A new way of rendering?

[Mats Öljare]

While working with the home made programs discussed in the previous
thread,  i got an idea that someone here might want to try realizing.

What if, instead of rendering a single point for each pixel on the
Mandelbrot set, rendered a full Julia set for that value, and used the
"mass" of that Julia set - the amount of attracted vs. escaping points
- to decide the color of the pixel on the Mandelbrot set?

The amount of computation required for this would be absolutely
MASSIVE, since a full Julia set has to be rendered for each pixel, but
it would be useful for creating a map for those strange functions
whose Mandelbrot set doesn't really describe the corresponding Julia
set, and it might make for a very interesting rendering of the
"regular" formulas, as well.

There would even be a corresponding "Super-Julia" set, where every
value of C is tested for each pixel representing a point on the Z-plane.

Anyone know if this have been tried already?

/Mats Öljare

[Mats Öljare]

Oh, and possibly, one could just use the real or imaginary "line" of Z
for each pixel, rather than the full two dimensional plane. That would
greatly reduce the computation needed while still being a novel way of
rendering.

/Ö

[webmonkey]

You might consider getting a used Cray supercomputer.  Or I am sure that
"Blue Gene" can probably be had for the right price ;-)

Sunday Humor....

Phil J.


On Sun, 2008-08-03 at 23:50 +0000, Mats Öljare wrote:

> While working with the home made programs discussed in the previous
> thread, i got an idea that someone here might want to try realizing.
>
> What if, instead of rendering a single point for each pixel on the
> Mandelbrot set, rendered a full Julia set for that value, and used the
> "mass" of that Julia set - the amount of attracted vs. escaping points
> - to decide the color of the pixel on the Mandelbrot set?
>
> The amount of computation required for this would be absolutely
> MASSIVE, since a full Julia set has to be rendered for each pixel, but
> it would be useful for creating a map for those strange functions
> whose Mandelbrot set doesn't really describe the corresponding Julia
> set, and it might make for a very interesting rendering of the
> "regular" formulas, as well.
>
> There would even be a corresponding "Super-Julia" set, where every
> value of C is tested for each pixel representing a point on the
> Z-plane.
>
> Anyone know if this have been tried already?
>
> /Mats Öljare
>
>
>
>
>  

[Jason Champion-3]

Considering that a lot of the original fractal software was meant to run on a 386 or less, it's not completely out of the realm of possibility that you could do this, especially with the average modern system having 1000x that processing power.

Even so, renders would take forever unless you used some form of "limited Julia set", such as only rendering a 16x16-pixel version to base your Mandelbrot on, or creating a set of pre-generated lookup tables from a set of known values for Julia and clamping your calculations to one of those lookup tables.

Sounds neat, though.  I think you should try it. :)

--- On Sun, 8/3/08, PhilJackson <pjackson1@...> wrote:
From: PhilJackson <pjackson1@...>
Subject: Re: [cnfractal_music] A new way of rendering?
To: cnfractal_music@...
Date: Sunday, August 3, 2008, 8:07 PM










   
            You might consider getting a used Cray supercomputer.  Or I am sure that

"Blue Gene" can probably be had for the right price ;-)



Sunday Humor....



Phil J.



On Sun, 2008-08-03 at 23:50 +0000, Mats Öljare wrote:

> While working with the home made programs discussed in the previous

> thread, i got an idea that someone here might want to try realizing.

>

> What if, instead of rendering a single point for each pixel on the

> Mandelbrot set, rendered a full Julia set for that value, and used the

> "mass" of that Julia set - the amount of attracted vs. escaping points

> - to decide the color of the pixel on the Mandelbrot set?

>

> The amount of computation required for this would be absolutely

> MASSIVE, since a full Julia set has to be rendered for each pixel, but

> it would be useful for creating a map for those strange functions

> whose Mandelbrot set doesn't really describe the corresponding Julia

> set, and it might make for a very interesting rendering of the

> "regular" formulas, as well.

