pt vs pmc

Sat Jun 09, 2012 7:56 am

more renders to come

edit:
few adjustments on each still.

Sat Jun 09, 2012 9:15 am

Amazing as always.
How did you do that carpet...that must be modeled.

Sat Jun 09, 2012 5:03 pm

Pullakk wrote:Amazing as always.
How did you do that carpet...that must be modeled.

thank you,
it's Max fur

Sun Jun 10, 2012 6:46 pm

Amazing job Gabriele!

I like your comparison between PMC and PT, all settings and materials where leaved the same?
why the toothbrushes in PMC doesn't have any color?, and why such a difference with the soap and perfume bottle refractions?

I'm in love with your last render, render time?

waitnig for your next job!

Saluti,

Paolo

Sun Jun 10, 2012 7:39 pm

paoloverona wrote:Amazing job Gabriele!
I like your comparison between PMC and PT, all settings and materials where leaved the same?
why the toothbrushes in PMC doesn't have any color?, and why such a difference with the soap and perfume bottle refractions?

I'm in love with your last render, render time?

waitnig for your next job!

Saluti,

Paolo

Hi, I did few adjustments on objects position and materials
The full resolution renders are at 2560x1810
PMC 10 hours
PT 7 hours

PMC renders have less noise but you need more hours

What I have discovered is:

In the case we have a good illumination we should select the PT kernel, on the contrary we will use the PMC.

I analyzed the best renders created with Arion and iray. The users don't use tiny emitters and use hdrs. They don't put in the scene the ceiling.
If we render an object inside a closed box we will get few samples/second. If we delete the ceiling and put an hdr the samples will be 10X more.

Mon Jun 11, 2012 9:35 am

super!, yours tests are lot helpful for the refractive's community

...unfortunately most of the time the ceiling couldn't be "hidden", should be fantastic if inside octane there was the option for the objects to don't cast shadows

hoping the best for the next releases of octane.

Saluti,

Paolo

Tue Jun 12, 2012 10:31 am

About that - you can set a slight transparency to the ceiling and the render will be faster and still see the ceiling.

Tue Jun 12, 2012 12:04 pm

lenreed wrote:About that - you can set a slight transparency to the ceiling and the render will be faster and still see the ceiling.

It's a good tip!

I use the same tip to create a slight transparent hemispherical sky passed by the sunlight.

Wed Jun 13, 2012 4:23 pm

great work as always gabrielefx. but 7 hours with 4 gtx 480.. is too much, no? i think you have same results using biased renders, like vray in brute force mode. just my opinion.

hugs

Thu Jun 14, 2012 6:54 am

lzanlorenzi wrote:great work as always gabrielefx. but 7 hours with 4 gtx 480.. is too much, no? i think you have same results using biased renders, like vray in brute force mode. just my opinion.

hugs

Oh yes.
If you look at the render with the rotated camera you will notice a more natural illumination.
The other renders were illuminate with 3 rectangular lights (two hidden) behind the camera.
I don't like artificial or tricky illumination, these results are similar to many Vray renders I do.
In Octane the light is distributed naturally, it goes everywhere.
In Vray using bruteforce for primary and secondary rays you can't achieve natural illumination using only one sky portals and two thin emitters behind the mirror.
In Vray you will get splotches, dark areas and a lot of noise except if you pump the dmc sampler.
Gpu render it's in a early state. I'm waiting for the next Kepler gpus to get 3x performances.