The perceivable beauty of algorithms
Absract:
Most computer systems that utilize perceptual models are invisible in the sense that the work they do is aimed at e.g. compressing the data with minimal perceivable differences for humans. ..
Introduction:
Several advanced algorithms have been and are actively developed for processing images and sound [ref]?. Often the purpose is to e.g. compress the data with minimal perceivable changes. To this end, psychological (and cognitional?) models of perception are applied. So some of these frameworks comprehend complex theories of e.g. human seeing or hearing (e.g. http://foulard.ee.cornell.edu/marcia/psych.html ). Besides data compression, another purpose for media processing algorithms is restauration, where e.g. small bitmap images are scaled up and processed to look less blocky, or unwanted objects are automatically removed from images (see below) or sound (in which noise removal is a common operation). Also there the goal is that the work done by the software should be invisible/-audible to humans. Furthermore, when rendering 3d scenes the target is often so-called photorealism, so also there much work has been done to both simulate reality accurately and mimic it non-accerately but so that believable images for the human eye can be made.
The purpose of the work here is reverse: to make those usually invisible algorithms visible, or more generally speaking, perceivable. A motivation is curiosity --- simply the want to see if it can be done, and if yes, what they will look or sound like. Additionally, intuition and preliminary observations suggest that there may be sudden beauty lurking in them --- after all, they do implement relatively complex perceptual models which might enable interesting outcomes. So the question is: how to make those different kinds of algorithms perceivable, or how to use them to produce interesting outcomes?
Different strategies outlined and experimented to get some demonstrations (a gallery).
First a review of related research to a) get an overview of what algorithms there are to try this with, and b) how would it be possible to work with them towards the purpose here.
Related research: media processing algorithms and their use
a) get an overview of what algorithms there are to try this with, and b) how would it be possible to work with them towards the purpose here.
Bringing out the artifacts made by compression algorithms
(tmu told that Shynola has used this in some of their videos)
..
Making renderings of impossible worlds to reveal the complexity under photorealism?
..
.. and/or ?
..
Discussion
Conclusions
Acknowledgements:
Acknowledgements to Joel Ryan of STEIM for pointing out how the logic of computer science is often backwards from an artistic point of view, hence laying out the basic idea of thinking about ways of reversing it. The title is a reference to the classic, the Algorithmic Beauty of Plants .
http://www-sop.inria.fr/odyssee/research/tschumperle-deriche:02d/appliu/index.html :ssa suurennoksesta http://www-sop.inria.fr/odyssee/research/tschumperle-deriche:02d/appliu/img/totoro_bloc.jpg lopputulos http://www-sop.inria.fr/odyssee/research/tschumperle-deriche:02d/appliu/img/totoro_shocked.jpg
Non photorealistic rendering has of course been a strong trend in recent years, see e.g. http://www.algorithmic.com/ - how to relate to this?
here's yet another example of an interesting
hidden algo:
11:25 < hexa> Kunii oli kehittänyt ryhmänsä kanssa sumeiksi kutsumiaan
algoritmeja, jotka suttasivat noita liian siistejä viivoja
11:25 < hexa> [NURBS]it oli liian tehottomia
11:26 < hexa> ja toi nervousness jotenkin soitti kelloa
11:26 < antont> hm ai siis halusivat ihmisen piirtämän näköisiä vai miksi piti
sotkea?
11:26 < hexa> ja ne esimerkit mitä se esitti oli aivan upeita
11:26 < hexa> joo hiukset
11:26 < hexa> yms.
11:27 < hexa> ei piirtämän vaan luonnollisen näköinen
11:27 < hexa> ne pelitti hirveen hyvin valojen ja muun semmosen kanssakin
11:27 < antont> joo, hämäännyin kun puhuit viivoista
11:27 < hexa> algoritmit siis
11:28 < hexa> viivoihin se polygonijoukkokin redusoituu
11:29 < hexa> ellen ole aivan väärin ymmärtänyt :)
11:29 < antont> ensin tuli mieleen non-photorealistic rendering, kun siellä on
tehty kaikkia jänniä missä matkitaan mm. eri taidesuuntauksia,
tunnetuimpana varmaan comic rendering, mut tuo orgaanisen(?)
näköiseksi tekeminen onkin hassusti taas photorealismin ts.
luonnollisuuden tavoittelua
Subject: NurbS
-
Ihmisiä rannalla jpg 7 prosentin tallennus ala irfanview - huomatkaa abstraktit perusmuodot, ja niiden yhdistyminen hahmotustasolla:
a source for sources --antont, Tue, 22 Mar 2005 20:09:47 +0200 reply
http://www.faqs.org/faqs/graphics/algorithms-faq/
ABugInOgre --antont, Thu, 26 May 2005 14:23:02 +0300 reply
and discussion about it seemed like an example of this
Lenna --antont, Fri, 30 Dec 2005 01:48:40 +0200 reply
or Lena (if not Leena, here in Finland .. she's a swede tho), http://www.ee.cityu.edu.hk/~lmpo/lenna/Lenna97.html quite a story
DebuggingWithColors? -- Wed, 08 Feb 2006 14:27:21 +0200 reply
http://www.cessen.com/storage/misc_files/shader_debug1.png is an example how colours can be used in debugging if shading (/ rendering) code,
< cessen> green channel = light hitting the surface if we ignore shadows,
< cessen> blue channel = do we even get to the shadow pass at all
< cessen> red channel = shadow factor
MotionEstimation? --antont, Wed, 08 Feb 2006 19:47:47 +0200 reply
harkyman: "in one version of the DivX? encoder, it pops up a status window that can be tasked to show motion vectors overlaying your video, only changed blocks, etc. Kind of interesting."