>

> There would even be a corresponding "Super-Julia" set, where every

> value of C is tested for each pixel representing a point on the

> Z-plane.

>

> Anyone know if this have been tried already?

>

> /Mats Öljare

>

>

>

>

>  




     

   
   
       
         
       
       








       


       
       

[Non-text portions of this message have been removed]

[Lauri Grohn-2]

The problem of using pictures for generating
music is not in the limeted  amount of data. Even
400 x 400 pixel photos contain 640000 bytes of
data. The problems are elsewhere.

Lauri Gröhn
metacomposer
www.synestesia.fi

[nosuch_tim]

 > The problem of using pictures for generating
 > music is not in the limeted amount of data. Even
 > 400 x 400 pixel photos contain 640000 bytes of
 > data. The problems are elsewhere.

That makes me think of the legendary saying about "who would want more than
640K of memory?"  :-)  If more data opens the mind of someone to a new
algorithm or compositional technique that hasn't been tried before, I say
"go for it", and let's check out the results.  I love seeing (and hearing)
people try new algorithmic/compositional techniques.

       ...Tim...    

[webmonkey]

Could be interesting!  Good thoughts on this idea.  I suppose it all
boils down to how this data, whether it be a 400px X 400px arbitrary
image or a billion J-Sets for each point on an M-Set; is interpreted and
acted upon programmaticly to produce the basic musical patterns.

Jason did make a very good point - some of the earlier stuff built to
run on systems with much more limited resources than current systems did
produce amazing results & must have made very efficient use of those
resources.  The Atari MIDI software you can run via STEEM / XSTEEM
certainly come to mind in that respect.  And ArtSong, Gingerbread,
FMusic, Musinum which were all, I think, originally running on 16-bit
very small machines.


Phil J.


On Mon, 2008-08-04 at 08:46 -0700, Tim Thompson wrote:

> > The problem of using pictures for generating
> > music is not in the limeted amount of data. Even
> > 400 x 400 pixel photos contain 640000 bytes of
> > data. The problems are elsewhere.
>
> That makes me think of the legendary saying about "who would want more
> than
> 640K of memory?" :-) If more data opens the mind of someone to a new
> algorithm or compositional technique that hasn't been tried before, I
> say
> "go for it", and let's check out the results. I love seeing (and
> hearing)
> people try new algorithmic/compositional techniques.
>
> ...Tim...
>
>
>
>
>  

[Mats Öljare]

> Could be interesting!  Good thoughts on this idea.  I suppose it all
> boils down to how this data, whether it be a 400px X 400px arbitrary
> image or a billion J-Sets for each point on an M-Set; is interpreted and
> acted upon programmaticly to produce the basic musical patterns.

Well, my suggestions was for a way to generate an image by summing the
Julia sets, not for sound.

[webmonkey]

So IBM sold you one of their older last generation "Blue Gene"'s to do
the processing with ;-)

You might run your ideas past this person:
http://www.ktaza.com/jsp/index.jsp

Stephen C. Ferguson, who has written a number of fractal graphics
systems that were quite good.  If you can get into contact with him,
perhaps he can provide feedback as to the feasibility of your idea.

Phil J.



On Wed, 2008-08-06 at 22:05 +0000, Mats Öljare wrote:

> > Could be interesting! Good thoughts on this idea. I suppose it all
> > boils down to how this data, whether it be a 400px X 400px arbitrary
> > image or a billion J-Sets for each point on an M-Set; is interpreted
> and
> > acted upon programmaticly to produce the basic musical patterns.
>
> Well, my suggestions was for a way to generate an image by summing the
> Julia sets, not for sound.
>
>
>
>
>  

[Mats Öljare]

Well, here's some examples i made, for those who are curious:

http://i161.photobucket.com/albums/t208/oljare/mand_real.gif

http://i161.photobucket.com/albums/t208/oljare/mandimag.gif

http://i161.photobucket.com/albums/t208/oljare/juliareal.gif