25:35 +0200
From: greg (Greg Ward)
Subject: Re:  Radiance input Previewer.

Hello Sumant,

Thank you very much for the previewer.  I just tried it out and it works
great!  It should come in very handy for anyone using Radiance.  Do you
mind if I include it at our ftp site so that people can pick it up?

Just a couple of minor suggestions.  It would be nice if the previewer
accepted multiple Radiance files, one after the other, or read from
the standard input if none were given.  This should not be very difficult.
I noticed that you adapted the object file reading routine, so just giving
it NULL makes it read from standard input.  Also, you can avoid the need
for specifying a bounding box by using the routines from oconv (bbox.c,
cone.c, face.c, misc.c) to compute the bounding box for you.  This would
require two passes on the input file, however, and it is nice that the
user can give a different bounding box to specify clipping so this
may not be so important.  For a long time, I required the user to include
the bounding box in the input files directly!

Thank you for such a wonderful service.  I still haven't gotten the model
library together, but you will be the first to know!

-Greg

Subject: Re: Radiance input Previewer.
Date: Sun, 26 May 91 11:44:01 +0530
From: [e-mail]

Dear Greg,

I'll do the needful.
I'll not try the  bounding box computation.

I am planning the following modifications.
1. The input parser in not rugged. I am now adapting it
   from "readobj.c".
2. The color of the objects are arbitrary now.
   I'll take the hints from the input file and color them accordingly.

I'll come bcak to U in a day or two.
I want that it is avaiable to public.
However, i havenot asked my boss yet. I'll do it when I am through.
I donot see any problem.
In any case most of the code used are assembled from book or PD
software. So it is really for public consumption. If there is any
problem I'll substitute all references of my name by "anonymous"
and send it to U.
---
sumant

Date: Tue, 28 May 91 08:56:45 +0200
From: greg (Greg Ward)
Subject: previewer

Helo Sumant,

Thank you for the new version of your previewer.  I tried it out and
it works very well indeed.  Thank you for getting (and granting)
permission to redistribute it.  I am in the process now of setting
up the public ftp site at hobbes.lbl.gov (128.3.12.38).  I will
put your software under pub/programs, with a one-line description
of its function.

At the same ftp site (by tomorrow, hopefully) you should find Radiance
objects and models that you may use under pub/objects and pub/models.

Thank you again for all of your help!

-Greg

Date: Tue, 28 May 91 13:44:39 +0530
From: [e-mail]

Dear Greg,
    Thank U for the message. I'll try to get the
new version of Radiance from your site. The earlier version's
(the version I got in the 1st Week of May) documentation
was a bit short. I hope it is improved now.

About the previewer, I must tell U the bugs I know of.
1. In X11.c line 103
    fprintf(..) has missing file pointer.
    Please correct it to
        fprintf(stderr,....)
2. The input parser does not support yet the description in
   the form
    "modifier alias identifier reference"
   and
      for the other description form
    "modifier type identifier
     n s1 s2 s3 ... sn
       0
     m R1 R2 ... Rm"
      the parser takes the sequence literally in terms of
      lines in which they appear.

   I'll correct this problem by adapting to your Parser and
   when I am ready I'll send U a copy. However, I'll be a bit
   late.

Thanks again.
Regards.
----
sumant
(email : [e-mail])
------------------------------------------------------------------
Sumanta Narayan Pattanaik
N.C.S.T.
Juhu, Bombay 400 049

=====================================================================
SPEC    Spectral distributions curves

Date: Tue, 28 May 91 09:16:16 -0400
From: [e-mail] (Stephen N. Spencer)
Subject: Radiance archive


I have a user here at OSU who has done some work with spectral distribution
curves.  He asked me if there any way to include these distribution curves,
either for light sources or objects, in RADIANCE.  I told him that I didn't
think so.  When your mail arrived just now I thought, well, it won't hurt
to ask the one person who can give a definitive answer.

Also, is there any .ies files available?  I've got the "ies*.{rad,dat}" but
don't have any IES input files.  Just curious.

We're having great fun with RADIANCE around here.  It's going to be a mix of
Industrial Design students and Art students using it (we have a sophisticated
scanline renderer as the main renderer but I've introduced RADIANCE as an
alternative).

steve

Date: Tue, 28 May 91 15:51:11 +0200
From: greg (Greg Ward)
Subject: Re:  Radiance archive

Hi Steve,

It is possible to make multiple passes with Radiance, using the three RGB
channels provided to mean some other selection of spectral samples.  For
example, you could have three sets of material files, each with slightly
different RGB values corresponding to different sample locations.  (I
recommend interleaving the samples.)  You then run rpict on each one
and produce three output pictures.  (Be sure to set -sj to 0 so the
pixels correspond.)  These three pictures have a total of 9 spectral
samples, which you can combine however you want using pcomb before the
final display.

