I assume that this is a known (and previously reported) bug, but mpov focal
blur doesn't work when the scene includes a hollow, transparent object (the
scene viewed through the object is not blurred). I'm using 3.1g.wmp.0.5a
under win95.

// sample code to show lack of blurring through object
#version unofficial MegaPov 0.5;
global_settings{post_process{focal_blur{1,2,6,0}}}
camera{location <0,0,-10> look_at 0}
light_source{10*y color rgb<1,1,1>}
sphere{0,2 pigment{rgbf <0.75,0.75,1,1>} hollow on}
plane{-z, -50 pigment{checker rgb 0, rgb 1 scale 2}}

Post a reply to this message

Tom Melly <tom### [at] tomandlucouk> wrote:
: I assume that this is a known (and previously reported) bug, but mpov focal
: blur doesn't work when the scene includes a hollow, transparent object

  It's not a bug. It's only a limitation that may be too difficult to
overcome.
  The post-process focal blur works only at recursion level 1 (the
"raycasting" process). It will not work for reflection, refractions or any
other higher recursion operations.

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):_;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message

"Warp" <war### [at] tagpovrayorg> wrote in message
news:39abe430@news.povray.org...
>   It's not a bug. It's only a limitation that may be too difficult to
> overcome.

Ah, what the hell - let's go the MS route and call it a "feature".

Post a reply to this message

Tom Melly <tom### [at] tomandlucouk> wrote:
: Ah, what the hell - let's go the MS route and call it a "feature".

  It's not that the limitation is there because people are too lazy to fix it.
You have to understand how the post-processing works in order to understand
why there is this kind of limitation and why it's just too difficult to
overcome.

  If you want working focal blur, use povray's own. It's slower, but it works
in all cases (and that's why it is slower, in fact).
  (If you want non-grainy focal blur, use 'variance 0'.)

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):_;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message

"Warp" <war### [at] tagpovrayorg> wrote in message
news:39ad1856@news.povray.org...
> Tom Melly <tom### [at] tomandlucouk> wrote:
> : Ah, what the hell - let's go the MS route and call it a "feature".
>
>   It's not that the limitation is there because people are too lazy to fix
it.
> You have to understand how the post-processing works in order to
understand
> why there is this kind of limitation and why it's just too difficult to
> overcome.
>

Oh, I wasn't disagreeing - in fact once you see it happen, even without
knowing the inner workings of post-processing, you slap your forhead and say
"of course" (and then thank the Gods, or Nathan, that the non-PP image is
kept as well, and then wonder how you are going to fill in all the detail
you missed out on the basis that focal blur would hide your sins).

Post a reply to this message

Tom Melly <tom### [at] tomandlucouk> wrote:
: Oh, I wasn't disagreeing

  Sorry, I misunderstood your response. I thought it was sarcastic :)

: (and then thank the Gods, or Nathan, that the non-PP image is
: kept as well, and then wonder how you are going to fill in all the detail
: you missed out on the basis that focal blur would hide your sins).

  Hmm... that's a new point of view about the usage of focal blur: You can
use it to hide all the tiny errors and mistakes in your scene :)

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):_;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message

>   (If you want non-grainy focal blur, use 'variance 0'.)

	Unfortunately 'variance 0' seems to be not enough to avoid
the grainy looking, especially along edges of some blurred object.
It can be a little bit better when using antialiasing method 2
(need unofficial version) while rendering focal blur. The CPU price
to pay is of course even bigger...
	By the way I always wondered why we get this grainy result
(again with variance 0 and high number of samples, e.g. 100). It's
never perfect. Also media and area-lights have basically the same
behaviour. Could be a problem with random number generator ? Code
in POV-Ray for that seems to be good enough... Can't understand
what's happening exactly.
	How do other (commercial) raytracers achieve so "pure"
effects ???? Maybe it's no more raytracing in this case...


*** Nicolas Calimet
*** http://pov4grasp.free.fr

Post a reply to this message

"Warp" <war### [at] tagpovrayorg> wrote in message
news:39ad1d85@news.povray.org...
> Tom Melly <tom### [at] tomandlucouk> wrote:
> : Oh, I wasn't disagreeing
>
>   Sorry, I misunderstood your response. I thought it was sarcastic :)
>

Nah, ironic. My sarcastic days are mon, tue, fri and alternative sats.

Post a reply to this message

Nicolas Calimet <pov### [at] freefr> wrote:
: 	Unfortunately 'variance 0' seems to be not enough to avoid
: the grainy looking, especially along edges of some blurred object.

  Then use a higher confidence.

  However, what I have tested, variance 0 gives almost always very smooth
focal blur, no matter what confidence. I don't know if confidence has any
effect if variance is 0.

-- 
main(i,_){for(_?--i,main(i+2,"FhhQHFIJD|FQTITFN]zRFHhhTBFHhhTBFysdB"[i]
):_;i&&_>1;printf("%s",_-70?_&1?"[]":" ":(_=0,"\n")),_/=2);} /*- Warp -*/

Post a reply to this message

>   Then use a higher confidence.

	Nope !

>   However, what I have tested, variance 0 gives almost always very smooth
> focal blur, no matter what confidence. I don't know if confidence has any
> effect if variance is 0.

	Yeahp ;-) Formula is a little bit complicated (at least to me)
but AFAI understand the threshold will be 0 as soon as variance is.
Also when variance is low enough but not 0, only very low confidence
will really affect the rendering (i.e. be worse).
	I noticed your "almost always" which translates for me to
"it's never perfect" in my previous post :-}


*** Nicolas Calimet
*** http://pov4grasp.free.fr

Post a reply to this message