On 7/16/2013 4:25 PM, clipka wrote:
> Am 23.06.2013 18:00, schrieb Todd Carnes:
>> On 11/8/2012 8:20 AM, clipka wrote:
>>> A side effect of radiosity is that you /can/ also get illumination from
>>> a sky sphere or background (this is automatic), and even objects (this
>>> requires a suitable finish setting; previously this could be done by
>>> using a non-zero "ambient" finish term, but this was originally just a
>>> hack and has always been troublesome, so as of POV-Ray 3.7 the new
>>> "emission" term needs to be used instead).
>>
>> I'm sure that "under-the-hood" emission probably works differently from
>> ambient, but from where I'm standing it looks like all they did was
>> trade one word for another.
> 
> "They" was me, and there is a fundamental reason why the change was made.
> 
> The original intention of the "ambient" term was to provide a means to
> (very roughly) approximate what the radiosity feature is doing at much
> more computational costs: The light that an object is receiving
> indirectly from other objects (IOW ambient illumination, hence the name
> of the parameter) and reflecting back diffusely.
> 
> (Here's already an initial flaw that has /not/ been tackled yet: How
> much ambient light an object reflects in this way obviously does not
> only depend on the material properties, but also on how much ambient
> light the object is receiving in the first place, which in turn depends
> on the geometry of the scene; making the ambient term a part of the
> material settings is therefore a questionable decision, and some future
> revsion of the SDL - say, some POV-Ray 4.0 SDL - should correct it.)
> 
> Users of course found out that this "ambient" term also provided a means
> to let objects appear to be glowing - which became even more exciting
> when the radiosity feature was added.
> 
> Unfortunately, this dual use of the "ambient" term led to a conflicting
> situation: While a non-radiosity scene would need all materials to have
> a non-zero "ambient" term (and a non-zero ambient_light setting) to look
> reasonably ok, a radiosity scene would need all non-glowing materials to
> have a zero "ambient" term (or to have ambient_light be zero, but this
> would prevent the simulation of glowing objects entirely). So you'd
> essentially need two versions of each material: One for non-radiosity
> scenes, and one for radiosity scenes.
> 
> The introduction of the "emission" setting solves this conflict, by
> providing different knobs for each of the purposes "ambient" had come to
> be used for: The "ambient" setting can now be used without limitations
> to simulate the effect of indirect light in non-radiosity scenes,
> without undesired side effects in radiosity scenes (as it is turned off
> entirely there, being replaced by the radiosity mechanism); at the same
> time, glowing materials can now be modeled without limitations using the
> "emission" setting instead, as it is available in both radiosity and
> non-radiosity scenes.
> 
> (As a side note, another - albeit less important - difference is that
> the "emission" setting is insensitive to the ambient_light parameter.)
> 

Thank you for the very detailed explanation. I've learned a lot from it. :)

Todd

Post a reply to this message