Still having trouble, new pics........

radiance-general
1

Firstly, to Lars grobe in reply to how I mapped the texture on to the
surface of the book from my last post. The book was made in Maya, nurbs
surface -> polygon -> obj2mesh , the texture mapping was also done in
maya, basically it's a planar map going straight through the book as
though you were reading it. The texture was just a photo looking
straight onto the book, and the planar map causes it to stretch and bend
on the surface the right amount. Wasn't tricky at all.

Secondly, was a bit confused by Georg's advice when you said -aa should
be increased, did you mean increased in accuracy? ie putting aa to a
lower value? Or did you mean increasing aa by setting it to a higher
value? Because the lower the aa value the more accurate the picture, so
I can only see setting aa lower than .15 as a good thing not higher?

Well anyway I definitely don't think aa is the problem anyway. Since the
last renders I sent out ive brightened the stained glass a bit so more
light gets in, and the results this time are pretty crazy! The glass is
rose window at the front is projecting loads of blotches of colour.

Check out these renders, in the first I used

-x 2560 -y 2048 -ar 128 -ad 512 -aa .15 -as 256 -ab 5

in the second I set the -aa to .10 and halved the ad and as , as the
rendering time would have been stupid

-x 2560 -y 2048 -ar 128 -ad 256 -aa .10 -as 128 -ab 5

The second has much larger colour splotches which suggests to me that
its not the ambient accuracy, and the splotch are being cause by to low
ad and as. So im doing a render now with ad and as up at 1024 and 512,
we shall see how that goes. If any one has any other ideas please say! I
need to get some results inside this building soon as my project is due
in soon!

https://www.bris.ac.uk/fluff/u/js0754/iBqEK1g3QV8ruqxb269Qsg7j/Output8.j
pg

https://www.bris.ac.uk/fluff/u/js0754/VW4ynFc05IgiAVwIbzPjuA7j/Output9pc
ondlowaa.jpg

Thanks guys!

John

2

just a quick check:
are you sure to be using illum in the window surfaces? you miss the typical
fake shadows this solution implies....
I guess you can use an illum surface outside of the windows, pointing
inside, close to the glass and with a good -pj of 0.6 ?0.8 and -ds < 0.1
eliminate all the fake shadow... doing this way the 'arlechine' effect
should go away... I guess
also... remember to set -av... seems to be a little high...

if you use -aa more than 0.3 with low values for -ad and -as you can save
time and eliminate artefacts... well artefacts are too big to be visible...
sometimes is a good trick to speed up renderings, other time it is not, just
try ;-))

cheers
giulio

PS another option is to use a good -ad value, more than 1024, may be you can
reduce a little ar to speed up, say -ar 64?96 ...
or the crazy solution: -aa 0 -ad 32 -as 4 -ab 1 -ps 1 -pj .8 (or if you
can 2)....

···

-----Original Message-----
From: John Sutherland [mailto:[email protected]]
Sent: 29 March 2004 19:01
To: [email protected]
Subject: [Radiance-general] Still having trouble, new pics........

Firstly, to Lars grobe in reply to how I mapped the texture on to the
surface of the book from my last post. The book was made in Maya, nurbs
surface -> polygon -> obj2mesh , the texture mapping was also done in maya,
basically it's a planar map going straight through the book as though you
were reading it. The texture was just a photo looking straight onto the
book, and the planar map causes it to stretch and bend on the surface the
right amount. Wasn't tricky at all.

Secondly, was a bit confused by Georg's advice when you said -aa should be
increased, did you mean increased in accuracy? ie putting aa to a lower
value? Or did you mean increasing aa by setting it to a higher value?
Because the lower the aa value the more accurate the picture, so I can only
see setting aa lower than .15 as a good thing not higher?

Well anyway I definitely don't think aa is the problem anyway. Since the
last renders I sent out ive brightened the stained glass a bit so more light
gets in, and the results this time are pretty crazy! The glass is rose
window at the front is projecting loads of blotches of colour.

Check out these renders, in the first I used

-x 2560 -y 2048 -ar 128 -ad 512 -aa .15 -as 256 -ab 5

in the second I set the -aa to .10 and halved the ad and as , as the
rendering time would have been stupid

-x 2560 -y 2048 -ar 128 -ad 256 -aa .10 -as 128 -ab 5

The second has much larger colour splotches which suggests to me that its
not the ambient accuracy, and the splotch are being cause by to low ad and
as. So im doing a render now with ad and as up at 1024 and 512, we shall see
how that goes. If any one has any other ideas please say! I need to get some
results inside this building soon as my project is due in soon!

https://www.bris.ac.uk/fluff/u/js0754/iBqEK1g3QV8ruqxb269Qsg7j/Output8.jpg
<https://www.bris.ac.uk/fluff/u/js0754/iBqEK1g3QV8ruqxb269Qsg7j/Output8.jpg>

https://www.bris.ac.uk/fluff/u/js0754/VW4ynFc05IgiAVwIbzPjuA7j/Output9pcondl
owaa.jpg
<https://www.bris.ac.uk/fluff/u/js0754/VW4ynFc05IgiAVwIbzPjuA7j/Output9pcond
lowaa.jpg>

Thanks guys!

John

___________________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses.

3

John Sutherland wrote:

Secondly, was a bit confused by Georg's advice when you said -aa should
be increased, did you mean increased in accuracy? ie putting aa to a
lower value? Or did you mean increasing aa by setting it to a higher
value? Because the lower the aa value the more accurate the picture, so
I can only see setting aa lower than .15 as a good thing not higher?

