Hi all!

I'm not sure if this has already been discussed and
what came of, nontheless, I've thought about
blurred reflections somewhat, and this is what
I've come up with:

A new keyword placed in the finish statement
would activate it, much like blur_samples
activates focal blur in the camera.
To control it, two values would be needed: one,
to control how great the difference from actual
reflection angle is allowed (either in degrees,
like 90 for hemispheric blurring, or in percentage,
where 100 would be spherical and 50 hemispherical),
and then a distribution value, to weigh the blurred
reflections towards the original reflection direction.

Still, I'm not sure if spherical is actually needed, and if
hemispherical would be enough. In the latter case,
the radiosity distribution might be used, and a value would
weigh the different samples different in accordance to
the discrepancy to the original reflection-direction.

The advantage of this would be that everything is done
as always, with the modification that not one reflection
ray is shot, but perhaps maybe 30 or 40. This would
get exponential when used on all surfaces, but thats
what always happens with raytracing, right? :-)

I've been wondering about this ever since using
blurred reflection via averaging textures. That
method doesn't supply enough control over the
reflection scattering and may yield in different
regions of good vs bad blurring, e.g. when
the normals averaged are too similiar to actually
scatter the reflections enough.

Sadly, I'm not much into POV-Code, so someone
else might have to do this...

As a side-note, why hasn't blurred reflections been
implemented in the official POV? Just curious...

Regards,
Tim


-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

Post a reply to this message

"Tim Nikias v2.0" wrote:
> 
> [...]
> 
> As a side-note, why hasn't blurred reflections been
> implemented in the official POV? Just curious...

http://www.students.tut.fi/~warp/povQandT/languageQandT.html#blurredreflection

For the other stuff you have written about - see:

Subject: Importance sampling of phong brdf
Date: Mon, 14 Apr 2003 20:12:50 +0200
Newsgroups: povray.unofficial.patches
From: "Mael" <mae### [at] hotmailcom>
news://news.povray.org/3e9afa23$1@news.povray.org
http://news.povray.org/3e9afa23$1@news.povray.org

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 28 Feb. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message

Well, I've known about how to do it, and the
QandT doesn't answer why it isn't in POV-Ray 3.5,
so that question is still unanswered.

As for the functionality: The described technique
doesn't really guarantee that the new samples
will be spread in a hemisphere, which is what
I'm after. In older pics and tests, I've noticed that
some regions have nice blurring, while others
are pretty much only blurred along a certain
direction.

The first link you gave works for Outlook to get
that exact post, but I've got to figure how to
access the entire thread. The second link doesn't
work, as the web-view is disabled for now.

But thanks for the links anyway, they might shed some
light for beginners, but to be honest, I haven't learned
anything new from them yet (got to figure Outlook
first... :-)

-- 
Tim Nikias v2.0
Homepage: http://www.digitaltwilight.de/no_lights
Email: Tim### [at] gmxde

>
>
> "Tim Nikias v2.0" wrote:
> >
> > [...]
> >
> > As a side-note, why hasn't blurred reflections been
> > implemented in the official POV? Just curious...
>
>
http://www.students.tut.fi/~warp/povQandT/languageQandT.html#blurredreflection
>
> For the other stuff you have written about - see:
>
> Subject: Importance sampling of phong brdf
> Date: Mon, 14 Apr 2003 20:12:50 +0200
> Newsgroups: povray.unofficial.patches
> From: "Mael" <mae### [at] hotmailcom>
> news://news.povray.org/3e9afa23$1@news.povray.org
> http://news.povray.org/3e9afa23$1@news.povray.org
>
> Christoph
>
> -- 
> POV-Ray tutorials, include files, Sim-POV,
> HCR-Edit and more: http://www.tu-bs.de/~y0013390/
> Last updated 28 Feb. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message

"Tim Nikias v2.0" wrote:
> 
> Well, I've known about how to do it, and the
> QandT doesn't answer why it isn't in POV-Ray 3.5,
> so that question is still unanswered.

Quoting:

"You can get blurred reflection which works pretty much like the one in
MegaPov, and which can even look better and be a bit faster to render..."

I think this qualifies for an answer (although it is not necessarily the
original reason why this megapov 0.x patch is not in POV-Ray 3.5).

