Nothing new, but:

In FAQ at 
http://tag.povray.org/povQandT/languageQandT.html#blurredreflection it 
is suggested that a good blurred reflection can be achieved using 
averaged textures. That is true, but they can also be achieved a lot 
faster using normals, for example:

finish {reflection 0.6}
normal {
    bumps 0.16    // This obviously controls the amount of blur
    scale 0.00001 // This must be small enough compared to scene scale
}

The key issue is to use good antialiasing with enough depth. Without AA 
it looks very, very bad no matter what values given above. With AA 
method 2 and depth=4 it starts to look very good. I think this should be 
added to FAQ?

Severi S.

Post a reply to this message

Severi Salminen wrote:
> The key issue is to use good antialiasing with enough depth. Without AA 
> it looks very, very bad no matter what values given above. With AA 
> method 2 and depth=4 it starts to look very good. I think this should be 
> added to FAQ?

How about everaging normals?  And using focal blur to round out the
whole effect.  It's what I'm using for my current project's lineoleum
texture... XD

-- 
Tim Cook
http://home.bellsouth.net/p/PWP-empyrean

-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GFA dpu- s: a?-- C++(++++) U P? L E--- W++(+++)>$
N++ o? K- w(+) O? M-(--) V? PS+(+++) PE(--) Y(--)
PGP-(--) t* 5++>+++++ X+ R* tv+ b++(+++) DI
D++(---) G(++) e*>++ h+ !r--- !y--
------END GEEK CODE BLOCK------

Post a reply to this message

In article <3ff89eaa$1@news.povray.org>,
 Severi Salminen <sev### [at] NOT_THISsibafi> wrote:

> In FAQ at 
> http://tag.povray.org/povQandT/languageQandT.html#blurredreflection it 
> is suggested that a good blurred reflection can be achieved using 
> averaged textures. That is true, but they can also be achieved a lot 
> faster using normals, for example:

This method tends to produce grainy results, and can be quite slow, 
which is why the average texture method was developed. The average 
texture method generally produces smoother results with fewer samples, 
and doesn't require that the entire image be rendered with high 
antialiasing settings.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: <chr### [at] tagpovrayorg>
http://tag.povray.org/

Post a reply to this message

In article <3ff8c0c9$1@news.povray.org>,
 Tim Cook <z99### [at] bellsouthnet> wrote:

> How about everaging normals?  And using focal blur to round out the
> whole effect.  It's what I'm using for my current project's lineoleum
> texture... XD

Averaging normals gives a very different effect. The result of the 
averaged normal is a single perturbed normal, which is then used for the 
texturing and lighting calculations. The result is not blurred at all.

An average texture computes the texture for each normal and averages the 
final colors, the antialiasing method just uses normals to generate fine 
detail which is averaged by high antialiasing settings.

-- 
Christopher James Huff <cja### [at] earthlinknet>
http://home.earthlink.net/~cjameshuff/
POV-Ray TAG: <chr### [at] tagpovrayorg>
http://tag.povray.org/

Post a reply to this message

Severi Salminen <sev### [at] not_thissibafi> wrote:
> In FAQ at 
> http://tag.povray.org/povQandT/languageQandT.html#blurredreflection it 
> is suggested that a good blurred reflection can be achieved using 
> averaged textures. That is true, but they can also be achieved a lot 
> faster using normals, for example:

  The blur trick uses normals as well, but it has a couple of advantages
compared to your suggested method:

  - Your suggestion requires quite heavy antialiasing to look good.
    This is a nuisance for several reasons. It makes rendering slower
    than necessary, and not only when rendering the object in question
    (your suggestion is *not* faster than the averaged textures trick due
    to this) but in the whole scene: Everything else will be antialiased
    heavily as well (which sometimes is good, but more often than not it
    will just be a waste of time since a lighter antialiasing often gives
    a result which is good enough).
      The trick works even without antialiasing (and antialiasing only
    makes it look better) and achieves the same result (because it also
    is based on normals) or even better.

  - Your suggestion produces a grainy blur. Sometimes this is the desired
    effect, but such grainy blur is also possible with the trick (but
    without the need for heavy antialiasing).
      The good thing is, however, that also smooth blur is possible (which
    is achieved by making all the normals go to the same directions in all
    points, which is achieved by scaling the normals large instead of small).

  - It's easy to control how many reflected rays are sent to calculate the
    blurred reflection with the trick. With your suggestion you don't have
    practically any control on this. (The only way to be sure that a certain
    number of reflected rays is always calculated would be to force
    antialiasing on each pixel, but that would be quite extreme since it
    would be done on the whole image and not only on the part where the
    reflecting object is located.)

