mkillum q's

I'm trying to use mkillum to render my house.. and I'm having trouble
understanding it.
I have broken the elements into various Rad files
Nov15.rad - the sky description
house_mat_lit.rad - materials and lights used in the house
House6nw.rad - house geometry without window geometry
furniture.rad - the (duh) furniture
windows.rad - the window geometry

RwR chapter 1 suggests using a .rif with an "illum=" line
scene= nov15.rad
scene= house_mat_lit.rad
scene= house6nw.rad
scene= windows.rad -- should this be here? or only as illum?
scene= furniture.rad
illum= windows.rad

# Options for mkillum
mkillum= -av 8 8 8 -ab 0

This is taking a long time to generate the .dat illum files - in fact I give
up after a while
I 'm thinking all I really need is the sky and the windows?
scene= nov15.rad
illum= windows.rad

This produces:
robfitzs% rad window.rif
                  oconv nov15.rad > window.oct
                  oconv -i window.oct windows.rad > window0.oct
                  mkillum window0.oct "<" windows.rad > il000924
                  oconv -f -i window.oct il000924 > window1.oct
                  rm -f il000924

this goes much faster, (creating 10 illum.dat files) but now how do I use
the illum data and render with all the rest of the house and furniture
without overwriting the previous simpler octree?

Could I add this to my rif?
octree= window1.oct

I also tried using mkillum manually ( for this, I combined the sky and
windows in 1 rad file) But I got an error
mkillum -ab 1 -ad 512 -as 256 -ar 16 < windowsSky.rad > winlit.rad
rtrace: fatal - command line error at '-ar'
mkillum: system - error reading from rtrace process

Does this make sense?

Rob Fitz

···

Fitzsimmons, Rob wrote:

scene= windows.rad -- should this be here? or only as illum?

No, the illum line is enough.
This results in a duplication of your window polygons in the
octree, which might result in all kinds of weird effects,
possibly including overly long simulation times.

illum= windows.rad

# Options for mkillum
mkillum= -av 8 8 8 -ab 0

This is taking a long time to generate the .dat illum files -
in fact I give up after a while

With -ab 0, all the other traditional mistakes (having the
polygons oriented the wrong way etc.) won't make much of a
difference, so the duplication is your best bet for the moment.

I 'm thinking all I really need is the sky and the windows?

If you have any geometry obstructing the sky when viewed from any
of your windows (roof overhangs, trees, neighbouring buildings),
then this will result in an overestimation. As is obvious from
the setup, you're basically determining the light your windows
would receive if they were floating up in the air somewhere.

-schorsch

···

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

Hi Rob F,

As Georg mentioned, you only need to reference your windows once, in the illum= line. Also, your "illum=" options are a little off, as Georg also mentioned. Greg Ward explained this to me rather well, in an email that is either on my office computer or on my PC upstairs, because it ain't on the laptop sitting on my lap at the moment (I really need to start using IMAP).

Anyway, you basically have to tell mkillum what's going on with your model in the illum= line, because all your input with QUALITY=, VARIABILITY= and DETAIL= in the .rif file mean absolutely nothing to the mkillum process. This is something that is not well documented, IMHO.

Make sure your illum= line includes sufficent -ab, and -ad, and -ar, and -as to ensure that the mkillum process is aggressively calculated. As Georg said, if you wimp out with the ambient calculation on your windows, the effect of the exterior environment will be largely ignored in terms of the distribution of the windows' luminance. Bottom line: just because you're using rad, that does not exempt you from having to be intimate with the various rpict switches for a proper rendering if you're using mkillum. (and please don't read this as some guru talking down to you; I'm still learning this stuff myself.)

I hope this helps.

Rob Gee.

-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-:-
   Rob Guglielmetti
[email protected]
www.rumblestrip.org

···

On Dec 2, 2003, at 8:39 PM, Fitzsimmons, Rob wrote:

I'm trying to use mkillum to render my house.. and I'm having trouble understanding it.

Thanks guys for your help

Rob - I did put the rtrace options in a mkillum= line, not the illum=
still rendering - we'll see how it turns out

A couple of other things I don't understand

If I use both the scene=sky.rad and illum=windows.rad
Are they both lighting the scene?
I assume not.. watching the process:

oconv nov15.rad house_mat_lit.rad house6nw.rad chandel_mat.rad furniture.rad

window.oct

        oconv -i window.oct windows.rad > window0.oct
        mkillum window0.oct "<" windows.rad > il000949
        oconv -f -i window.oct il000949 > window1.oct
        rm -f il000949
        rm -f win.amb

It seems
window.oct is the full scene
window0.oct is windows.rad
window1.oct is the full scene with the windows.rad turned into illums?

after mkillum creates the illum1.dat, illum2.dat etc.
How do I then reference that info?
I assume when using rad, it is somewhat automated.
but If I were to manually use mkillum, does it alter the windows.rad file to
use the illum.dat files?

(and please don't read this as some guru talking
down to you; I'm still learning this stuff myself.)
not at all! I appreciate the help on this list

Rob Eff

Fitzsimmons, Rob wrote:

Thanks guys for your help

Rob - I did put the rtrace options in a mkillum= line, not the illum=
still rendering - we'll see how it turns out

A couple of other things I don't understand

If I use both the scene=sky.rad and illum=windows.rad
Are they both lighting the scene?
I assume not.. watching the process:

oconv nov15.rad house_mat_lit.rad house6nw.rad chandel_mat.rad furniture.rad
  > window.oct
oconv -i window.oct windows.rad > window0.oct
mkillum window0.oct "<" windows.rad > il000949
oconv -f -i window.oct il000949 > window1.oct
rm -f il000949
rm -f win.amb

It seems
window.oct is the full scene
window0.oct is windows.rad

That's window.oct with windows.rad added to it (the -i option).

window1.oct is the full scene with the windows.rad turned into illums?

Yes (same principle).

after mkillum creates the illum1.dat, illum2.dat etc.
How do I then reference that info?

You shouldn't need to, rad does that.
I'm not sure if you can convince rad *not* to delete "il000949"
again after creating the final octree, because that's where you'd
find how it's done... It should be a brightfunc referencing
illum.cal, but I don't have an example sitting around either.

I assume when using rad, it is somewhat automated.
but If I were to manually use mkillum, does it alter the windows.rad file to
use the illum.dat files?

The "il000949" (or whatever name it will get on the next run)
is the altered version of windows.rad.

-schorsch

···

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