Le 18-09-27 à 07:08, Jim Holsenback a écrit :
> On 9/26/18 8:12 AM, clipka wrote:
>> Am 26.09.2018 um 12:10 schrieb Jim Holsenback:
>>
>>> since i used radiosity, area lights, focal blur and am3...next time i
>>> have 62 hrs of free computer time i /might/ give it some more attention
>>
>> While quality settings normally have a tendency to stack up pretty
>> brutally, am3 is somewhat different in this respect, as it lets you get
>> away with lower settings on other features - provided they can be set to
>> produce random noise instead of biased artifacts.
>>
>> For example, area lights can be set to produce random noise by turning
>> on jitter. If you're using am3 anyway, you can then crank down on the
>> other area light quality settings.
>>
>> (Radiosity, on the other hand, is an example where this doesn't work, as
>> its artifacts are biased, not random-noise-ish.)
>>
>>
>> This comes in particularly handy when combining multiple "jitterable"
>> features, e.g. area lights + media + SSLT + micronormals (the latter two
>> always being "jittery" by nature).
>>
> 
> yeah i think i might give relaxing area light parameters and see if i 
> can see any gains in time. you didn't say anything about focal blur. not 
> jitter per se but it's mechanism noise based... right?

It's possible that you may be able to use adaptive 0 for those. If it's 
not good, try adaptive 1.
With adaptive 0, you can use some insane values for your area_light, like :
area_light x y  1025 1025 circular orient adaptive 0

and still get good performances.
Using adaptive 0 don't turn the feature off, neither will using adaptive 
off.
The only way to turn adaptive off is to remove adaptive from the definition.

Post a reply to this message