POV-Ray : Newsgroups : povray.beta-test : Sven’s Bug : Re:_Sven’s_Bug Server Time
5 May 2024 18:10:47 EDT (-0400)
  Re:_Sven’s_Bug  
From: Alain
Date: 18 Feb 2016 17:55:44
Message: <56c64bf0$1@news.povray.org>
Le 16-02-18 15:38, clipka a écrit :
> Am 18.02.2016 um 20:29 schrieb Stephen:
>> I’ve named this Sven’s Bug because I noticed it running his code posted
>> in “15 Shapes Taking Forever” in povray.general
>> And so he gets his name on something. :-P
>>
>> There is a difference between renders with and without using Render
>> Block Size.
>> The images are with the windows QUICKRES.INI [800x600, AA 0.3] with and
>> without +BS8
>> I compared them in PaintShop Pro (XOR shows it best)
>
> Nothing in there that would surprise me.
>
> I see some tiny fluctuations in the results, which are normal -- and a
> few more pronounced differences at the objects' edges, which are quite
> obviously related to anti-aliasing.
>
> POV-Ray 3.7 performs anti-aliasing at the render block level, /before/
> assembling the blocks to an entire image. As a consequence, the decision
> to supersample any given pixel not only depends on the neighboring
> pixels, but also on whether the pixel is at a block boundary.
>
> Consequently, reducing the render block size, and therefore adding more
> block boundaries, will change the behaviour of anti-aliasing, and
> increase the risk of aliasing artifacts.
>
> So this is not a bug, just a drawback of how things are implemented.
>
>
> (Hey, what do you expect if you associate Sven's name with it?! :P)
>

Is it possible that using antialiasing method 2 could reduce that kind 
of artefacts?
With method 1, you sample the center of each pixel, and compare that 
with the neibors. With method 2, you start with samples at each corners 
that are then compared.


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.