Le 18-04-19 à 20:35, Cousin Ricky a écrit :
> I thought I'd give Jadzia's image a try.  I did include a mirror: a 
> glass plate with aluminum backing.
> 
> Elapsed time was 15.5 hours.  Here are some stats:
> 
> Number of photons shot:        78466522
> Surface photons stored:         1060010
> Gather function called:        51789326
> 
> Render Time:
>    Photon Time:     11 hours 34 minutes 58 seconds (41698.012 seconds)
>                using 11 thread(s) with 164250.003 CPU-seconds total
>    Radiosity Time:   0 hours  4 minutes  2 seconds (242.569 seconds)
>                using 8 thread(s) with 1918.047 CPU-seconds total
>    Trace Time:       3 hours 51 minutes 53 seconds (13913.227 seconds)
>                using 8 thread(s) with 108057.917 CPU-seconds total
> 
> Most of that time was photons interacting with the mirror, and that even 
> by using photons { target 5 }.  Based on draft runs, I estimate that 
> about 10 of the 11.6 photons hours was just the mirror.  But some 
> caustics from the mirror are visible on the walls, so I kept the mirror 
> as a photons target.
> 
> Attachment: cube_of_cubes.jpg

For those mirrors, you could save a lot of photons time by using copies 
of the mirrors as projected through objects with lights mirrored 
relative to the mirrors. Don't forget to remove the protons{target} from 
the mirrors. The mirrors will still reflect any secondary photons they 
receive.

Another photons trick : For the inner cubes, don't set them as target. 
The photons shoot at them are lost because they are intercepted by the 
outer cubes. Have at most two layers of your cubes set as target.

Post a reply to this message