Page 1 of 1

Just gzavye's tests topic

Posted: Sun Sep 05, 2010 4:20 pm
by gzavye
I open a place to post somme of my tests with Octane.
I begin with a subject than I have been working on since few weeks; it's around wood toys.
Elephnat done with Maya. Racing cars with 3dsmax.
1. the elephant
elephant01.png
First test in early version of Octane
toyRoom_elephants01.png
integration in a room when bump arrived

2. the racing cars
voitureCourse03_kodachrome25.png
one of several exposure tests I did
4voitures03_comp01.png
recent version with textures; no exposure cause Octane crashes
voitureCourse04_comp01.png
other version; wood floor made with floor generator

comments & critics allowed :mrgreen:

Re: Just gzavye's tests topic

Posted: Mon Sep 06, 2010 3:55 am
by jakchit
very great ideas and amazing renders. you have an eye for details that's for sure

Re: Just gzavye's tests topic

Posted: Tue Sep 07, 2010 3:17 pm
by gzavye
update on the last image :
new texture for the car
new texture for floor; to obtain more colorcontrast on image.

to do :
tweak the wheels
work on helmet

render at 1920x1280
voitureCourse04.png
futura 400CD response
voitureCourse05_futura400CD.png
kodachrome 64CD response
voitureCourse06_kodachrome64CD.png

Re: Just gzavye's tests topic

Posted: Tue Sep 07, 2010 9:24 pm
by kubo
I love the car. Your glass chair needs some more samples to get solved properly, you could try radiance trick of downsizeing http://www.refractivesoftware.com/forum ... f=21&t=878 or taking a second shots changing rrprob value for half the samples than the first one and composing them in photoshop using darker color layer for the lowest sample value render.

Re: Just gzavye's tests topic

Posted: Wed Sep 08, 2010 9:24 am
by gzavye
kubo : it's not the final version so I'm not going to cook a 4K version now. ;)
Can you tell me more about the rrprob trick?
What are the good value to have a correct specular plastic render?

Re: Just gzavye's tests topic

Posted: Wed Sep 08, 2010 12:19 pm
by kubo
the rrprob (russian roulette probability) is the chance of a ray to bounce (you can search the forum, or check the manual for a more in depth explanation). You should leave it as default, but one side side effect of changing is that it will render a different solution for the same shot, where the fireflies and noise in glass will be in a silght different position. Hence if you comp both (or as many as you need) in photoshop or gimp or the like, and add the second layer as darker color you'll get rid of those fireflies and have a cleaner glass. The thing is that you need less samples than with a single image, you use a fair amount for the first one, then half for the second one or 1/3 for the extra 2 and so on.
Radiance's more complete explanation about rrprob: http://www.refractivesoftware.com/forum ... lit=rrprob
Here is a little discussion about this trick http://www.refractivesoftware.com/forum ... 3&start=20

Re: Just gzavye's tests topic

Posted: Tue Sep 14, 2010 8:45 pm
by gzavye
kubo : thanks for the explanation

New scene with emiters done in 2.3_4
MT_loft07_comp01.png
C&C are welcome

Re: Just gzavye's tests topic

Posted: Sat Sep 18, 2010 3:21 pm
by gzavye
Various renders with the same scene

cameras research
MT_loft05_pre23_3.png
MT_loft06_pre23_3.png
same vue that in previous post with emiters
lot of noise instead of long renders (3h for 7000 s/px)
MT_loft08-agfa400PlusCD_pre23_4.png

Re: Just gzavye's tests topic

Posted: Thu Dec 30, 2010 2:39 pm
by gzavye
New scene; not yet finished.
I'm out of ideas for a better image :|
C&C are welcome :)

First version with sun. I was not very happy with that
early version
early version
I prefer that one, but something is missing; I can't put the finger on.
present version
present version
With other response
fauteuil08_Ekta64CD_pre23-5.png
extra views
detail01
detail01
detail02
detail02
detail03
detail03
And the post production one
photoshoped image
photoshoped image
that's it for now ;)