Le 18-08-07 à 06:33, Stephen a écrit :
> On 07/08/2018 11:29, Stephen wrote:
>> No, don't follow my maths.
>>
>> Using the data from this site. I did a quick model.
>> http://www.alcyone.com/max/reference/scifi/ringworld.html
>> I adjusted the Shadow square ring radius until the day/night cycle 
>> looked equal.(multiplied the radius by 0.587587699)
> 
> A close up of the terminator on the Ringworld.
> The area light does make a difference.
> Using
>   light_source {  // Light_Source2
>    < 0.000000, 0.000000, 0.000000 >, colour rgb <1.000,1.000,1.000>*1.500
>    area_light <3.400000,0.000000,0.000000>, 
> <0.000000,3.400000,0.000000>, 23, 23
> 
>    circular
>    orient
>    adaptive 5
> 

The 23, 23  get bumped up to 33, 33

adaptive 5 mean start at subdivision 5, or start with 33, 33 samples. 
This effectively disable the adaptive optimization.
In the present case, adaptive 0 is good.

adaptive 0 start at 2 by 2
adaptive 1 start at 3 by 3
adaptive 2 start at 5 by 5
adaptive 3 start at 9 by 9
adaptive 4 start at 17 by 17
adaptive 5 start at 33 by 33
adaptive 6 start at 65 by 65
...

You need a number of samples at least 2 step up to get some appreciable 
benefit. For adaptive 0, use at least 5 by 5, and 9 by 9 or more in most 
cases.
For adaptive 1, better to start around 17 by 17.
For adaptive 5, you start to see an improvement around 129 by 129...

Post a reply to this message