DEFINITION AND ANIMATION OF BI- AND TRIDIMENSIONAL MANIFOLDS
BY MEANS OF PSEUDO-PROJECTIONS,
PICTURE SELF-TRANSFORMATIONS
Jean-François COLONNA
CMAP (Centre de Mathématiques APpliquées), Ecole Polytechnique, CNRS
france telecom, France Telecom R&D
91128 Palaiseau Cedex
France
[Site Map, Help and Search [Plan du Sîte, Aide et Recherche]]
[The Y2K bug [Le bug de l'an 2000]]
[Do you believe that Real Numbers exist for a computer and that floating point computations are safe?]
[Please, visit A Virtual Space-Time Travel Machine, the place where you can find more than 3230 pictures between Art and Science]
(this page was created on 12/24/2004)
(this page -belonging to the CMAP28 site- was last updated on 03/01/2010 20:51:53 -CET-)
[en français]
Abstract: Tridimensional surfaces -bidimensional manifolds- can be defined by means of three matrices
and then by means of three grey scale pictures -or again one color picture-.
An arbitrary dynamics of a tridimensional surface could then be defined by means of an animation.
This can be extended to higher dimensions and used to define picture self-transformation methods.
Keywords: Holographic Principle, Pseudo-Projection, Tridimensional Surfaces, Bidimensional Manifolds, Tridimensional Manifolds, Picture Self-Transformations.
Contents of this page:
1-PICTURES:
A black and white picture is defined as a rectangular array (or again a matrix) of pixels.
Each of them contains a numerical value; the smallest and biggest possible
values (0 and 255 most of the time) represent the black and the white levels
respectively.
A color picture can be defined as a set of three black and white pictures
each one of them corresponding to one of the primary colors (red,
green and blue respectively).
2-TRIDIMENSIONAL SURFACES (BIDIMENSIONAL MANIFOLDS):
Many surfaces -bidimensional manifolds- in a tridimensional space
can be defined using a set of three equations:
X = F (u,v)
x
Y = F (u,v)
y
Z = F (u,v)
z
with:
u E [U ,U ]
min max
v E [V ,V ]
min max
For example:
F (u,v) = R.sin(u).cos(v)
x
F (u,v) = R.sin(u).sin(v)
y
F (u,v) = R.cos(u)
z
with:
u E [0,pi]
v E [0,2.pi]
defines a sphere with R as the radius and the origin of the coordinates as the center.
[Umin,Umax]x[Vmin,Vmax] then defined a bidimensional rectangular domain D.
v ^
|
V |...... ---------------------------
max | |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
| |+++++++++++++++++++++++++++|
V |...... ---------------------------
min | : :
| : :
O------------------------------------------------->
U U u
min max
If D is sampled by means of a bidimensional rectangular grid (made of Nu.Nv points),
the three {X,Y,Z} coordinates can be defined by means of three rectangular matrices:
X = M (i,j)
x
Y = M (i,j)
y
Z = M (i,j)
z
with:
i = f(u,U ,U ,N )
min max u
j = g(v,V ,V ,N )
min max v
where 'f' and 'g' denote two obvious linear functions...
Then it is possible to define a tridimensional surface -a bidimensional manifold-
by means of three matrices.
It is noteworthy to recall that a picture is a matrix and then,
a tridimensional surface
can be defined by means of three (arbitrary) black and white pictures
or again by means of one (arbitrary) color picture: something like
an holographic principle. That will allow us to do with the
surfaces all that can be done with the pictures (filterings, interpolations,
arithmetic operations,...).
Here are some examples:
-
== 
a plane.
-
== 
a "pseudo-gaussian plane".
-
== 
a "spiraling plane".
-
== 
a Peano "fractal plane".
-
== 
a "fractal plane" with overhangings.
-
== 
a "fractal plane" with overhangings.
-
== 
the relief -modulus- of the function exp(1/z).
-
== 
the relief -modulus- of the function sin(z).
-
== 
a sphere.
-
== 
a crumpled sphere.
-
== 
a crumpled torus.
-
== 
the modulus of the complex sine function.
-
== 
the Mobius strip.
-
== 
the Klein bottle.