This is obviously less efficient than having more spectral samples in
the calculation itself, but until I have the spectral response data,
I can't see the sense in implementing more samples right at the moment.

Regarding IES luminaire data, I put what I have into the archive directory
pub/iesdata.  Thank you for reminding me of this.  Although I didn't have
much to contribute myself, maybe others will pitch in.

-Greg

From: [e-mail] (Stephen N. Spencer)
Subject: Radiance archive

Thanks!  I don't think this is what my user will want to hear (it sounds
rather time-consuming) but hey, it WILL work.

steve

======================================================================
DAYF    Daylight factors and unknown programs

[The following is in response to a letter from John Mardaljevic at Leicester
Polytechnic.  I am too lazy to retype the letter here, but John was
having some problems with Sun Open-Windows and Radiance version 1.3,
and he also had some questions about undocumented programs and computing
daylight factors.  Incidentally, I have added documentation for cnt,
neat, and total and plan to have a daylight factor calculation script
ready in the next release.]

Date: Wed, 29 May 91 10:02:09 +0200
From: greg (Greg Ward)
To: [e-mail]
Subject: Radiance

Hello John,

Your letter was just forwarded to me regarding your questions on v1.3.1.
The problem you are having is an initialization problem that is associated
with the open look window manager (olwm).  It has been fixed in version
1.4, which is available by anonymous ftp from hobbes.lbl.gov (128.3.12.38).
If you do not have access to ftp and do not want to wait for another
upgrade via mail, I can send you the new version of x11image.c by e-mail
and you can recompile it yourself.  Alternatively, you can run twm or some
other window manager instead of olwm and x11image should work.
X11image never has a command window, so that is normal.  The manual page
for x11image is identical to ximage, and I even renamed the program to
ximage in the next release.  Sorry for the confusion.  Unfortunately,
the problems with xshowtrace haven't been fixed in release 1.4,
although I've fixed them since 1.4 and the program will work in the
next release.  Rtrace should work fine, since it doesn't depend on
the window system unless you run it in conjunction with x11image.

As for the missing manual pages, I apologize.  I never did write up
some of the programs I include on the distribution since I figure
most people won't be using them.  There are manual pages for calc,
rcalc and ev in ray/src/cal/man.  I am sorry they are not in the
expected place.  The only documentation for the others is the source
code, but here is a one line description of each:

    cnt        - integer counter, try "cnt 2 5"
    colorscale    - generates color scale picture
    genwindow    - calculates light from window with venetian blinds
    greyscale    - generates grey scale picture
    lam        - joins lines from multiple files
    lookamb        - examines contents of Radiance ambient file
    mt160r        - output driver for Mannesman-Tally dot matrix printer
    neat        - neatens up columns of numbers and aligns decimals
    oki20c        - output driver for OkiData OkiMate 20 color printer
    paintjet    - output driver for HP color paintjet printer
    sun.com, sundev    - driver programs used by rview for sun windows
    total        - sums up columns of numbers

As for daylight factors, the most efficient method is to use rtrace with
the -oi option (will be -I in 1.5) and give it your list of workplane
points with up vectors (ie. 0,0,1).  Then, take the output numbers and
divide them by the ambient level given to you by gensky (multiplied by
pi).  Here is a more complete example:

Let's say we have a room that we have compiled into room.oct and we
want to calculate daylight factors on a 10x8 grid running from x=1
to x=10 and y=5 to y=11 at a height z=3.  First, we run gensky manually
for the day and time we are interested in, like so:

    gensky 5 29 10 -a 47 -o -7 -m -15

Of course, you would adjust the latitude, longitude and standard meridian
to correspond with your site location.  The above produced the following
comment in its output:

    # Ground ambient level: 7.582072

This ground ambient level corresponds to the irradiance/pi due to the sky
without direct solar, which is what we will divide into our irradiance
values from rtrace to get daylight factors, thus:

    cnt 10 8 | rcalc -e '$1=$1+1;$2=$2+5;$3=3;$4=0;$5=0;$6=1' \
    | rtrace -oi room.oct | rcalc -e '$1=(.3*$1+.59*$2+.11*$3)/PI/7.582' \
    > room.df

The factors of .3, .59 and .11 are to get from RGB to brightness.
Note that the values in the output file do not have their associated input
values with them.  You can add them in again like so:

    cnt 10 8 | rcalc -e '$1=$1+1;$2=$2+5' | lam - room.df \
    | neat 4.9 > room.df.final

Notice that I tried to use as many of the programs you asked about as I
could!  Notice also that doing real calculations with Radiance
currently requires long, unreadable command lines.  We hope to make
some of this easier in the next release by adding shell scripts to do
these kinds of useful things without requiring so much user guidance.