-- 
plane{-x+y,-1pigment{bozo color_map{[0rgb x][1rgb x+y]}turbulence 1}}
sphere{0,2pigment{rgbt 1}interior{media{emission 1density{spherical
density_map{[0rgb 0][.5rgb<1,.5>][1rgb 1]}turbulence.9}}}scale
<1,1,3>hollow}text{ttf"timrom""Warp".1,0translate<-1,-.1,2>}//  - Warp -

Post a reply to this message

Warp wrote:

>   The blur trick uses normals as well, but it has a couple of advantages
> compared to your suggested method:
> 
>   - Your suggestion requires quite heavy antialiasing to look good.

Yes, I thought this to be non-issue as I want to use quite heavy AA 
normally. But for test renderings it will be an issue. And you are 
correct that the heavy AA makes this otherwise fast method not so fast 
afterall :(

>   - Your suggestion produces a grainy blur. 

It will be _totally_ smooth but only with heavy AA. Now I'm doing a test 
rendering with "+a0.0 +am2 +r5" and yes, it is very smooth :) And 
slooooowww...

>   - It's easy to control how many reflected rays are sent to calculate the
>     blurred reflection with the trick. 

Also true. I think it I like(d) this single bumps method more because of 
the simpler looking syntax. Averaging textures looks uglier in the scene 
file and with "my" method one only has to alter one value to alter the 
smoothness. And if there was a way to apply object related AA settings 
it would be maybe more usable. If I can propose a new feature, this 
would be it: object based AA. That way one could define high AA to very 
small detailed objects without affecting the whole scene.

The single bumps method also works well if one does not want to make 
_very_ reflective surface. I'm now constructing a decent parquet and the 
  simpler method looks good even without very heavy AA. It all depends 
on the case. But most likely the averaging method is/can be allways 
faster with more control. I have to try that one too.

Severi S.

Post a reply to this message

There is a clear tendency that using one normal scaled very small with good
AA, is much faster than averaging a lot of textures. Sure, in theory, the
opposite is true. The micro-normal method isn't efficent because of the AA
that covers the entire scene, but nevertheless, in all my scenes it IS the
fastest method - by far!

The result is a little grainy, assuming you use only +r2  (which is 25
samples pr pixel)  but this may not bother anyone. By using the other method
(averaging textures) there is a price to pay as well: It can produce very
inaccurate results, at least when the normals have random directions and you
don't average all 255 possible textures.

Severi:  Use the method you think is best, and fastest.

Try this pov file that illustrates what I mean:
http://news.povray.org/povray.binaries.scene-files/29562/

Regards,
Hugo

Post a reply to this message

"Severi Salminen" <sev### [at] NOT_THISsibafi> wrote in message
news:3ff95bd4$1@news.povray.org...

> If I can propose a new feature, this
> would be it: object based AA. That way one could define high AA to
very
> small detailed objects without affecting the whole scene.

   Now wouldn't that be good! Nice idea.

     Steve


> Severi S.

Post a reply to this message

>  If I can propose a new feature, this 
> would be it: object based AA. That way one could define high AA to very 
> small detailed objects without affecting the whole scene.

The word I was looking for was of course "specific": object specific AA.

Severi

Post a reply to this message

Severi Salminen wrote:

> Warp wrote:
> 
>>   The blur trick uses normals as well, but it has a couple of advantages
>> compared to your suggested method:
>>
>>   - Your suggestion requires quite heavy antialiasing to look good.
> 
> 
> Yes, I thought this to be non-issue as I want to use quite heavy AA 
> normally. But for test renderings it will be an issue. And you are 
> correct that the heavy AA makes this otherwise fast method not so fast 
> afterall :(

If I may make an addition. If one DOES use heavy AA usually then the 
single bumps method produces, of course, a lot faster reflections. Using 
AA and many averaged textures slows thing down _a lot_! This depends 
also the amount of reflecting surfaces in final image: the more area 
there is, the faster the single bumps method will be. And I still think 
the averaging and creating randomly moved normals looks and feels very 
ugly in scene sources.

Newerthless, I strongly suggest that _both_ methods should be covered in 
Pov-FAQ and TAG-Pov-FAQ as both methods have distinct advantages and 
disadvantages depending on the actual situation. Would this be possible, 
Warp?

Severi S.

Post a reply to this message