Subject: Re: [magick-users] ImageMagick demo
From: Florent Monnier <fmonnier@linux-nantes.fr.eu.org>
Date: Tue, 28 Feb 2006 13:27:53 +0100
To: ImageMagick User List <magick-users@imagemagick.org>

> Wow. Neat video.  I like the way you added the plasma to the swirled
> hex bumpmap background before unswirling it for the start of the video.

The foreground is swirled in the middle but not at the beginning.  It is the
opposite for the background which is swirled at the beginning but not in the
middle. The sens of the rotation is the same so people don't always notice this
point, but without this trick the effect doesn't give the same felling, which
is that there is not a swirled point and an unswirled point.

Moreover the maximum degree of the swirl is not the same for the embossed
plasma background, and for the shaded hexagons foreground. These are
respectively 180 and 120. This provide (from my point of view) a nice sliding
sensation.

Even if you don't notice it, the image would look more static without it.  And
I think that if it would be more visible, the image would be less simple- and
lose its strength.


The plasma is embossed to make it more soft.

In the link I gave, the parameters are animated with a linear function, but
with the earlier version of Grinnm (see below) they are animated with a cosinus
function so the result is more fluid.

All source code is released under the GPL,
   http://www.linux-nantes.org/~fmonnier/OCaml/Grinnm/

here is the real source used for the video:
  http://www.linux-nantes.org/~fmonnier/OCaml/Grinnm/grinnm.B.ml.html

but it was made with an earlier version of Grinnm where the code and the data 
are mixed together.

Here is an equivalent with the last version of Grinnm:
  http://www.linux-nantes.org/~fmonnier/OCaml/Grinnm/script.B2.ml.html

Grinnm[1] is just a very early sandbox. The goal of Grinnm will be to provide
an interactive UI. For example with the 'make' utility, only the modified files
are compiled, not all. The goal of Grinnm will be to provide the same concept
for multi-layers structures. Which means while modifying a single parameter,
only the modified layers will be recompiled, not all, so we could see the
result without waiting too long. That's the goal, but it is far not done yet
;-)

The operands and the parameters labels have the same names than in bash and
Perl so there will be no difficulty to make versions from the others IM
interfaces.
http://www.linux-nantes.org/~fmonnier/OCaml/Grinnm/script.B2.ml.html

The only tricky part would be to make the equivalent of the cos function which 
makes evolve the parameters. I've never used the -fx argument yet but it 
doesn't seems hard ; I have seen that it provides the cos() function needed.
It is also possible to make a version without this tricks, with a more simple 
linear evolution rather than a sinusoidal one.