high reflectance & roughness metals

hi, althought i guess it has been discussed already at the forum, i've
having some problems with high reflectance & roughness metals to simulate a
diffuse metal (alluminium) garage door

my first atempt:

rpict -x 4000 -y 3000 -vf vistas/e3 -af ambext -ab 2 -ad 200 -as 30 -ar
300 -sj 1 -st 0 -dj .5 -pj .5 -t 60 exterior.oct>img/exteriore3.pic
pfilt -r .6 -x 800 -y 600 img/exteriore3.pic>img/exteriore3pf.pic

with the material
void metal jaimito_l_3dpgaraje
0
0
5 0.65 0.65 0.65 0.5 .1

and the result is
http://www.alfanature.com/avis/exteriore3b.jpg

as you can see there are some artifacts (horizontal bands) , are them due to
low-discrepancy?

the second attempt is using pure montecarlo

void texfunc metro
4 gloss_dx gloss_dy gloss_dz gloss.cal
0
3 .02 .02 .02
metro metal jaimito_l_3dpgaraje
0
0
5 0.65 0.65 0.65 0.5 0

and the result is
http://www.alfanature.com/avis/exteriore3.jpg

any way to improve that?
anyway i prefer pure montecarlo to default radiance, do you?
in other threads it was recommended to decrement sj and increment st, but
this isn't it a way to elude the problem?

thanks all

Hi Ignacio,

If you prefer standard Monte Carlo sampling over the low-discrepency method, then you should recompile with -DMC as a flag in your rmake script. The only way to reduce the noise is to increase your sampling density (initial rendering resolution).

-Greg

···

From: Ignacio Munárriz <[email protected]>
Date: February 25, 2005 12:40:40 AM PST

hi, althought i guess it has been discussed already at the forum, i've
having some problems with high reflectance & roughness metals to simulate a
diffuse metal (alluminium) garage door

...

I was disturbed with the horizontal lines that was present in the metal
http://www.alfanature.com/avis/exteriore3b.jpg

son i took a look at the code normal.c, first discovering that the random
functions
lrand48(), drand48() in random.h
did not work in my computer (return always 0), so i changed them for:
#define random() rand()
#define srandom(s) srand((unsigned)(s))
#define frandom() (random()*(1./2147483648.))
then random works but continued with the artifact, then i changed normal.c
in this way:

instead of multisamp(rv, 2, d);
rv[0]=frandom();
rv[1]=frandom();

i think that multisamp creates a pattern in the random because the randoms
numbers i got were rare( some too exacts .50000 0.00000 .7500000)

and the result is
http://www.alfanature.com/avis/exter.jpg

···

----- Original Message -----
From: "Gregory J. Ward" <[email protected]>
To: "Radiance general discussion" <[email protected]>
Sent: Friday, February 25, 2005 6:24 PM
Subject: Re: [Radiance-general] high reflectance & roughness metals

Hi Ignacio,

If you prefer standard Monte Carlo sampling over the low-discrepency
method, then you should recompile with -DMC as a flag in your rmake
script. The only way to reduce the noise is to increase your sampling
density (initial rendering resolution).

-Greg

From: Ignacio Mun�rriz <[email protected]>
Date: February 25, 2005 12:40:40 AM PST

hi, althought i guess it has been discussed already at the forum, i've
having some problems with high reflectance & roughness metals to
simulate a
diffuse metal (alluminium) garage door

...

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

and a more smoother result is achieved using urand in rv[0]
rv[0]=urand(ilhash(dimlist,ndims)+samplendx);

rv[1]=frandom();

http://www.alfanature.com/avis/exter3.jpg

···

----- Original Message -----
From: "Ignacio Mun�rriz" <[email protected]>
To: "Radiance general discussion" <[email protected]>
Sent: Tuesday, March 01, 2005 11:20 AM
Subject: Re: [Radiance-general] high reflectance & roughness metals

I was disturbed with the horizontal lines that was present in the metal
http://www.alfanature.com/avis/exteriore3b.jpg

son i took a look at the code normal.c, first discovering that the random
functions
lrand48(), drand48() in random.h
did not work in my computer (return always 0), so i changed them for:
#define random() rand()
#define srandom(s) srand((unsigned)(s))
#define frandom() (random()*(1./2147483648.))
then random works but continued with the artifact, then i changed normal.c
in this way:

instead of multisamp(rv, 2, d);
rv[0]=frandom();
rv[1]=frandom();

i think that multisamp creates a pattern in the random because the randoms
numbers i got were rare( some too exacts .50000 0.00000 .7500000)

and the result is
http://www.alfanature.com/avis/exter.jpg

----- Original Message -----
From: "Gregory J. Ward" <[email protected]>
To: "Radiance general discussion" <[email protected]>
Sent: Friday, February 25, 2005 6:24 PM
Subject: Re: [Radiance-general] high reflectance & roughness metals

Hi Ignacio,

If you prefer standard Monte Carlo sampling over the low-discrepency
method, then you should recompile with -DMC as a flag in your rmake
script. The only way to reduce the noise is to increase your sampling
density (initial rendering resolution).

-Greg

> From: Ignacio Mun�rriz <[email protected]>
> Date: February 25, 2005 12:40:40 AM PST
>
> hi, althought i guess it has been discussed already at the forum, i've
> having some problems with high reflectance & roughness metals to
> simulate a
> diffuse metal (alluminium) garage door
>
> ...

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

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