|
 |
William F Pokorny <ano### [at] anonymous org> wrote:
> An image result using povr's new reflection block inv_pnormal_cure
> keyword. Using inv_pnormal_cure 5.22, so 22% of the perturbed rays are
> forced into the spheres as if the surface somewhat porous to light.
>
> More detail in povray.pov4.discussion.general.
>
> Bill P.
Looks good, I have no idea what it means ;-) Is this similar to the SSLT
feature? If so is it quicker to render?
Sean
Post a reply to this message
|
|
|
|
On 2/16/22 09:33, s.day wrote:
> William F Pokorny <ano### [at] anonymousorg> wrote:
>> An image result using povr's new reflection block inv_pnormal_cure
>> keyword. Using inv_pnormal_cure 5.22, so 22% of the perturbed rays are
>> forced into the spheres as if the surface somewhat porous to light.
>>
>> More detail in povray.pov4.discussion.general.
>>
>> Bill P.
>
> Looks good, I have no idea what it means ;-) Is this similar to the SSLT
> feature? If so is it quicker to render?
>
> Sean
>
Thanks.
While pushing feature boundaries in povr, I'm often not sure what it
means. :-) What might users do after a new option is available. How
might some new enabled behavior interact with other features. I might
have ideas but, I don't really know.
I added this mode(5.probability) thinking about reflections with porous
surfaces, say, a fabric weave on a lamp shade(a). Some of the rays at a
ray-surface intersection reflect through openings to what is behind.
For this image I was thinking, some, about a faked subsurface effect
but, as modeled, it's only one surface behind the outer one. A simple
set up like this is probably faster than SSLT(b). However, other than
fixing some SSLT bugs which were really solver / code set up bugs, I've
not much used SSLT - so I don't know for certain.
Your question got me thinking. What if we model more surface layers? I
don't think there is any fundamental reason we couldn't model more
depth, with more layers. Could we get a result close to or, perhaps
sometimes, better than the existing SSLT? There are issues for more
complex geometries where I think the current SSLT code is not very
accurate.
Bill P.
(a) - This happens too where certain wavelengths of light see through a
surface which to other wavelengths might be opaque, etc.
(b) - This method uses both the new 'normal { micro }' pattern and the
inv_pnormal_cure option which grabs some portion of the reflected rays
and pushes (re-directs) them through the surface. The overall effect
only 'activates' with anti-aliasing (AA) on because the micro normal
pattern is set up to lean on the AA mechanism to control the number of
samples. Where AA is off or at a coarse setting, the renders will be
much faster but, look noisy.
Post a reply to this message
|
|
