Alain <aze### [at] qwertyorg> wrote:

> > It was mentioned in another thread that beta 3.7 treats radiosity differently.
> > How has it changed, or where can I find information on that? Of the greatest
> > importance to me, is how radiant intensity calculated for radiosity is
> > influenced by diffuse and ambient settings (both global and object).
> >
> >
>
> Mostly, all object must have ambient set to zero unless they are to
> actualy emit light. This have not changed since radiosity was introduced.
>
> Diffuse determine how an object will affect it's surrounding depending
> on the illumination it receives. No major changes from previous versions
> in the basic principles.
>
> The main differences is in HOW the sampling is actualy done and the
> multi-threaded capability allowing you to benefit from multicore CPUs
> and multiple CPUs.
> Also, some radiosity settings work a little diferently in the beta.
>
> /Most/ radiosity should render the same as with version 3.6.
>
> For all radiosity scenes, whatever the version used, you should have:
> #default{finish{ambient 0}} early in your scene. Best place: before any
> #include and #declare of texture and material.
>
> Seting ambient_lights 0 or some very small value is NOT something to do,
> as it will screw things whenever you have some ambient light source.
> The purpose of that setting is to tint the ambient component of the finish.
>
> Then, check any include and alter them so that they don't set any
> ambient unless the finish obviously should emit light, as the Luminous
> finish.
>
>
> Alain

I've implemented that setting. I now use:
#default{finish{ambient 0 diffuse 1}

It's a good idea, but I run into trouble with skies. Since radiosity (by my
experiment) uses both ambient and diffuse light, I get into a juggling act with
trying to illuminate both the camera and the scene evenly. It's hard to balance
the direct light from the sky to make it look natural, and have the right
illumination level in the shadows. I use light groups to colour balance as well,
so things can get complicated.

I believe this setting would be over-ridden by libraries, which isn't too great.
Of course, it won't affect textures defined in your sources.

The weird thing for me now is that radiosity doesn't seem to respect (or
understand) light groups. I know this is a curious complaint, and even way ahead
of my own needs, but it raises interesting questions in my mind...

Such as, should the radiosity for an object be calculated using only the light
sources in its light group?
Currently, it seems to me that this is not the case.

Post a reply to this message