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

> > I am working on a scene with a Christmas tree with tiny white lights all over
> > it.  Since there are too many individual lights to model them as light sources,
> > I am using media emission and radiosity.
> >
> > When I crank up the emission brightness enough so that the lights light up the
> > surrounding tree branches realistically, they also create large bright spots on
> > the walls surrounding the tree, which is not very realistic.  If I were modeling
> > them as light sources, I would just add fading to solve the problem, but it
> > doesn't look like there is a way to do that with radiosity and emissive media.
> > I've played with max_sample and adc_bailout, but they uniformly reduce the
> > effect of the bright objects, so I lose the illumination of the tree branches.
> >
> > I was thinking maybe the fade_distance and fade_power in the interior statement
> > might work, but they seem to apply to light passing through the object, not
> > emissions from the media in the object.
> >
> > Am I missing something, or is there just no current support for fading and
> > radiosity from emissive objects?
> >
> > -- Chris R.
> >
> >
>
> You are chasing the wrong things, using the wrong setings.
> What you need to do is to increase the count. You may need to increase
> it to over 2000.
> Please, use version 3.8, as it offers the importance feature.
>
> In your global_settings radiosity block, crank up the count. Use the two
> values format : count 10000, 111111
> Then, add a #default statement :
> #default{ radiosity{importance 10000/75} }
> That'll ensure that on average, you'll use about 75 samples.
> I often use #declare for that.
>
> #declare MaxCount = 1e6; // Yes, count can now get that big.
> #declare AverageCount = 98;
> #default {radiosity { importance AverageCount / MaxCount } }
> global_settings{
>    radiosity{
>   count MaxCount, MaxCount * 1.11+1
>   //any other radiosity settings
>   }
>  }
>
> THEN, add this radiosity block to your radiosity lights :
> radiosity{ importance 1 }
>
> There is no tweak for the fading of the radiosity effect of emissive
> objects. There IS fading as a side effect of how radiosity works. The
> fading do follow the inverse square law, but is also affected by the
> actual surface of the emissive object. If the emissive object is a plane
> or very big sphere surrounding the scene, then, the illumination is
> independent from the distance.
>
> Rule of thumb :
> If the object is small relative to the distance, it's contribution is as
> the inverse square law. Distance is at lease 20 times the size of the
> object. Should mostly be your case.
> If the object is large relative to the distance, it's contribution will
> gradually change to straight inverse of the distance. Distance similar
> to the size up to a tenth of the size.
> If the object is very large relative to the distance, then, the distance
> stop being a factor. Size of the object is at least 100 times greater
> than the distance.

Thank you for the detailed explanation.  I have removed the fading settings for
the interior of the bulbs and added in the radiosity { importance 1 } setting to
them instead.  In addition, I bumped the count up considerably higher and set
the default importance to a lower value.  The attached image shows now that near
the top of the tree, where the lights are further away from the walls, there is
little illumination from the lights.  As you move further down the tree and the
branches are closer to the walls, there is more illumination, and the floor is
the most highly illuminated.  The tree branches are illuminated as I would
expect being that close to the lights as well.  (Note: there are not light
sources in this scene, only the emitting media in the light objects and
radiosity.)


-- Chris R.

Post a reply to this message

Attachments:
Download 'christmas_tree.png' (657 KB)

Preview of image 'christmas_tree.png'