RC5 25% faster than SP2 and little better quality!

[dlparisi]

just letting you know dmitri, but your sky is different between your two images and the RC5 is slightly warmer on the direct lit geometry, also probably due to the difference in the sky.

I'm not seeing this. Comparing in photoshop they look exactly the same other than some noise difference. 5x5 averaged eyedropper samples (to get rid of the noise) yield exactly the same numbers.

[gilpo]

Percy,
I thought the same thing too. So I opened them both in Photoshop, layered them and they are identical. I tracked the problem to my LCD. I have a Samsung and it seems colors drift a little from the top to the bottom of the screen. So, by having both visible in the post, the top image looked warmer than the bottom. Could be your problem as well.

Back on-topic: by layering them in PS and turning the top layer on and off, the quality seems to be the same, just the noise pattern changes. I wouldn't say that one is better than the other.

[^Lele^]

indeed, a difference of the two renders from Dmitri leads to noise-only differences, which peak at 3 (of an rgb255), and extend up to 8 (if they become a lot less pixels).
Allright with pixel-fu**ing, but come on, who can see 3 points of rgb difference in a noise pattern?
I'm planning some tests myself, this time on the sampler itself.

[Morbid Angel]

its Dmitry...
Im actually running a high quality renders right now. 1/64 at 0.001 and also same test with glossy reflection. Tonight we will see how it turns up.

[cecofuli]

Hi Dmitry,

i don't know why logaritmic is ON [???]... And the gradient ramp is only for background, not for GI. For GI i use a VRay Environment color. Anyway, simply scene is the best for testing. But in the "original scene" the difference in the rendering time remains. I don't know why... Maybe is DOF fault?

Here, the sky is the same.

[^Lele^]

Lol Dmitry, sorry for the typo XD

Well, the RC5 had rQMC, the SP2 has DMC.
Reason enough for two identical settings, with two different samplers, to produce different render times without a stark difference in quality.

[Morbid Angel]

cecofuli, indeed the times are different in your scene. Hance I reapproached it. Normally when going from older version to a newer version you MUST reset renderer. At one point when I was working on your scene, vray told me - unknown path sampler, using default. This ment that when I opened your RC5 scene in SP2, some how the path sampler remained from the old one, and was not recognized. That said, god knows what else has not ported over. That is why a full reset is in order. Im sure that neither the DOF or anything else is going to make a significant difference.

[RErender]

Just as an aside, http://www.popularscience.co.uk/features/feat16.htm is why the skies look different

[percydaman]

its probably drift in my monitor. I kinda actually suspected it might be the case, but was too lazy to bring em into photoshop.

[jujubee]

I tracked the problem to my LCD. I have a Samsung and it seems colors drift a little from the top to the bottom of the screen. So, by having both visible in the post, the top image looked warmer than the bottom. Could be your problem as well.

Thanks for this test - now I know my monitors are crappy.

[RErender]

You guys are funny, the difference is apparent on a good monitor too, it's an optical illusion... If Morbid would edit his post to add some blue space between the images you all could pull your monitors back out of the trash pile.

[Morbid Angel]

hahahaha
sorry I didnt mean to spark another debate

[pskdude]

haha

Lmao ! thas awesome

[ene.xis]

Lol Dmitry, sorry for the typo XD

Well, the RC5 had rQMC, the SP2 has DMC.
Reason enough for two identical settings, with two different samplers, to produce different render times without a stark difference in quality.

exactly, and i am betting dmc to produce identical quality at the same rendertime or even less

[Morbid Angel]

All tests are done with 1/64 dmc/qmc aa, at noise threshold of 0.001
I can conclude that rQMC defenently had a perfomace gain over DMC, but a minute one.