PDA

View Full Version : NET | SMblur -slower- with less samples!?


tcastudios
01-22-2006, 06:28 PM
I'm checking out SceneMotionBlur and first did a 16 samples net rendering. It took 15 min on 2 quads and 1 Dual G5. The same scene and switching to 9 samples it took 45 minutes!
Switching back to 16 samples it took 15 minutes again.

What could possible be the cause if this?

Cheers
Lennart

dann_stubbs
01-22-2006, 06:50 PM
I'm checking out SceneMotionBlur and first did a 16 samples net rendering. It took 15 min on 2 quads and 1 Dual G5. The same scene and switching to 9 samples it took 45 minutes!
Switching back to 16 samples it took 15 minutes again.

What could possible be the cause if this?

Cheers
Lennart

maybe like GI - it takes more time to process the in between data from fewer samples

there is a sweet spot in GI that may be the same - too few samples and it takes longer to calculate the differences between them, and too many samples and the extra sample calcuations don't add any visible difference.

you probably could do a small matrix of sampels and results and narrow down the sweet spot - i'm sure as well as with GI it probably vary's per scene so there is no magic number to always use.

dann

tcastudios
01-22-2006, 07:04 PM
Hi Dann. There is only a simple light with softshadows involved. No GI.
Ten Figures, one transparant Cube and one reflecting Tube and a floor.
I' did a 25 samples render and it took 25 min.
I'll do a 5samples render and see what that brings..

Cheers
Lennart

tcastudios
01-22-2006, 07:38 PM
..and a 5 samples render took 25 minutes.. hmm I can only relate this to the only thing that is animated in the scene, a camera. Looking in the render loggs, each frame took about the the same amount of time to render, regardles of the sample settings.

My -guess- would be that the time added for the low sample settings is because the animation have to be calculated. Lower samples = More calculation, Higher samples= less calculation.
Doesn't fully explain the 9 sample time thou.

Cheers
Lennart

tcastudios
01-22-2006, 08:18 PM
OK. So it seems as it is the "AntiAliasing Restriction" option that makes this happen.
If checked, Cinema decides by itself the level of antialiasing.
Trying again and this time only running the first frame of the animation (to make the animation not kick in as a time factor).

AA restriction ON (As under the NET test)
5 samples > 28sec
9 samples > 52sec
16 samples > 18sec (restriction kicked in)
25 samples > 28sec (restriction kicked in)

AA restriction OFF.
5 samples > 28sec
9 samples > 52sec
16 samples > 1min32sec
25 samples > 2min25sec

So that must have been it. It's good to learn your tool, and better late then never:)

Cheers
Lennart

CGTalk Moderation
01-22-2006, 08:18 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.