Fabrice
NEYRET -
Maverick team, LJK, at INRIA-Montbonnot
Joelle
THOLLOT
-
Maverick team, LJK, at INRIA-Montbonnot
Romain
VERGNE -
Maverick team, LJK, at INRIA-Montbonnot
One of the Graals of Computer Graphics is the realistic rendering of ultra large and detailed scenes, possibly with real-time walk-through. Virtual exploration of galaxies is a challenging example of it : ultra complex, large and detailed, with very little corresponding 3D data and physical models, while people are already accustomed to detailed Hubble photos at various scales from whole galaxy ( left image - NGC3810 ) to nebula (right).
Procedural noise such as Perlin noise (see refs) is a classical
CG way to produce cheaply very large and detailed natural looking
stochastic patterns. But many real-world patterns don't really fit
their look, in particular the dark fractal clouds of galactic dust.
Also, deep zooming or anti-aliasing filtering require that the look
doesn't shift as you zoom: the procedural pixels values must mimic
the rendering of smaller scales (or of higher resolution).
Also,
the dust is not a pure same all-scale fractal: At large scales it is
organized as spurs (or plumes) within arms. At small scale H||
nebulae carve explosion holes illuminated by
ionization ( right image; see also this shadertoy
). A true multi-scale model must be able to account for or adapt to
these.
From our previous researches we already have some ingredients: the concept of unitary multiplicative noise ( middle images; see also this shadertoy ), an isolated nebula model (right image), many inputs from the astrophysical world, plus a strong experience of filterable models and real-time realism.
The main purpose of the subject is to explore and study the
various options to make a complete model from the new unitary
multiplicative noise technique, in 2D and in 3D,
ensuring the continuity through scales, and the various user controls
that can be offered to tuned the resulting look. Note that while the
first target is galactic dust cloud, such a new procedural tool can
have broader applications.
Then we want to study the integration
with (illuminated) perturbations such the nebulae. A last and more
open goal is the modelisation of spurs at arm levels, i.e.
half-stochastic half-organized patterns.
General culture in Computer Graphics and Math ( textures, proceduralism, Perlin noise, fractals would be a plus)
C/C++,
GLSL and OpenGL would be a plus, but these can be learn easily during the practice or before.
.jpg){kind=link}