Exterior wip

[Chakib]

Hi all,

I'm working on an exterior scene lately and i wanted to render it using Corona to test, so after converting and checking all materials, and here is the result.

Problem : the renders took 10 hours and the noise stills here :( i know i have poor cpu : core2quad q6600 )

I used :
Raytracing core : Embree tris : Precise
Low thread priority unchecked

PT-hd cache
path tracing samples : 16
max sample intensity : 10
Max records : 8000

So any tips to optimise speed  and clear up the noise would be  a big welcome !

[lacilaci]

are all those lightsources encapsulated in fully closed object with refractive material? I see at least one that seems to be enclosed in the interior... situations like this might be a part of the problem. I would lower down max sample intensity to 3 and increase light samples to 4... what version do you use?

[Chakib]

are all those lightsources encapsulated in fully closed object with refractive material? I see at least one that seems to be enclosed in the interior... situations like this might be a part of the problem. I would lower down max sample intensity to 3 and increase light samples to 4... what version do you use?

If you mean the exterior sphere lights : No they don't have refraction materials they are just simple emission material.
I use the legacy latest build for 2013 in the dropbox ( daily builds )

[lacilaci]

hm... don't know. But I'm working on a similar exterior scene with corona right now so I guess I'll soon be in the same situation myself... did you try pt+pt and bidir?

[Chakib]

hm... don't know. But I'm working on a similar exterior scene with corona right now so I guess I'll soon be in the same situation myself... did you try pt+pt and bidir?

No, will test them tonight
maybe the openexr envirenement map  is the problem, i will test a simple corona daylight sun and see if it helps

[Ondra]

definitely try increasing the direct lighting sampling multiplier

[Ludvik Koutny]

This could be related to the opacity bug...  that produces intense grain... 

[Chakib]

definitely try increasing the direct lighting sampling multiplier

tested it ( increased to 6 ) and after an hour it seems that it doesn't help much :s

[Chakib]

This could be related to the opacity bug...  that produces intense grain...

i think you're right because i did use the opacity map for the trees leafs, do you think it's the problem source?

[Ludvik Koutny]

Most probably, similar scene should take about 30 minutes to get reasonably clean on an i7 machine...  so maximally an hour on quad ;)

[Chakib]

Most probably, similar scene should take about 30 minutes to get reasonably clean on an i7 machine...  so maximally an hour on quad ;)

that's what i've told myself because an interior scene was clean in 2h with the q6600 so logically the exterior should take less than interior rendertime.

Also i forget to mention that the leafs have the transluency +opacity+ reflection+ bump.. lol will delete them and see the result

[Chakib]

Here is a 30 min render with daysystem corona without disabling the leaf opacity.

looks faster than the night version.

Problem : weird white dots in windows ! maybe because i did decrease the max samples intensity to 3.

second remark : the sun intensity is set to 1, but hete i see that the sun intensity is a little bit low, should i increase the intensity of the sun or leave it alone ?

[michaltimko]

dots = low MSI, try to increase or let it cook longer. If  you set PT+PT with MSI 0, corona is running in unbiased mode

And yes, you can set sun multiplier to any number, also try to play with sun size, bigger sun is producing nice smooth soft shadows.

[lacilaci]

Well I got the same problem now. I'm working on an exterior too and I'm too using latest legacy build and it seems that it is indeed related to opacity since I'm having weird bright dots on screen when I render trees with opacity mapped leaves. When I turn off opacity then it works as expected....

[lacilaci]

Ok... Forget what I just wrote.. My problem seems to be elsewhere. Turning off opacity did just greatly reduce the defect, but I'm still getting some bright dots. I guess it's the material (possibly reflection cause turning off translucency didn't help) testing different settings now.