I hope some of this helps.
-Greg

=================================================================
VISION    Vision-3D modeler for the MacIntosh

Date: Fri, 17 May 91 9:58:22 NZT
From: [e-mail]
To: [e-mail]
Subject: Mac modeller

The public domain modeller Vision-3D for the Mac II family is about to
support Radiance data files as an export option. This has already been
done but the copy on our FTP site hasn't yet been updated (I want to put
some more features in the next release) If anyone is interested however
the current version of Vision-3D with Radiance file export can be made
available.
--
 |   Paul D Bourke        [e-mail]         (130.216.1.5)   |


Date: Fri, 17 May 91 11:16:42 +0200
From: greg (Greg Ward)
Subject: Re:  Mac modeller

Hi Paul,

Thank you for putting a Radiance output option into your modeller!  I have
had a student working on a translator from Sculp 3D RIB files to Radiance
with limited success.  It would be great to get a hold of a real 3D
modeller for the MacIntosh II, since that's the computer I use most
often.  Does your program work with A/UX 2.0?  I have Radiance working
on the Mac under this operating system now.

-Greg

Date: Sat, 18 May 91 10:38:22 NZT
From: [e-mail]
Subject: Re: Mac modeller

>
> Thank you for putting a Radiance output option into your modeller!  I have
> had a student working on a translator from Sculp 3D RIB files to Radiance
> with limited success.  It would be great to get a hold of a real 3D
> modeller for the MacIntosh II, since that's the computer I use most
> often.

Well, Vision-3D is a shareware modeller...MicroStation is a real modeller.
The RayShade and Radiance export facility has received so much attention
that I will put a preliminary copy of the next version of Vision-3D in our
FTP directories early next week. The readme file will indicate when this
has been done. In case you don't know we are ccu1.aukuni.ac.nz (130.216.1.5)
The directory you want is mac/architec.

> Does your program work with A/UX 2.0?

Don't know. What are the issue, I use Think C (never even seen AUX)

> I have Radiance working on the Mac under this operating system now.

Is it possible to generate a version that would run under the finder OS.
This would be of great interest to many users.
--
 |   Paul D Bourke        [e-mail]         (130.216.1.5)   |

Date: Tue, 21 May 91 08:55:41 +0200
From: greg (Greg Ward)
Subject: Re: Mac modeller

Hi Paul,

Unfortunately, I don't have a version of Radiance for the MacOS, and I don't
expect to in the near future.  The rendering program in Radiance was really
meant to run in the background and may require large amounts of memory for
complex scenes, so I don't think it would be very simple to port it to the
native Macintosh OS.  You are welcomed to try!  Frankly, I haven't got much
patience for menus and that sort of programming.  A few years ago, someone
I hired wrote a 3d editor for Radiance files on the Mac and the code was
bigger than my renderer!  It never quite worked to my satisfaction (it
crashed a lot) so I didn't advertize it much.  This is another problem
with the Mac environment -- frequent crashes that spell disaster for
background processes.  Also, the system does not allow that much CPU
time to such programs, preferring to reserve as much as possible for
the application running in the foreground.  Perhaps this will change
a bit with version 7, but I doubt it.

I will pick up the modeller as soon as you have the right release on there.
Thanks again.

-Greg

Date: Wed, 22 May 91 7:48:49 NZT
From: [e-mail]
Subject: Re: Mac modeller

> Unfortunately, I don't have a version of Radiance for the MacOS, and I don't
> expect to in the near future.  The rendering program in Radiance was really
> meant to run in the background and may require large amounts of memory for
> complex scenes, so I don't think it would be very simple to port it to the
> native Macintosh OS.

Yes, I tried to render some landscapes woth 2 million polygonal facets, no
luck yet and our SGI has 64MB !

> You are welcomed to try!  Frankly, I haven't got much
> patience for menus and that sort of programming.  A few years ago, someone
> I hired wrote a 3d editor for Radiance files on the Mac and the code was
> bigger than my renderer!

