Milk (regular) optimized

[Meelis]

Hi

I made Milk material
At the moment i don't have free time to render good quality preview.
If you can render it, you may upload it under your name to MDB.

~ 1 h / 900 S/px

[zeitmeister]

Great, thank you very much!

[Meelis]

Great, thank you very much!

Thx

[Pibuz]

Hi Meelis! Thank you!
..but why the white dots?

[dag]

Hi Meelis! Thank you!
..but why the white dots?

Expired milk.

[Meelis]

Hi Meelis! Thank you!
..but why the white dots?

Expired milk.

Maybe just noisy. The preview was in path tracing mode, 1000 MNCR.

The milk will get better if live it on table for few weeks

MNCR 10000
Supersampling 2
MLT Bidir
http://static3.nagi.ee/i/p/690/79/172698739745d7_o.jpg

[Godzilla]

Very cool scene Meelis. If you would like, I could render it until it's noise free for you?

[Meelis]

Very cool scene Meelis. If you would like, I could render it until it's noise free for you?

Thanks

Are you shure, it's not that cool as it's slow (+ that ugly shadow on backwall created by emitter plane)
I'm rendering it on network with C2D e4500 2,2 GHz + C2D e7400 2,8 GHz.

I will upload packed indigo scene tomorrow morning (~ 8 hours from now)

Here is update
http://static3.nagi.ee/i/p/691/35/1728391997140d_o.jpg
http://static3.nagi.ee/i/p/691/35/172838374d21ec_o.jpg

[Meelis]

Here is the unpacked pigs (80,5 MB) and repacked with 7zip (36,1 MB)
https://docs.google.com/leaf?id=0B1ETge ... NmNk&hl=en

Hope it still works

[Pibuz]

Hi Meelis! Thank you!
..but why the white dots?

Expired milk.

Maybe just noisy. The preview was in path tracing mode, 1000 MNCR.
http://static3.nagi.ee/i/p/690/79/172698739745d7_o.jpg

..to me, they look like fireflies..
That's why I asked. Do they go away after further calc time?

[pixie]

..to me, they look like fireflies..
That's why I asked. Do they go away after further calc time?

after 24 on 2 i7 I couldn't see much off an improvement..

[Pibuz]

..mmmmmhhh..
I guess by that time the "regular noise" should be gone by far...

[Meelis]

Maybe i need more super samples than 2. (i need to add ram in slave computer for that)

If more super samples doesn't help i might try:
* I used 0.9 white, so i can go with 0.7
* I should lower SSS uniform value from 500 to 200 maybe.
* Can MNCR 10000 be problem, i go with default 1000

111 h

[suvakas]

Hmm.. Wasn't there much cleaner SSS in some earlier Indigo builds long long time ago ?
11k samples should be crystal clear. I remember having decent sss with 6k spp. If sss works like this now, then it's not usable at all. Ono? Lyc? Any comments ?

[lycium]

Wow, that's quite a noisy result for such an incredibly long rendertime...

My suggestion is to use MLT path tracing, non-bidir. Bi-dir tries to make many connections inside the SSS medium that don't contribute any light paths, this is somewhat inefficient even if MLT is helping it along.

edit: upon further investigation of the render settings, i suggest to never use gaussian splat or downsize, it's slow and doesn't look as good as fastbox + MN; also, you should prefer circular apertures unless you really want a particular bokeh effect, since rendering with generated apertures is considerably less efficient than circular ones.