POV-Ray : Newsgroups : povray.bugreports : Media Bug Server Time
31 Oct 2024 19:27:49 EDT (-0400)
  Media Bug (Message 1 to 2 of 2)  
From: Ken
Subject: Media Bug
Date: 4 Mar 1999 03:39:28
Message: <36DE4634.312579ED@pacbell.net>
Versions affected:
   Pov-Ray v3.1.watcom.win32.r1 (Pentium II optimized)
   Pov-Ray v3.1d.msvc.win32     (Pentium II optimized)

   See attached file for evidence of problem.

   Discription - Observations:
   If renderd as is a visible horizontal defect is seen running through
   each object. The camera position was choosen to get the best angle so
   that the defect was seen in all 6 objects at once. If the y camera
   direction is changed by even a small +/- amount, the position or the
   appearance of the defect changes accordingly.

   There are two different density patterns available for comparison but
   the defect manifests it self equaly in either type used. As such it
   appears that it is not a pattern dependent defect.

   It renders equaly the same in both Pov-Ray versions listed above.

   Verification:
   Before submission this file was sent to Dan Connelly who verified
   that he too observed the problem. It was his opinion, as well as my
   own, that there was a likelyhood that it is a bug and should be
   reported.

   Possible causes are :
   1. Arises from using a lathe object for a media container
   2. Media intereaction - see 3
   3. Proximity of media density patterns to each other
   4. Proximity of lathe objects to each other
   5. Lathe type used - some testing was done using a quadratic spline
      and the defect remained - As the quadratic spline type greatly
      increased the render time it was ommitted from this example to
      increase the speed for testing of the defect.

  Solutions: ?

  My system: Pentium 200 mmx, 512k cache, 128m edo ram, on Win98 platform.

-- 
Ken Tyler

mailto://tylereng@pacbell.net


Post a reply to this message


Attachments:
Download 'media_bug_report.pov.txt' (2 KB)

From: Thomas Willhalm
Subject: Re: Media Bug
Date: 8 Mar 1999 04:51:40
Message: <qqmpv6kmhqo.fsf@deutwang.fmi.uni-konstanz.de>
Ken <tyl### [at] pacbellnet> writes:
> Versions affected:
>    Pov-Ray v3.1.watcom.win32.r1 (Pentium II optimized)
>    Pov-Ray v3.1d.msvc.win32     (Pentium II optimized)

Add the generic Unix version 3.0 and the inofficial linux version 3.1
to your list.

>    See attached file for evidence of problem.
> 
>    Discription - Observations:
>    If renderd as is a visible horizontal defect is seen running through
>    each object. The camera position was choosen to get the best angle so
>    that the defect was seen in all 6 objects at once. If the y camera
>    direction is changed by even a small +/- amount, the position or the
>    appearance of the defect changes accordingly.
> 
>    There are two different density patterns available for comparison but
>    the defect manifests it self equaly in either type used. As such it
>    appears that it is not a pattern dependent defect.
> 
>    It renders equaly the same in both Pov-Ray versions listed above.
> 
>    Verification:
>    Before submission this file was sent to Dan Connelly who verified
>    that he too observed the problem. It was his opinion, as well as my
>    own, that there was a likelyhood that it is a bug and should be
>    reported.
> 
>    Possible causes are :
>    1. Arises from using a lathe object for a media container
>    2. Media intereaction - see 3
>    3. Proximity of media density patterns to each other
>    4. Proximity of lathe objects to each other
>    5. Lathe type used - some testing was done using a quadratic spline
>       and the defect remained - As the quadratic spline type greatly
>       increased the render time it was ommitted from this example to
>       increase the speed for testing of the defect.

This bug has nothing to do with media. This if evident when you remove
the media from your object and add a colour to it. (I have attached the
modified file to this post - without the candle sticks.)

This means, that you rediscovered the well known "lathe bug". I appears
when you look more or less orthogonal to a lathe object.

Jochen Lippert (dar### [at] uni-bremende) had a similar problem, when he
implemented his "sphere sweeps patch". As far as I know, he solved
the problem in his case.
(http://www.informatik.uni-bremen.de/~vader/pss/pss.html)
He wrote me that he wanted to look at the lathe code to see whether
this bug can be removed with similar techniques. Probably, he didn't
find the time to do so, because I never heard about a fix for it.

Thomas

---- POV code follows

#declare Candle =
union{
 lathe{
  linear_spline 27,
    <0.000000,0.000000>,<0.072407,0.030038>,<0.084233,0.039200>,
    <0.094412,0.050470>,<0.102887,0.063629>,<0.109666,0.078300>,
    <0.114786,0.094229>,<0.118402,0.111670>,<0.120700,0.131000>,
    <0.121833,0.152533>,<0.121833,0.176333>,<0.120700,0.202400>,
    <0.118244,0.231237>,<0.113522,0.265363>,<0.105400,0.307800>,
    <0.093248,0.360060>,<0.078451,0.417617>,<0.062900,0.474433>,
    <0.048229,0.525421>,<0.035070,0.569290>,<0.023800,0.605700>,
    <0.014733,0.634413>,<0.007933,0.655608>,<0.003400,0.669566>,
    <0.001007,0.676908>,<0.000125,0.679613>,<0.000000,0.679999>
   texture{ pigment{rgb <1,1,0> } }
   scale<1.25,2.0,1.25>*3 
   translate y*1.25
 }
}

 camera{location<0,3.8,-20> look_at 0}

 light_source{<0,0,-500>rgb 1}

#declare Set = 
  union{
    #declare A = 0;
    #while  (A < 3)
     object{Candle
       translate<-1.5,-0.75,0.0>
       translate<A*1.25,A,0>
     }
     #declare A = A+1;
    #end
  }

object{Set rotate   0*y translate< 3,1,0>}
object{Set rotate 180*y translate<-3,1,0>}


-- 
http://www.fmi.uni-konstanz.de/~willhalm


Post a reply to this message

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