You shouldn't be very surprised at this, a modeller is much more complex
and varied than a renderer.

> It never quite worked to my satisfaction (it
> crashed a lot) so I didn't advertize it much.  This is another problem
> with the Mac environment -- frequent crashes that spell disaster for
> background processes.

Can't remember my last crash! I am afraid that the impression people have
of the Mac as unstable is almost always due to substandard software
especially INITs, games, utilities in the public domain. The system
software and toolbox is only resposible for a small fraction of the
problems people have with Macs.

> Also, the system does not allow that much CPU
> time to such programs, preferring to reserve as much as possible for
> the application running in the foreground.  Perhaps this will change
> a bit with version 7, but I doubt it.

No it hasn't really changed. An application can specifiy how often a
background task is "looked" at and for how long. Most developers tend
to give background tasks only a peek every now and then because they
want maximum performance for themselves. I have only seen one application
which allowed the user to specify the sharing load, a good idea though.

> I will pick up the modeller as soon as you have the right release on there.
> Thanks again.

I have placed a beta version of the modeller that supports Radiance in our
archive. There is an administrative problem though with our site, they are
worried about the huge amount of traffic they are experiencing.
--
 |   Paul D Bourke        [e-mail]         (130.216.1.5)   |

Date: Wed, 22 May 91 09:26:03 +0200
From: greg (Greg Ward)
Subject: Re: Mac modeller

Hi Paul,

I just picked up your modeller yesterday, and I must say I'm impressed.
I hope it took you a long time to write and document.  Please don't tell
me you did it during donut breaks on Thursdays.

If you want to render large scenes with Radiance, you need to somehow
break them down into repeatable groups using instances.  Of course, this
doesn't work if what you're rendering doesn't have any inherent repetition
inherent repitition.  Almost everything I do does.

I agree that crashes are caused by poor inits more than bad system software,
but I have little control over what other people do to their machines and
they seem to like those silly little bastards.  I've been using a shared
machine here at EPFL and every day I switch it on it seems like there's
another one.  Maybe they breed overnight.  I go for a bite while the thing
boots up.

I didn't find the export option for Radiance in the modeller.  It's not
in the expected place (File Export...).  Are you sure it was on yesterday's
release?

If network traffic is a problem, I would be happy to put Vision 3D on
my ftp site in California.  That is, if you have no objections.  From
the manual, it sounds like you were (thinking?) of marketing the program
at one time.  I'd be interested to hear what your current attitude is on that.

By the way, I ran Vision 3D under A/UX to try it out, and everything I
tested seemed to work fine.  The only problem I noticed was that the
cursor disappeared after I quit.  There must be some deinitialization that's
required for A/UX that the normal Finder takes care of.  I've noticed other
strange cursor happenings in the past with other programs as well.  The
other thing that wasn't quite right was that the 32-bit clean flag was not
set on the program.  That's a function of the compiler, I suppose, though
it is possible to set this flag yourself using ResEdit.  The requirements
for applications running under A/UX (as I understand it) besides 32-bit
clean are that you stick with Apple's guidelines (which I guess you have)
and that you don't access certain little-used toolbox routines that the A/UX
people haven't implemented yet.  Most of these have to do with strange
color table manipulations as near as I can tell.

Once I get the correct version of Vision 3D (assuming I ain't got it),
I'll do a more thorough testing under A/UX so I can tell you if there are
any other problems.  I don't expect there will be too many, anyway.

Do you know Robert Amor?  How did New Zealand get so many hot shots when
they have so little money?  As I said before, I'm impressed.

-Greg

Date: Wed, 22 May 91 20:32:14 NZT
From: [e-mail]
Subject: Re: Mac modeller

> I just picked up your modeller yesterday, and I must say I'm impressed.
> I hope it took you a long time to write and document.  Please don't tell
> me you did it during donut breaks on Thursdays.

I did work on it fairly well full time for a few months. I would love to
have time to rewrite the "definitive" modeller now that I know know how. ie:
Vision-3D was my first 3D software but it has been left behind because of
other commitments. I would love to write a good modeller specifically for
RenderMan but it's probably a years work!

> I didn't find the export option for Radiance in the modeller.  It's not
> in the expected place (File Export...).  Are you sure it was on yesterday's
> release?

Sorry, humble...humble... my mistake, don't know how it happened, etc etc...
I will fix up the directory, I might try mailing you the application.

