POV-Ray : Newsgroups : povray.binaries.images : MIKA Pink Floyd's next album cover? - 2 attachments : Re: MIKA Pink Floyd's next album cover? - 2 attachments Server Time
14 Aug 2024 22:16:16 EDT (-0400)
  Re: MIKA Pink Floyd's next album cover? - 2 attachments  
From: Safari
Date: 3 Oct 2002 20:27:01
Message: <slrnappnuk.4b5.y7pt9001@safari.homelinux.net>
On Mon, 30 Sep 2002 21:52:23 -0400,
Christopher James Huff <chr### [at] maccom> wrote:
> In article <Xns### [at] 204213191226>,
>  Marc Champagne <nos### [at] pleasecom> wrote:
> 
>> I'm feeling pretty damned proud of how these have turned out !
>> 
>> These are quick renders without AA
> 
> Very nice...
> The top one seems to have very distinct bands of colors, maybe 
> increasing the dispersion samples would help. The second one could use a 
> little higher number of photons/lower spacing, but both are simple and 

I used the same photon count as in the original .pov (10000000)
and dispersion_samples (30) ... 

how noticeable are the 'jerky' edges on that image in your opinion --
in the middle, around the white beam and between 'the rainbow colors'?
I think they're due to the PRNG used in Povray 3.5, OR,
due to the fact that FRAND can return only(?) 32768 different values.

whichever the reason, if someone could sacrifice a minute in looking
and comparing these two pictures...

http://iki.fi/safari/prng/pink_1280_orig.jpg
http://iki.fi/safari/prng/pink_1280_sha1.jpg
(also .png's, 1.5 MB each)
mirror in case of trouble
http://kotisivu.mtv3.fi/sami.farin/pink_1280_orig.jpg
http://kotisivu.mtv3.fi/sami.farin/pink_1280_sha1.jpg
(NO .png)

...and tell me what causes those 'jerky' edges?

oh, those _sha1.* files are made with a little patch to
Povray 3.5, which uses SHA-1 as the PRNG:
http://iki.fi/safari/prng/povray-3.5-sha1-prng-and-stuff.diff.txt
that patch has also GSL patch (which probably nobody wants)
and some misc bug fixes I found on the 'net.
only SHA-1 part of it:
http://iki.fi/safari/prng/povray-3.5-sha1-prng.diff.txt

BTW, one call to get a new pseudo-random number takes about 180
cycles on a Celeron.
I didn't optimize for speed any places of this SHA-1 usage,
like cells_pattern().

why are there calls to rand() in some places?

> beautiful. (of course, I've always been fascinated with optics)

too bad I ran almost out of mem with those settings..
 
> BTW, what is the MIKA in your message subjects for?

-- 
Safari - y7p### [at] sneakemailcomgovinvalid
"Talk is cheap. Show me the code." - Linus Torvalds


Post a reply to this message

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