> I'm afraid <Orchid> is right. I expierenced the same problem in my 
> picture (look 2 threads before; "ouch..") with "mesh"-glasses. The first 
> dozend photon renders gave always some kind of grid like pattern or even 
> isolated sparkles, until I raised the gather radius to smooth them out. 
> But be aware, that your render time might get quit long by doing this.

As I understand it, "gather radius" effectively controls how "big" each 
individual photon is. I would have expected that to require shooting 
*fewer* photons, and hence be *faster*...

OTOH, I haven't actually tried it.

Post a reply to this message

Orchid XP v2 wrote:
> 
> As I understand it, "gather radius" effectively controls how "big" each 
> individual photon is. I would have expected that to require shooting 
> *fewer* photons, and hence be *faster*...
> 
> OTOH, I haven't actually tried it.

I'm never ever looked at the code so far, but as I understood the docs and the 
discussions in this forum, the radius and the gather values doesn't influence 
the number of photons to be shot at each induvidual object, this is done by the 
count/spacing value together with the target multiplier.
Radius and gather kick in when allready all photons are shot. They influence 
which of the stored photons should be taken into account for each induvidual 
point in your render.

To few photons will result in a coarse photon map which won't reveal much 
details. On the other hand lots of photons will give a finer photon map, but 
unless you adjust the gather values not all off them will be "collected" while 
searching through the photon map (the max value limits the number of photons to 
be collected and therefore to be taken into account). AFAIUI the gather min 
value decides if the the search radius has to be expanded adaptively to collect 
at least a minimal number of photons, and can cause problems when not enough 
photons are present.

to put it other words, I imagine the code looks roughly like this:

   -sort all photons stored in the photon map by means of the distance to the
    point rendered

   -take those photons which distance is less then the search radius

   -if the number of this photons is less then gather min value expand the
    search radius until the minimal number of photons to be collected is reached

   -if the number of photons found is larger then gather max value take those
    photons nearest to the point rendered

   -average the collected photons weighted by there distance


please correct me if this is nonsense; the more I understand the concept and how 
it's implemented the better (I think) I can make use of it and know which value 
have to be changed to get what I want.

... dave

Post a reply to this message

Hasan3 wrote:
> My photon settings:
> 
> global_settings {
>   max_trace_level 35
>   assumed_gamma 2.2
>   ambient_light rgb< 0,0,0>
> 
>   photons {
>     spacing 0.05
>     autostop 0
>     jitter 0.4
>     count 1000000
>     media 0
>     max_trace_level 25
>   }
> 
> }
> 
> 
> 

the docs ("3.6.3.2.1 Photon Global Settings") recommend to use either
<spacing> OR <count> ...

the number photon shot using <spacing> depends on object dimension, while with 
<count> the spacing will be estimated to fit the number of photons to be shot.

In your case I don't know which does have the higher priority ???

...dave

Post a reply to this message