W3C home > Mailing lists > Public > www-svg@w3.org > November 2013

RE: Gabor noise [was RE: Perlin and simplex noise]

From: David Dailey <ddailey@zoominternet.net>
Date: Tue, 12 Nov 2013 22:29:26 -0500
To: "'Michael Mullany'" <michael@sencha.com>, "'Erik Dahlstrom'" <ed@opera.com>
Cc: "'www-svg'" <www-svg@w3.org>, <public-fx@w3.org>
Message-ID: <004601cee020$8e61ea40$ab25bec0$@net>
Hi  Michael,

 

Cool!

 

It is a very nice review article done by people who know a lot about the
subject, which I don’t.  I know there are folks on the WG who are pretty
knowledgeable, though. I guess the thing to do is to convince folks that it
is worth the time and energy needed to research the subject and spec things
out.  Input from the 3D/GPU community might be helpful too since the GPU
folks have possibly already implemented some of these methods in hardware.
I’d be willing to invest a bit of time making inquiry about the hardware
implementation issue, if there were sufficient interest. Of course, it’s
entirely possible that my thoughts are na´ve and that 3D/GPU implementations
might be only relevant to the 3D world and have nothing to do with SVG, but
a cursory glance at the article you mentioned suggests that there might be a
good degree of overlap.

 

One issue that comes to mind as I think about it is the following: the folks
at Pixar who developed all the wonderful 3D stuff for Toy Story and beyond,
were a bit dismayed to find that they had made obsolete a class of highly
professional and talented 2D illustrators at Disney.  Sometimes, the 2D
“simulation” of 3D is actually preferable, from an artistic, cognitive and
educational perspective to the full-blown 3D approach to the same topic. Not
that the SVG WG needs to be reminded of this, of course, but I can imagine
readers of the list who might forget the value of effective planar
communication on things that are generally 2D display devices. Is using 3D
noise to make a tornado, really better, for example, than displaying the
vortices of vectors that comprise the physics of the tornado? One looks more
realistic (in a European Renaissance – post development of the lens –
perceptual sort of sense), the other might BE more realistic (in a gongbi /
cognitive sort of sense).

 

Cheers

David

 

From: Michael Mullany [mailto:michael@sencha.com] 
Sent: Tuesday, November 12, 2013 6:37 PM
To: David Dailey; Erik Dahlstrom
Cc: www-svg; public-fx@w3.org
Subject: Re: Gabor noise [was RE: Perlin and simplex noise]

 

I am not a graphics engineer, so this may be review for folks on this list,
but I found this 2010 survey of procedural noise generation algorithms from
to be helpful in understanding state of the art and trade offs among various
noise functions.

 

State of the Art in Procedural Noise Functions

A. Lagae1,2 S. Lefebvre2,3 R. Cook4 T. DeRose4 G. Drettakis2 D.S. Ebert5
J.P. Lewis6 K. Perlin7 M. Zwicker8

1Katholieke Universiteit Leuven 2REVES/INRIA Sophia-Antipolis 3ALICE/INRIA
Nancy Grand-Est / Loria

4Pixar Animation Studios 5Purdue University 6Weta Digital 7New York
University 8University of Bern

 

http://www-sop.inria.fr/reves/Basilic/2010/LLCDDELPZ10/LLCDDELPZ10STARPNF.pd
f

 

On Tue, Nov 12, 2013 at 3:48 AM, David Dailey <ddailey@zoominternet.net>
wrote:

Hi folks,

Nikos A., from Canon, gave a recent and very well-documented talk at The
Graphical Web (some of you may remember it as SVG Open?) on diffusion curves
[1]. I found the rationale for their inclusion in a future spec quite
compelling, btw, but that's a different issue.

He also mentioned something quite cool: Gabor noise (related to Gabor
filters). [2]

I'm not sure of the computational complexity of it relative to Perlin noise,
but at the end of the paper [2] one can see a large variety of very
interesting textural effects that are simply not expressible with
feTurbulence at present, short of, for example, overlaying two or more
feTurbulences with different frequencies, octaves and directionality, and
thence using an feDisplacement to combine them. Likely, this strategy would
be computationally more intense than unpacking the rather dense-looking math
of Gabor filters.

But Gabor noise appears to be scalable, like feTurbulence, and, most
importantly, it adds a rich set of textures that would enable more
naturalistic approximations to the sorts of things humans might want to
draw.  Combine Diffusion curves with Gabor noise, and I suspect you'd go a
very long way toward enabling very economical approximations to many real
world images (hence, phenomenal image compression) and while image analysis
is not, per se, a part of the SVG WG's purview, it is, nonetheless, the
ancestor of zillions of use cases for SVG. And, of course, there is always
the ability to draw pretty pictures with it!

Regards
David

[1] I gather he's been working with Cyril, and mentions this paper.
Apologies, if I misunderstood the relation between this work.
http://biblio.telecom-paristech.fr/cgi-bin/download.cgi?id=14268

[2]
http://peterwonka.net/Publications/pdfs/2011.EG.Jeschke.DiffusionCurveFittin
g.AdditionalMaterial.pdf


-----Original Message-----
From: Erik Dahlstrom [mailto:ed@opera.com]
Sent: Friday, August 31, 2012 4:58 AM
To: www-svg@w3.org
Cc: public-fx@w3.org
Subject: Re: Perlin and simplex noise

On Thu, 30 Aug 2012 18:51:32 +0200, David Dailey <ddailey@zoominternet.net>
wrote:

> I know you all found the concepts of declarative randomness a bit
> distasteful (we're giving a talk on it at the conference in
> Switzerland,
> soon) and given the history of society's reaction to randomness, one
> can perhaps understand if not appreciate that reaction.
>
>
> However, I wondered if thought has been given to Simplex noise in
> addition to Perlin noise[1]?

Some thought has gone into that[2], but there's no concrete proposal for it
yet. It's listed as issue 15 in the filter draft [3].

I would like to see a more hardware friendly noise algorithm in the spec,
e.g simplex noise (or something with the same characteristics), but if we
want existing content to look the same we can't just switch the algorithm in
feTurbulence since the algorithm in the spec and the simplex algorithm
generate slightly different results. But, it may well be that the
differences are small enough that it would be acceptable, anyhow I think
that needs to be investigated.

What do other people think? The computational cost of the noise algorithm in
SVG 1.1 is fairly high, and that does limit what you can use it for in
practice. If we chose to go for a new noise algorithm I would also like to
be able to animate the noise continously (I think this means we'd need the
3d version of the algorithm). That is, I'd like to simulate say fire or
smoke, and link the z dimension in the noise algorithm to the time dimension
so that the animation is continous (without strange gaps and without it
looking like the result is scrolled along either or both of the x and
y-axis).


[1] http://en.wikipedia.org/wiki/Simplex_noise
[2] http://www.w3.org/2009/02/05-svg-minutes.html#item01
[3]
https://dvcs.w3.org/hg/FXTF/raw-file/tip/filters/index.html#feTurbulenceElem
ent

--
Erik Dahlstrom, Core Technology Developer, Opera Software Co-Chair, W3C SVG
Working Group Personal blog: http://my.opera.com/macdev_ed










 

-- 
www.sencha.com

"Amazing Apps with Web Technologies"

 
Received on Wednesday, 13 November 2013 03:30:24 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 March 2017 09:47:34 UTC