Le 2021-05-02 à 19:55, Cousin Ricky a écrit :
> On 2021-05-02 1:21 PM (-4), Alain Martel wrote:
>>
>> For the object acting as your light sources, add :
>> radiosity{importance 1}
>>
>> That way, the sampling count will average about 100 for most of the
>> scene, but, those bright object will receive a lot of needed attention.
> 
> I tried radiosity importance, and the result was disastrous.  Maybe I
> used it wrong?
> 

Did you use the default directive ? It is crucial to use it to set the 
average importance so that most of the scene gets a reasonable sampling 
density. Also, the base count value seems to need to be larger than 10000.
I typically use count values from 60000 to 500000. Don't hesitate to 
increase that if needed. Just decrease the default importance accordingly.
Next, I tend to use pretrace_end of about 0.005 to 0.00125.
Use adaptive nearest_count, like 20, 3. This tend to give good results.
I also like to set nearest_count just under my pretrace_end value. So, 
something like 0.0045 to 0.0012.

NEVER forget to set radiosity{importance 1} for very bright objects. 
Also, set it to about 0.5 to 0.75 for bright objects that are relatively 
small.
In some cases, you may also need to increase it for luminous walls. 
Here, a value around 0.1 should be enough.

The following tend to give me good results :
#declare MXCount = 100000;// Set the maximum count size
#declare AVCount = 111; // Set the average radiosity sampling density

#default radiosity{importance AVCount/MXCount}

global_settings{
	radiosity{
		pretrace_end 0.0025
		count MXCount, MXCount*sqrt(2)
		nearest_count 20 3 // Can even use 20 2
		minimum_reuse 0.0022
// it seems that this NEED to be smaller than pretrace_end
		error_bound 0.3
// May need a smaller value in some cases
// Smaller value = need larger MXCount, but usually same AVCount
		low_error_factor 0.3
		adc_bailout 0.0001
		}
	}

Post a reply to this message