> As for the functionality: The described technique
> doesn't really guarantee that the new samples
> will be spread in a hemisphere, which is what
> I'm after.

I don't understand.  You surely can't distribute the samples uniformly on
a hemisphere around the reflection ray direction because then samples
would be below the surface.  Maels patch allows detailed control over how
wide the samples are distributed although the density distribution as he
has written it corresponds to the phong model.

> 
> The first link you gave works for Outlook to get
> that exact post, but I've got to figure how to
> access the entire thread. The second link doesn't
> work, as the web-view is disabled for now.

If the information i have printed is not sufficient to locate the thread
i'm afraid i can't help much.  This is what my search script spits out and
i think this is pretty much already.  Using the subject and the search
function of your newsreader or just the date of the post should be enough
to locate the thread.

Christoph

-- 
POV-Ray tutorials, include files, Sim-POV,
HCR-Edit and more: http://www.tu-bs.de/~y0013390/
Last updated 28 Feb. 2003 _____./\/^>_*_<^\/\.______

Post a reply to this message

> Quoting:
>
> "You can get blurred reflection which works pretty much like the one in
> MegaPov, and which can even look better and be a bit faster to render"
>
> I think this qualifies for an answer (although it is not necessarily the
> original reason why this megapov 0.x patch is not in POV-Ray 3.5).

The "averaging normals" trick is more flexible than MegaPOV's blur patch (it
allows anisotropical reflections and a variable strength of blur) but in
certain cases, it's also much slower. I have discussed this earlier and
prooved it. I can't find my post at the moment when the webview is disabled.

Averaging the normals are slower when two or more reflective surfaces causes
the ray to ping-pong. I won't try to explain the technical reason, when I
don't know the source code for POV-Ray. But the speed difference is huge, in
those cases.

Regards,
Hugo

Post a reply to this message

Tim Nikias v2.0 <tim### [at] gmxde> wrote:
> A new keyword placed in the finish statement
> would activate it

  No. AFAIK a more generic solution has been planned for the distant future.

  (Which is good, really. Better give a generic solution which you can use
for this and tons of other effects than cluttering the SDL with tons of
keywords for tons of unique, rigid, non-configurable features.)

-- 
#macro N(D)#if(D>99)cylinder{M()#local D=div(D,104);M().5,2pigment{rgb M()}}
N(D)#end#end#macro M()<mod(D,13)-6mod(div(D,13)8)-3,10>#end blob{
N(11117333955)N(4254934330)N(3900569407)N(7382340)N(3358)N(970)}//  - Warp -

Post a reply to this message

I might be out of the subject, but it seems to me that Jaime Piqueres 
did similar effects using micro-normals. You can have a look at:

http://www.ignorancia.org/t_experiments.php

JC

Tim Nikias v2.0 wrote:
> Hi all!
> 
> I'm not sure if this has already been discussed and
> what came of, nontheless, I've thought about
> blurred reflections somewhat, and this is what
> I've come up with:
> 
> A new keyword placed in the finish statement
> would activate it, much like blur_samples
> activates focal blur in the camera.
> To control it, two values would be needed: one,
> to control how great the difference from actual
> reflection angle is allowed (either in degrees,
> like 90 for hemispheric blurring, or in percentage,
> where 100 would be spherical and 50 hemispherical),
> and then a distribution value, to weigh the blurred
> reflections towards the original reflection direction.
> 
> Still, I'm not sure if spherical is actually needed, and if
> hemispherical would be enough. In the latter case,
> the radiosity distribution might be used, and a value would
> weigh the different samples different in accordance to
> the discrepancy to the original reflection-direction.
> 
> The advantage of this would be that everything is done
> as always, with the modification that not one reflection
> ray is shot, but perhaps maybe 30 or 40. This would
> get exponential when used on all surfaces, but thats
> what always happens with raytracing, right? :-)
> 
> I've been wondering about this ever since using
> blurred reflection via averaging textures. That
> method doesn't supply enough control over the
> reflection scattering and may yield in different
> regions of good vs bad blurring, e.g. when
> the normals averaged are too similiar to actually
> scatter the reflections enough.
> 
> Sadly, I'm not much into POV-Code, so someone
> else might have to do this...
> 
> As a side-note, why hasn't blurred reflections been
> implemented in the official POV? Just curious...
> 
> Regards,
> Tim
> 
>

Post a reply to this message