Increasing a parameter means to give it a higher numeric value.
Increasing -aa will result in lower accuracy. But since you have
-ad and -as at relatively inaccurate settings, that would result
in a better balance. Alternatively, you can increase -ad and -as,
which also results in better balance, with better accuracy
overall, but taking (possibly much) more time to render.
In either direction you will eventually find a combination that
reduces or even eliminates the artifacts.

Well anyway I definitely don't think aa is the problem anyway.

There's hardly ever a single parameter responsible for specific
artifacts. They all need to play together in the right way.

Your new pictures are a nice demonstration of what happens when
you disturb the balance even more than was first the case.
I actually suspect that the original builders of that space
would have been very happy if they could have produced such
an effect in reality! :wink:

-schorsch

···

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

4

Firstly Guilio,
  Yes im using mkillum , the sources are about the equivalent of a
few cm behind the windows and they are shining in through the glass. Do
you still think there might be something up to do with the mkillum?

Secondly Georg , I was following the rendering parameters suggested by

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

Which suggests that 512 and 256 for ad and as is accurate? My mistake if
they are not, but im trying double that now anyway. Cheers guys.....
John

···

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Georg
Mischler
Sent: 29 March 2004 19:27
To: Radiance general discussion
Subject: Re: [Radiance-general] Still having trouble, new pics........

John Sutherland wrote:

Secondly, was a bit confused by Georg's advice when you said -aa

should

be increased, did you mean increased in accuracy? ie putting aa to a
lower value? Or did you mean increasing aa by setting it to a higher
value? Because the lower the aa value the more accurate the picture,

so

I can only see setting aa lower than .15 as a good thing not higher?

Increasing a parameter means to give it a higher numeric value.
Increasing -aa will result in lower accuracy. But since you have
-ad and -as at relatively inaccurate settings, that would result
in a better balance. Alternatively, you can increase -ad and -as,
which also results in better balance, with better accuracy
overall, but taking (possibly much) more time to render.
In either direction you will eventually find a combination that
reduces or even eliminates the artifacts.

Well anyway I definitely don't think aa is the problem anyway.

There's hardly ever a single parameter responsible for specific
artifacts. They all need to play together in the right way.

Your new pictures are a nice demonstration of what happens when
you disturb the balance even more than was first the case.
I actually suspect that the original builders of that space
would have been very happy if they could have produced such
an effect in reality! :wink:

-schorsch

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

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

5

John Sutherland wrote:

I was following the rendering parameters suggested by

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

Which suggests that 512 and 256 for ad and as is accurate? My mistake if
they are not, but im trying double that now anyway. Cheers guys.....

The optimal parameters strongly depend on the characteristics
of each specific scene.

Any recommended values can only be a starting point, from where
on you need to check in which direction to correct them.

-schorsch

···

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

6

last idea.... if you want to try,
change the windows material to (it maps the sky luminance to the windows and
use them like a visible light source):

skyfunc brightfunc light_of_the_sky
2 winxmit winxmit.cal
0
0

light_of_the_sky light your_window
0
0
3 .7 .7 .7

have a look to the rendering....
use this settings:
-ab 2 -aa .08 -ad 1024 -as 512 -ar 64 -ps 1 -dj .8 -ds .1
(it takes a while...)
may be just check that the normal are fine before trying....
... I always forgot about this, £$"*@#{{ ! ;-))

···

----------------------------------------------------------------------------
----------------------------------------------------------------------------
------
if it is the same splotchy image, stop here; if it is not go further.
----------------------------------------------------------------------------
----------------------------------------------------------------------------
--------
just add this:

void colorpict colour_of_the_glass
.........this is yor texture for the glass

colour_of_the_glass glass the_glass_matherial
0
0
3 .76 .76 .76

skyfunc brightfunc light_of_the_sky
2 winxmit winxmit.cal
0
0

##now the material is illum and not light...

light_of_the_sky illum your_window
1 the_glass_matherial
0
3 .7 .7 .7

hope it helps
cheers
giulio

PS of course you should change .76 and .7 according to your glass
description...

___________________________________________________________________
Electronic mail messages entering and leaving Arup business
systems are scanned for acceptability of content and viruses.

7

Hi John,

Firstly, you should move your illum's to just inside the windows, rather than having them on the outside. This should reduce the bizarre color artifacts you're seeing. Secondly, you should increase your -ar parameter a bit and simultaneously increase -ad and -as parameter values. This should reduce the splotches you are seeing in corners. Decreasting the -aa value might help as well, but I would try that next if these suggestions don't solve the problem.

-Greg

From: "John Sutherland" <[email protected]>
Date: March 29, 2004 10:01:29 AM PST

Firstly, to Lars grobe in reply to how I mapped the texture on to the surface of the book from my last post. The book was made in Maya, nurbs surface -> polygon -> obj2mesh , the texture mapping was also done in maya, basically it’s a planar map going straight through the book as though you were reading it. The texture was just a photo looking straight onto the book, and the planar map causes it to stretch and bend on the surface the right amount. Wasn’t tricky at all.

Secondly, was a bit confused by Georg’s advice when you said –aa should be increased, did you mean increased in accuracy? ie putting aa to a lower value? Or did you mean increasing aa by setting it to a higher value? Because the lower the aa value the more accurate the picture, so I can only see setting aa lower than .15 as a good thing not higher?

Well anyway I definitely don’t think aa is the problem anyway. Since the last renders I sent out ive brightened the stained glass a bit so more light gets in, and the results this time are pretty crazy! The glass is rose window at the front is projecting loads of blotches of colour.

...