> If network traffic is a problem, I would be happy to put Vision 3D on
> my ftp site in California.  That is, if you have no objections.  From
> the manual, it sounds like you were (thinking?) of marketing the program
> at one time.  I'd be interested to hear what your current attitude is
> on that.

Please put it somewhere in the US and let me know where. There has been
huge traffic loads to our site, I am becoming unpopular with the Computer
Centre administrators. The current position is that the program is shareware,
which means in my book that you should feel free to copy and distribute the
software. If you keep and use it for any sort of financial gain then I would
appreciate $120 NZ (approx US$80)

> By the way, I ran Vision 3D under A/UX to try it out, and everything I
> tested seemed to work fine.  The only problem I noticed was that the
> cursor disappeared after I quit.  There must be some deinitialization that's
> required for A/UX that the normal Finder takes care of.  I've noticed other
> strange cursor happenings in the past with other programs as well.  The
> other thing that wasn't quite right was that the 32-bit clean flag was not
> set on the program.  That's a function of the compiler, I suppose, though
> it is possible to set this flag yourself using ResEdit.  The requirements
> for applications running under A/UX (as I understand it) besides 32-bit
> clean are that you stick with Apple's guidelines (which I guess you have)
> and that you don't access certain little-used toolbox routines that the A/UX
> people haven't implemented yet.  Most of these have to do with strange
> color table manipulations as near as I can tell.

Yeah, I have just installed system 7 and all my programs seem to work OK. I
do use my own cursors when the mouse is in "my" windows (it looks like the
normal crosshair cursor but it's not). You will probably find that if you
quit from Vision-3D with the cursor away from my windows all will be fine.
I'll put cursor initialising on my list of things to do.

> Once I get the correct version of Vision 3D (assuming I ain't got it),
> I'll do a more thorough testing under A/UX so I can tell you if there are
> any other problems.  I don't expect there will be too many, anyway.

Thanks

> Do you know Robert Amor?  How did New Zealand get so many hot shots when
> they have so little money?  As I said before, I'm impressed.

Yes, how do you know him? I met him last week, he's come up to the Computer
Science dept here to do some brief contract work on an expert system for
a building industry firm. His address here is [e-mail]
He just gave me 40MB of radiance examples...!
--
 |   Paul D Bourke        [e-mail]         (130.216.1.5)   |

Date: Thu, 23 May 91 09:02:01 +0200
From: greg (Greg Ward)
Subject: Re: Mac modeller

Hi Paul,

Thank you for sending the beta version of Vision3D.  I know what you mean
about wanting to start over.  I've put the modeller just by itself on
anonymous ftp at hobbes.lbl.gov (128.3.12.38) under pub.  You can put the
other hqx files there yourself if you feel like it.  I am planning to get
the whole thing organized with drop off points and pick up points and
read me files and so on, in a day or two.  (How many weeks have I been
saying this?)  Actually, it was largely Robert Amor's idea to have a
shared Radiance archive, and I recently got a lot of interest at the
Eurographics workshop on rendering.

Robert came by Berkeley at some point last year and we talked a bit.
He knew an awful lot about Radiance and had some good suggestions and
one or two programs for me.  Also, we had been exchanging e-mail for
some time on the topic of building data representation, a shared interest.

I had a quick look at the Radiance export files from Vision3D and they
look great.  I haven't run any tests, yet, though, so I'll have to let
you know how they turn out.  I read quickly through your little article
on CAD formats and took some of the things you said to heart.  My next
release of Radiance will be much more forgiving of cones and spheres with
"illegal" radii so there will be a little less for you to worry about.
I would be interested in any specific comments or criticism you have
about the Radiance scene description format.  It was basically written
for my own (ie. the implementer's) convenience rather than ease of use,
so I realize it is somewhat brain dead.  (Scanf is my parser and printf
is my formatter...)

-Greg

Date: Sun, 26 May 91 12:02:00 NZT
From: [e-mail]
Subject: Triangulate

Something else of mine that now supports Radiance export...!

Triangulate is a Mac II utility that takes a "random" distribution of samples
of a surface and generates a triangulated (Delauney) or gridded mesh
(user specified resolution) representation of the surface. We use it
extensively for generating terrain models, the data is generated either from
manual entry from site surveys or by digitizing existing contour maps. A
number of export formats are supported, DXF, Super3D text, and now Radiance.

The archive I uploaded has been passed through Stuffit (.sit) and then
BinHex (.Hqx). It contains the application and the user manual in MS Word 4
format.
--
 |   Paul D Bourke        [e-mail]         (130.216.1.5)   |