stm31415 wrote:
> You can smooth out the photons by using a small image to render a photon
> map, like two pass radiosity. Or are you already doing that?
> 
> -s
> 5TF!
> 

AFAIK it's a little bit more tricky with phothons, as in radiosity there will be 
new samples when the solution is insufficient while photon precalculation is 
independend from render dimensions. The size of an single photon sparkle is (so 
I think) estimated by the radius keyword, which specify the space to look for 
stored photons and average them weighted by distance. With gather you can fine 
tune this step by specifying the min/max number of photons to take into account.

What I actually did is something you could call a "2-pass method"; first I build 
the photon map by altering the spacing/count value until I think the solution is 
sufficient eventhough often quit blotchy. (spacing strongly depends on the 
dimension of your objects)
In a second pass I reload the map and adjust radius und gather until artefacts 
are smoothed out but caustics remain well defined. To large values will lead 
caustic details to vansih.

All in all there are normally multible passes before final render:
  - 2 for radiosity solution with disabled micro-normals (blurred reflection)
    and disabled photons @low-resolution
  - 2 for photons @med-resolution and aditional close-ups to verify caustic
    details.
  - 1 combined radiosity, photon render @low-resolution to adjust light level

But I'm affraid that there are allways a lot more prerenders as I always forget 
to switch on/off features so that I spoil the result.


... dave

Post a reply to this message