UPDATED BENCHMARK (ITS FASTER I PROMISE)

[RErender]

my test with vray 1.5 demo. Slower than before verison and something weird with reflections is going on.

Too fix the reflection turn on "sub-pixel mapping" in the color mapping roll-out. That will fix the edge but they are still a lot brighter, hopefully that is more correct.

As far as the qmc not sure yet what the factor of change is, I changed the noise thresh from .01 to .1 with only a slight change and what the heck is a "Latin Super Cube"?

[Paul Oblomov]

and what the heck is a "Latin Super Cube"?

A new method of sampling. Slower than default Halton, but ... well dunno why it's there I've read already about it on *.edu site, still don't remember which of 'em it said to be one of the method to organize groups/nets of smth...
As I remember - LSS can work in a more efficient way, than QMC. That's why Vlado had introduced rQMC (randomQMC) with LSS. In one word - it's about high dimensional integration problems.

edit
http://www.informs-cs.org/wsc98papers/076.PDF
Maybe this paper will help in understaing.

edit #2
I turned sub-pixel already on, to avoid these "edges"

[RErender]

thanks Paul,

must some scenario where it is advantageous?

[jujubee]

Wow. Just read the paper. Makes perfect sense.

[Paul Oblomov]

thanks Paul,

must some scenario where it is advantageous?

Don't even ask me I got the same "sense" as jujubee

[jujubee]

lol. You skipped to the last paragraph too?

[RErender]

lol me too

[vlado]

This particular scene would render with different speed compared to the older version, since there were some bugs in the 1.47.03 build (particularly with refractive materials) that made it take more shortcuts than it should (particularly for back-side reflections, I think).

As for the Latin supercube sampling, it is another way of combining pure qmc sampling with some random numbers to avoid some potential sampling artifacts. This is most obvious in PPT mode, where the default sampling may produce moire patterns over the image. Those go away with more samples, but can be somewhat distracting. The Latin supercube method does not suffer from this particular problem, although it tends to be slightly slower.

Best regards,
Vlado

[timmatron]

Hey Lucetius, thanks for the specs. I ordered one of those remote management cards for our new blade I'm putting together. Very exciting!

[crossbow]

Anyone have a file version that can be opened on max5.1? Ya I know, I'm living in the dark ages. I just model with primitives!

[GregM]

Hi all! My first post here as a registered V-Ray newbie. I just purchased V-Ray about 2 weeks ago, after testing it against Brazil. I was going to post these last week but it took awhile to get access to this forum. Unfortunately, I haven't had much time to actually learn how V-Ray works so I can start using it

BTW, I was fortunate enough to meet (and chat with) Vlado and Peter at SIGGRAPH. Real nice guys!

-Greg Montague

[Syclone1]

something odd in the 2nd image where the water passes over the light??

-dave

[GregM]

something odd in the 2nd image where the water passes over the light??

-dave

Those were done with the same (older) scene, not the new scene from Paul. Look at lukx's 1.5 demo render on this page - the same thing happened.

-Greg

[DaForce]

As mentioned in previous posts on this page you need to tick "sub-pixel mapping" in the color mapping rollout to fix up those funny bits in the RC2 render

Also, welcome GregM

[Syclone1]

ah, tkx daForce I didnt see that one

Just back from a week off of work...

-dave