Eurographics/SIGGRAPH Symposium on Computer Animation (2003)

D. Breen, M. Lin (Editors)

Particle-Based Fluid Simulation for Interactive Applications

Matthias Müller, David Charypar and Markus Gross

Department of Computer Science, Federal Institute of Technology Zürich (ETHZ), Switzerland

Abstract

Realistically animated fluids can add substantial realism to interactive applications such as virtual surgery sim-
ulators or computer games. In this paper we propose an interactive method based on Smoothed Particle Hydro-
dynamics (SPH) to simulate fluids with free surfaces. The method is an extension of the SPH-based technique
by Desbrun to animate highly deformable bodies. We gear the method towards fluid simulation by deriving the
force density fields directly from the Navier-Stokes equation and by adding a term to model surface tension effects.
In contrast to Eulerian grid-based approaches, the particle-based approach makes mass conservation equations
and convection terms dispensable which reduces the complexity of the simulation. In addition, the particles can
directly be used to render the surface of the fluid. We propose methods to track and visualize the free surface using
point splatting and marching cubes-based surface reconstruction. Our animation method is fast enough to be used
in interactive systems and to allow for user interaction with models consisting of up to 5000 particles.
Categories and Subject Descriptors (according to ACM CCS): I.3.7 [Computer Graphics]: Three-Dimensional
Graphics and Realism

1. Introduction

1.1. Motivation

Fluids (i.e. liquids and gases) play an important role in every

day life. Examples for fluid phenomena are wind, weather,
ocean waves, waves induced by ships or simply pouring of
a glass of water. As simple and ordinary these phenomena
may seem, as complex and difficult it is to simulate them.
Figure 1: Pouring water into a glass at 5 frames per second.
Even though Computational Fluid Dynamics (CFD) is a well
established research area with a long history, there are still
many open research problems in the field. The reason for
the complexity of fluid behavior is the complex interplay of niques for fluids are medical simulators, computer games or
various phenomena such as convection, diffusion, turbulence any type of virtual environment.
and surface tension. Fluid phenomena are typically simu-
lated off-line and then visualized in a second step e.g. in
1.2. Related Work
aerodynamics or optimization of turbines or pipes with the
goal of being as accurate as possible. Computational Fluid Dynamics has a long history. In 1822
Claude Navier and in 1845 George Stokes formulated the
Less accurate methods that allow the simulation of fluid famous Navier-Stokes Equations that describe the dynam-
effects in real-time open up a variety of new applications. ics of fluids. Besides the Navier-Stokes equation which de-
In the fields mentioned above real-time methods help to scribes conservation of momentum, two additional equa-
test whether a certain concept is promising during the de- tions namely a continuity equation describing mass con-
sign phase. Other applications for real-time simulation tech- servation and a state equation describing energy conserva-

c The Eurographics Association 2003.

