PDA

View Full Version : G4 G5 Radiosity



Paul Goodrich
04-13-2005, 05:05 PM
Hi,
I'm rendering a scene (animation) with Radiosity/Monte Carlo 2x6 (yes I must be a masochist). I was wondering if I split the render up between my G5 and several G4's does anyone know if there will be any visual difference in the render? Thanks
Paul Goodrich

NigelH
04-13-2005, 11:02 PM
Just render a couple of sample frames on the G5 and one or two G4's then take the renders into Photoshop and view them (layered) in 'difference' mode. If they're exactly the same, the image should show completely black.

Captain Obvious
04-14-2005, 03:47 AM
Monte Carlo doesn't use random samples, as far as I know, and should thusly look exactly the same regardless of what you render it on.

Lynx3d
04-14-2005, 04:41 PM
Monte Carlo doesn't use random samples, as far as I know, and should thusly look exactly the same regardless of what you render it on.

In case of LW (8) you seem to be right, but strictly spoken "Monte Carlo" by definition uses random samples...but since you can't calculate true random numbers and quasi-random numbers can be better anyway that's what's usually used, and called "Quasi-Monte Carlo"...

Ozzie
04-14-2005, 09:17 PM
In case of LW (8) you seem to be right, but strictly spoken "Monte Carlo" by definition uses random samples...but since you can't calculate true random numbers and quasi-random numbers can be better anyway that's what's usually used, and called "Quasi-Monte Carlo"...

Yeah that's what I was going to say. :D

CHEERS.

Mark

Ge4-ce
04-15-2005, 08:20 AM
Speaking about random"ness"

radiosity was never a problem for me, but just plain procedural textures were.

Is there a list of procedural textures that act "random"?

I know Ripples move, And I think "Dented" also has a random something and there are others. I just don't exaclty know wich ones..

I once messed up a complete render over multiple machines because the textures jumped when rendered on a different computer.

Captain Obvious
04-15-2005, 02:09 PM
In case of LW (8) you seem to be right, but strictly spoken "Monte Carlo" by definition uses random samples...but since you can't calculate true random numbers and quasi-random numbers can be better anyway that's what's usually used, and called "Quasi-Monte Carlo"...
Well, yes, but... bah. I did mean the so-called Monte Carlo radiosity in Lightwave, not the actual Monte Carlo statistical sampling method... :p