Page 3 of 3
Re: 62MT
Posted: Thu Apr 11, 2013 8:28 pm
by thiagobulhoes
great job !!!but, i agree with the flags.. the simulations can be much better, and i think is not necessary the final shot with the island in the ocean ( not so real as the other scenes - reveals thats a CG image). great work anyway, good job.
Re: 62MT
Posted: Thu Apr 11, 2013 10:34 pm
by nuno1980
No caustics?!!

But sorry you need ~4000-5000 or more samples to appear caustics.
Re: 62MT
Posted: Fri Apr 12, 2013 1:20 am
by ROUBAL
Great animation. Very nice work.
Re: 62MT
Posted: Fri Apr 12, 2013 8:50 am
by gabrielefx
Thank you.
pmc generates caustics...but not so much...
the sea is a sandwich made with the Dreamscape ocean object, a circular sea and a third circular plane that simulates the ocean's floor.
Because the Dreamscape's ocean id a trapezoidal mesh that adapts to the camera position you notice in the reflections this piece of geometry. Then the empty space must filled by another geometry.
This complex and large mesh generates caustics, If I delete the second circular mesh I get a caraibic effect, the sand must be positioned at -10 meters (the dreamscape sea object stays always at zero)
I will post a shot.
Regarding the flag I used and old simulation that I created 5 years ago into another cg animation (in that time I used vray)
I'm not an expert about the winds but when the scirocco wind blows you see the sea calm and the flag moves very fast.
Unfortunately Octane (now) doesn't support image sequence because I can generate a bmp sequence of foam that intersect the ship or the island beaches.
The ***REALLY BAD*** problem in this work is the exposure flickering.
Octane for Max now can't be used in Pro productions because the pmc kernel generates different exposed frames.
I used several brightness stabilizer in After Effects to reduce this issue, and several frames were deleted during the editing.
I could use the PT kernel but in this scene is impossible to use it because the sun and a very large reflective object....you can't believe how many fireflies generates the PT kernel with caustic blur set to 1.
Re: 62MT
Posted: Fri Apr 12, 2013 11:37 am
by neo83_gr
Top quality as always!
Re: 62MT
Posted: Fri Apr 12, 2013 11:51 am
by merid888
amazing Gabrielle, impressive !!!!
Re: 62MT
Posted: Sat Apr 13, 2013 7:11 pm
by gabrielefx
this example shows why we need no flicker in animations rendered with the pmc kernel.
It's impossible to use the pt kernel because it generates a lot of noise.
Also the pmc kernel produce faster and better results.
I used a caustic blur set to 0.03
5 bounces
parallelism set to 1
0,5 hot pixel remove
some photoshop retouch
the images were rendered with four GTX680
Compare yourself the two images
edit:
the last one image was rendered in "just" 36 minutes (pmc 4000M/s)
Re: 62MT
Posted: Mon Apr 15, 2013 2:36 pm
by nuno1980
gabrielefx wrote:this example shows why we need no flicker in animations rendered with the pmc kernel.
It's impossible to use the pt kernel because it generates a lot of noise.
Also the pmc kernel produce faster and better results.
I used a caustic blur set to 0.03
(...)
Yeah. But I keep no caustic blur (0.00) due to be more realistic.
the images were rendered with four GTX680
Compare yourself the two images
edit:
the last one image was rendered in "just" 36 minutes (pmc 4000M/s)
Thank!

Not 4000 M/s but yes ~4 megasamples/s or total 4000 samples during 36 minutes?

Re: 62MT
Posted: Mon Apr 15, 2013 8:40 pm
by gabrielefx
oh yes 4000 samples/sec and not 4000 M
the animation was rendered with caustic blur set to 0.3 and not 0.03
Re: 62MT
Posted: Mon Apr 15, 2013 9:13 pm
by nuno1980
gabrielefx wrote:oh yes 4000 samples/sec and not 4000 M
4000 samples/s are too slow!? lol But yes probably 4000 kilosamples/s.