While tracing images and doing some test runs
for the IRTC-Still, I've noticed that using
focal-blur with 20 samples takes far longer
than Antialiasing method 2 with depth set 2
(which may use 25 samples). I haven't run
with depth set to 3 (since on thumbnail-tests
thats not worth it), but it still is a huge amount
of time in comparison.
I guess it has to do with the algorithm, since
focal-blur probably uses all samples, whereas
the antialiasing only kicks in when needed.

Any suggestions on how to improve focal-blur?
 A major trace is running for one and a half days
now, and since RL will get in the way, I'll probably
have to jump to antialiasing anyway. But I thought
a nice discussion could clear some things up for me.

Thanks in advance!

--
Tim Nikias
Homepage: http://www.digitaltwilight.de/no_lights/index.html

Post a reply to this message

Antialiasing and focal blur are very different. If you use focal blur and no
antialiasing you'll get 'pixel grainy' effects or some rough, noisy blurs
:-)
No vocal blur and high antialiasing leads to very sharp, almost perfect
edges.

Post a reply to this message

You have to take into account that there's one big difference between
focal blur and antialiasing:
  Antialiasing is triggered when the color of *adjacent* pixels differ more
than a threshold value. There are many places in the image where only one
ray is shot for the current pixel and that's it.
  However, focal blur shoots many rays for each pixel (I don't know exactly
how the algorithm works) and then (I think) shoots even more rays if the
difference between the previous ones was large enough.
  Also the "samples" taken by the initial rays are dispersed a lot more than
antialiasing pixels (focal blur can extend over an area equivalent to tens of
pixels) which means that there will be a lot more variation and a lot more
chances for the samples to differ over the threshold.

-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message

Apache <apa### [at] yahoocom> wrote:
: If you use focal blur and no
: antialiasing you'll get 'pixel grainy' effects or some rough, noisy blurs

  Not necessarily. For example, see
http://www.students.tut.fi/~warp/pics/GlowTest.jpg

-- 
#macro M(A,N,D,L)plane{-z,-9pigment{mandel L*9translate N color_map{[0rgb x]
[1rgb 9]}scale<D,D*3D>*1e3}rotate y*A*8}#end M(-3<1.206434.28623>70,7)M(
-1<.7438.1795>1,20)M(1<.77595.13699>30,20)M(3<.75923.07145>80,99)// - Warp -

Post a reply to this message

I like that picture, but in my opinion focal blur has a totally different
objective than pixel-oriented antialiasing has...

BTW, I'm having problems typing that word 'antialiasing' all the time. I
keep typing things like 'anitalisaing' and things like that. Maybe we should
change that word into something else?  :-)   Maybe I'm too tired...

Post a reply to this message

Thanks! I agree, it looks like focal blur isn't useful
for antialiasing AND little focal depth, but using
a lot of blurring can enhance images (if you wanna
see one, look in my gallery for the chess image, I
think the first on the page). But the image I was
tracing for the IRTC-Stills round now took
2 days, and it reached the difficult part about a day
ago and has traced about 40 lines since.

Since I don't have all February for my entry -
we all know RL - I'll stick to antialiasing for
the entry. All I wanted was a slightly blurred moon
and clouds in the background, but I think, the image
doesn't really need it. It can go without that.

--
Tim Nikias
Homepage: http://www.digitaltwilight.de/no_lights/index.html

Post a reply to this message

Apache wrote:

> No vocal blur and high antialiasing leads to very sharp, almost perfect
> edges.

Hey, "vocal" blur? I thought we were talking about images,
not wave-files... ;)

But I do agree, too few samples get a grainy blur-look, at least
when used with sharp contrasts.

--
Tim Nikias
Homepage: http://www.digitaltwilight.de/no_lights/index.html

Post a reply to this message

On Thu, 7 Feb 2002 22:35:21 +0100, "Apache" <apa### [at] yahoocom> wrote:
> Antialiasing and focal blur are very different. If you use focal blur and no
> antialiasing

Note: you can't use it together with 3.5

> you'll get 'pixel grainy' effects or some rough, noisy blurs
> :-)

Have you seen http://www.wozzeck.net/images/DoF.html ?

> No vocal blur and high antialiasing leads to very sharp, almost perfect
> edges.

Unfortunatelly reality is not perfect :-)

ABX

Post a reply to this message

Tim Nikias <tim### [at] gmxde> wrote:
: But I do agree, too few samples get a grainy blur-look, at least
: when used with sharp contrasts.

  The main parameter affecting graininess is variance. AFAIK its default
value was lowered (from 1/128 to 1/10000) in order to get a much better result
with the defaults (previously it didn't matter how many samples you defined,
20 or 2000, you always got a grainy result due to the default value of
variance).
  Then, of course, the number of samples and the aperture are things that
affect this (the higher the aperture, the more samples you'll need).

-- 
#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

> > If you use focal blur and no antialiasing
> > you'll get 'pixel grainy' effects or some rough, noisy blurs
>
> Note: you can't use it together with 3.5

Correct. There is no sense in using both AA and focal_blur at the same time!
Btw, I found focal blur to be quite efficient and fast. I doubt I'll be
using AA a lot, in the future.. Focalblur is much better.

> keep typing things like 'anitalisaing'

You have girls on your mind, that's what's going on. :o)  Anita... Lisa...

Hugo

Post a reply to this message