|
![](/i/fill.gif) |
Well I believe POVMan can do "proper" shading within pov, though I've never
had a play with it. Basically a ray-tracer is very well suited to shading,
even when using radiosity, because all you're ever doing is calling a shader
with an array of light colours coming from different directions (either from
radiosity or light sources) and asking it to tell you the colour emitted in
a specific view direction (for either a view ray, reflection, refraciton, or
radiosity sample). Of course I've oversimplified a little but it should be
pretty straightforward, just needs the syntax.
BTW, megapov's angle-of-incidence (aoi) pattern lets you fake lots of BRDF
style shading, so you can overlay 2 of those patterns with one based on the
angle to the camera and one based on the angle to the light.
--
Tek
http://evilsuperbrain.com
"Tim Nikias" <JUS### [at] gmx netWARE> wrote in message
news:45450fb6@news.povray.org...
> Tek wrote:
>> Actually as far as "shading" goes pov's pretty lame. e.g. conserve_energy
>> only works on transparent things, which means if you want realistic
>> gloss-paint you need a layered texture, and pov won't let you layer over
>> patterned textures... Seriously these kind of restrictions have no place
>> in a shader style language. In fact the whole syntax of the finish
>> statement means you have to bend over backwards to do anything clever
>> (like anisotropic shading, never mind BRDFs). Uh... oops looks like you
>> touched a nerve... :)
>
> Hehe, yeah, after a couple of courses at my university I was thinking the
> same thing. BRDFs, better layering etc would really be handy and to
> current standards. Then again, there are always some that implement new
> stuff, and in the recent years advances were rather fast-paced. People
> like to use the new stuff, but there aren't that many able to understand
> how they really work, let alone properly implement them into a versatile
> and powerful environment such as POV-Ray (and no, I don't want half-a**
> implementations like in Maya, where stuff crashes all the time). In some
> regard, there's also the overhead to be considered: we all know how
> radiosity and photons can start crawling with enough samples, simply
> because a raytracer has to raytrace and there aren't as many quick,
> GPU-accelerated approaches on that sector...
>
> Personally, I think that once POV-Ray would allow for multi-pass rendering
> (e.g. with switches for lights on certain passes, object on/off etc),
> people would be able to work around a lot of issues simply by rendering a
> couple dozen passes, e.g. to simulate various light-samples and later on,
> composite them. That's how I actually achieve quite a few effects anyways:
> rendering the scene once or twice and using the results to enhance the
> image (e.g. bloom or glares). I don't like how it's a a crude approach and
> people using different software might get different results.
>
>> But the fact that you can describe objects using a language, so you have
>> loops and conditionals and stuff, is just awesome.
>
> Total agreement on that.
>
> --
> aka "Tim Nikias"
> Homepage: <http://www.nolights.de>
Post a reply to this message
|
![](/i/fill.gif) |