scene optimization

[lllab]

thanks to a tip from strat, i do it a little different approach:

for indoor first get colormapping plugin (dark multiplier 1.2+1.6 bright 1 liniear p.e.)

for GI:
i use strenght abou 60-90%
accuracy 95-100%
prepass 1/4for test -1/2final
diffuse 1 or maximum 2
stoch 50-100
min1-3
max 15-30
(last 3 points depends stronglyx on scene size, so test)

for dayligth scenes use very lowkey fill lights

i first start with 95% accuracy and go up until the sample distribution looks dense enough and good.

use rendertag and set all glas to be unseen by GI, set also tiny high reflective things to be unseen by gi. use rendertag and set all far distance envirements to ower GI accuracy.

save your GI solution once you fixed 90% of you image and reuse it then.
for big reflective areas use a tag check unseen by GI, make a copy of the original object, give it a material with no reflection, but same color and brightness and ckeck seen by GI and uncheck al others. also for displaced grass make a copy of the plane with no displacemant, use the simpel version for GI calculation.

just a few tipps for GI.

after all tests i can say that GI can be realtively fast in cinama, the sad thing is that when reflections and transpareny is used cinema seems to be slow with that. i think reflections are c4ds weekest part regarding speed, at least in combination with GI.

be sure to check the otions, lower reflection depth, shadow depth and ray depth as much as possible. i often use 15/1(2)/6 for that.

hope that helpes

cheers
stefan
lllab

edit for AA: playes a big role in speed: i use sinc for sharper look and best with 1x1, but lower the threshold to 5 or less. i find it renders faster and looks better@1x1 3%than @1x4 10%. i only give rendertags at points that might need higher AA settings( mostly 1x2 is enough)

[adem]

thanks for those tips, they are all very helpful,

esspesially the aa stuff, because I dont know much about that yet

thanks

[sundialsvc4]

Compositing is your single greatest friend, even for still-shots. It allows you to break down the shot in layers, back to front, and to render exactly what each layer of the scene requires. Obviously it also makes it much simpler to divide the work among several computers simultaneously.

Aggressively look for alternatives to computer-intensive operations, particularly including radiosity and ambient occlusion. Computer-graphic lights can do magical, impossible things like shining right through solid objects. They can be positioned right in front of the camera and never be seen. With compositing, you can light each major thing in the picture separately without worrying where the extra light might fall.

This is going to take a substantial amount of your careful time to set up, but each render will be much simpler and, when inevitable changes are needed, each one can be dealt with separately and you can "get back to looking at a complete image" much more quickly than you otherwise could. The finished result might not have the "theoretical perfection" of a complex radiosity/AO solution, but it will be pretty gosh-darned close and it will be ready soon.

[adem]

yeah I just begining to realise the importance of compositing & post processing etc..

[AdamT]

The thing that sticks out like a sore thumb is your use of diffuse depth of 20. That will absolutely kill your render times. As Ilab says, use a value of 1 or 2. If that doesn't give you enough illumination use some low-level fill lights.

[CGTalk Moderation]

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.