Thanks Alain

thats very helpful advice. I have taken on board your suggestions and get good
scenes now.

One thing I have noticed that is strange in a scene I have created is
absorption. If I shoot photons towards a box with ior = 3.2 that has a thinner
box inside it with same ior but with exponential absorption (alpha) and
thickness (t_box2) I get unexpected results. I use 'difference' to cut two
45degree angles through the two boxes, with the angled sides facing each other
(resulting shape like a truncated pyramid). If I shoot photons so that they hit
an angled side at a position in the midpoint of the absorbing box the photons
are reflected towards the other 45degree face where they are again reflected and
are sent back towards the light source (offset slightly but parallel).

From my results of this it looks like the reflected rays are not absorbed by the
absorbing layer (the path length through the absorber for these photons should
be long because the reflection is in the long dimension of this absorber, rather
than through its smaller thickness).

In a more simple scene without the 'difference' when i shoot rays at the boxes
the returned intensity looks right (absorption is exp(-2*t_box2/alpha) and
multiple reflections seem to be accounted for too).

Is it possible that absorption is calculated in a way that means using
difference to shape an object wont allow correct estimate of pathlength of
photons through the layer?

Alain <kua### [at] videotronca> wrote:

> > Thanks Alain
> >
> > the reason for the large gather value and the low adc_bailout is because I found
> > that I need these for a the scene I am trying to model. The code I attached in
> > my first posting was a simple version of this scene - in my main scene the
> > object that reflects light back to the diffuse screen which is viewed by the
> > camera is more complex - it is several layers of different ior, and i use
> > fresnel reflection
> >
> > stu
> >
> >
>
> If you just want a very, almost zero, adc_bailout, why not use
> adc_bailout 0. That way, the tracing depth will be controled only by
> max_trace_level, that you can set up to 255.
>
> I don't think that you realy, even with 100 stacked objects having all
> different ior, need to play around with the gather parameter.
>
> Finaly, fresnel only affect the reflectivness of your object reletive to
> it's ior and the angle of the incident rays or photons. It have no
> effect on the photons distribution.
> When using fresnel, you should add conserve_energy in the finish of that
> object.
>
> Only the first object to receive the light need the photons block. It
> controlls where the photons are shot. Secondary photons, after any
> refraction of reflection, WILL be affected by any refractive /
> reflective object they emcounter. If you don't want those to collect
> photons and have them be visible on it's surface, add photons{collect off}.
>
> When using count, the photons are distributed between all target
> objects, and any object behind any other will have photons shot toward
> it, but those photons will get intercepted by the intervening object and
> get lost.
> It may be why you need 100 millions photons.
> Also, if you loose over 90% of your photons, it may explain why you
> needed to play around with gather.
> Let say, 20 stacked target objects and the projection plane also set as
> target, means that your count is divided among 21 objects, but only
> those received by the front object realy count.
>
>
>
> Alain

Post a reply to this message