-
== 
a crumpled Klein bottle.
-
== 
the double Jeener bottle.
-
== 
the Jeener's triple Klein bottle.
-
== 
the Jeener's quadruple bilateral bottle.
-
== 
the Jeener's quintuple Klein bottle.
-
== 
the Bonan-Jeener's triple Klein bottle.
-
== 
a shell (Jeener surface 1).
-
== 
the Boy surface.
-
== 
a fractal surface (two iterations).
-
== 
a fractal surface (three iterations).
-
== 
a fractal surface (four iterations).
-
==
a fractal surface (five iterations).
-
== 
a fractal surface (six iterations).
-
== 
a fractal surface (numerous iterations...).
-
== 
a gaussian mixing of a sphere and of the Mobius strip.
-
== 
a Fourier interpolation between a sphere and the Bonan-Jeener's triple Klein bottle.
-
== 
the addition of a sphere and of the Bonan-Jeener's triple Klein bottle.
-
== 
a fractal interpolation between a sphere and the Bonan-Jeener's triple Klein bottle.
3-ANIMATION OF TRIDIMENSIONAL SURFACES (BIDIMENSIONAL MANIFOLDS):
Thus, a tridimensional surface can be defined by means of one color picture.
An arbitrary dynamics of a tridimensional surface could then be defined using a
set of color pictures or again a color animation.
Here are some examples:
At last, it is obvious that
a set of tridimensional surfaces could be defined using
the juxtaposition of color pictures or again
a set of color pictures -ie. a color animation-.
Here are some examples:
Holes and distorsions of the 'u' and 'v' coordinates can be added as on these examples:
-
a set of twisted spheres.
-
a set of spheres with holes.
At last, this process facilitates the interpolation between surfaces as on this example:
-
from a torus to a cylinder.
4-TRIDIMENSIONAL MANIFOLDS:
This can be extended to tridimensional manifolds. Thus:
Many tridimensional manifolds in a tridimensional space
can be defined using a set of three equations:
X = F (u,v,w)
x
Y = F (u,v,w)
y
Z = F (u,v,w)
z
with:
u E [U ,U ]
min max
v E [V ,V ]
min max
w E [W ,W ]
min max
[Umin,Umax]x[Vmin,Vmax]x[Wmin,Wmax] then defined a tridimensional rectangular domain D.
If D is sampled by means of a tridimensional rectangular grid (made of Nu.Nv.Nw points),
the three {X,Y,Z} coordinates can be defined by means of a set of Nw rectangular matrice triplets:
k
X = {M (i,j)}
x
k
Y = {M (i,j)}
y
k
Z = {M (i,j)}
z
with:
i = f(u,U ,U ,N )
min max u
j = g(v,V ,V ,N )
min max v
k = g(w,W ,W ,N )
min max w
where 'f', 'g' and 'h' denote two obvious linear functions...
Then it is possible to define a tridimensional manifold by means
of a set of Nw matrice triplets.
It is noteworthy to recall that an animation is a set of N color pictures and then,
a tridimensional manifold
can be defined by means of one (arbitrary) color animation...
Here are some examples:
5-PICTURE SELF-TRANSFORMATIONS:
Then, any color picture can be used to define a surface. Moreover,
this picture (seen as a texture) can be mapped on its associated surface (using the {u,v} coordinates
as the cartesian texture coordinates). For practical reasons,
it is suggested to symmetrize this texture to avoid discontinuities.
For example, the sphere is defined
with the following picture
as described earlier. After its symmetrization, it gives birth
to the so-called canonical texture of the sphere: 
.
This texture picture is distorted during the mapping;
this process is the so-called self-transformation
of the initial picture: 
.
Here are some examples of picture self-transformations without displaying the associated surfaces:
-
=> 
-
=> 
-
=> 
- =>
- =>
- =>
-
=> 
(in this example, a Fourier filtering was applied to the initial picture in order to smooth it
).
This process can be applied to color animations thus giving birth to complex texture animations.
Here are some examples:
Copyright (c) Jean-François Colonna, 2004-2010.
Copyright (c) France Telecom R&D and CMAP (Centre de Mathématiques APpliquées) / Ecole Polytechnique, 2004-2010.