⃝
 Müller et al / Particle-Based Fluid Simulation for Interactive Applications

tion are needed to simulate fluids. Since those equations are of each particle using radial symmetrical smoothing kernels.
known and computers are available to solve them numeri- According to SPH, a scalar quantity A is interpolated at loca-
cally, a large number of methods have been proposed in the tion r by a weighted sum of contributions from all particles:
CFD literature to simulate fluids on computers. Aj
AS (r) = ∑ m j W (r − r j , h), (1)
Since about two decades, special purpose fluid simula- j ρj
tion techniques have been developed in the field of computer
where j iterates over all particles, m j is the mass of particle
graphics. In 1983 T. Reeves17 introduced particle systems as
j, r j its position, ρ j the density and A j the field quantity at
a technique for modeling a class of fuzzy objects. Since then
r j.
both, the particle-based Lagrangian approach and the grid-
based Eulerian approach have been used to simulate fluids The function W (r, h) is called the smoothing kernel with
in computer graphics. Desbrun and Cani2 and Tonnesen22 core radius h. Since we only use kernels with finite support,
use particles to animate soft objects. Particles have also been we use h as the radius of support in our formulation. If W is
used to animate surfaces7 , to control implicit surfaces23 and even (i.e. W (r, h) = W (−r, h)) and normalized, the interpo-
to animate lava flows20 . In recent years the Eulerian ap- lation is of second order accuracy. The kernel is normalized
proach has been more popular as for the simulation of fluids if
in general18 , water 4, 3, 21 , soft objects14 and melting effects1 . !
W (r)dr = 1. (2)
So far only a few techniques optimized for the use in in-
teractive systems are available. Stam’s method18 that is grid The particle mass and density appear in Eqn. (1) because
based is certainly an important step towards real-time simu- each particle i represents a certain volume Vi = mi /ρi . While
lation of fluids. For the special case of fluids that can be rep- the mass mi is constant throughout the simulation and, in our
resented by height fields, interactive animation techniques case, the same for all the particles, the density ρi varies and
are available as well6 . needs to be evaluated at every time step. Through substitu-
Here we propose a particle-based approach based on tion into Eqn. (1) we get for the density at location r:
Smoothed Particle Hydrodynamics to animate arbitrary fluid
motion. ρj
ρS (r) = ∑ m j W (r − r j , h) = ∑ m jW (r − r j , h). (3)
j ρj j
1.3. Our Contribution
In most fluid equations, derivatives of field quantities ap-
We propose a method based on Smoothed Particles Hydro-
pear and need to be evaluated. With the SPH approach, such
dynamics (SPH)9 to simulate fluids with free surfaces. Stam
derivatives only affect the smoothing kernel. The gradient of
and Fiume first introduced SPH to the graphics community
A is simply
to depict fire and other gaseous phenomena19 . Later, Des-
brun used SPH to animate highly deformable bodies2 . We Aj
∇AS (r) = ∑ m j ∇W (r − r j , h) (4)
extend his method focussing on the simulation of fluids. To j ρj
this end, we derive the viscosity and pressure force fields di-
while the Laplacian of A evaluates to
rectly from the Navier-Stokes equation and propose a way
to model surface tension forces. For the purpose of interac- Aj 2
∇2 AS (r) = ∑ m j ∇ W (r − r j , h). (5)
tivity, we designed new special purpose smoothing kernels. j ρj
Surface tracking and surface rendering at interactive rates
are difficult problems for which we describe possible solu- It is important to realize that SPH holds some inherent prob-
tions. lems. When using SPH to derive fluid equations for particles,
these equations are not guaranteed to satisfy certain physi-
cal principals such as symmetry of forces and conservation
2. Smoothed Particle Hydrodynamics of momentum. The next section describes our SPH-based
Although Smoothed Particle Hydrodynamics (SPH) was de- model and techniques to solve these SPH-related problems.
veloped by Lucy9 and Gingold and by Monaghan5 for the
simulation of astrophysical problems, the method is general 3. Modelling Fluids with Particles
enough to be used in any kind of fluid simulation. For in-
In the Eulerian (grid based) formulation, isothermal fluids
troductions to SPH we refer the reader to Monaghan10 or
are described by a velocity field v, a density field ρ and a
Münzel13 .
pressure field p. The evolution of these quantities over time
SPH is an interpolation method for particle systems. With is given by two equations. The first equation assures conser-
SPH, field quantities that are only defined at discrete parti- vation of mass
cle locations can be evaluated anywhere in space. For this ∂ρ
purpose, SPH distributes quantities in a local neighborhood + ∇ · (ρv) = 0, (6)
∂t

c The Eurographics Association 2003.

⃝
 Müller et al / Particle-Based Fluid Simulation for Interactive Applications

while the Navier-Stokes equation15 formulates conservation Since particles only carry the three quantities mass, posi-
of momentum tion and velocity, the pressure at particle locations has to be
" # evaluated first. This is done in two steps. Eqn. (3) yields the
∂v
ρ + v · ∇v = −∇p + ρg + µ∇2 v, (7) density at the location of the particle. Then, the pressure can
∂t
be computed via the ideal gas state equation
where g is an external force density field and µ the viscosity
of the fluid. Many forms of the Navier-Stokes equation ap- p = kρ, (11)
pear in the literature. Eqn. (7) represents a simplified version
where k is a gas constant that depends on the temperature.
for incompressible fluids.
In our simulations we use a modified version of Eqn. (11)
The use of particles instead of a stationary grid simplifies suggested by Desbrun2
these two equations substantially. First, because the number
p = k(ρ − ρ0 ), (12)
of particles is constant and each particle has a constant mass,
mass conservation is guaranteed and Eqn. (6) can be omit- where ρ0 is the rest density. Since pressure forces depend on
ted completely. Second, the expression ∂v/∂t + v · ∇v on the the gradient of the pressure field, the offset mathematically
left hand side of Eqn. (7) can be replaced by the substantial has not effect on pressure forces. However, the offset does
derivative Dv/Dt. Since the particles move with the fluid, influence the gradient of a field smoothed by SPH and makes
the substantial derivative of the velocity field is simply the the simulation numerically more stable.
time derivative of the velocity of the particles meaning that
the convective term v · ∇v is not needed for particle systems.
3.2. Viscosity
There are three force density fields left on the right hand
side of Eqn. (7) modeling pressure (−∇p), external forces Application of the SPH rule to the viscosity term µ∇2 v again
(ρg) and viscosity (µ∇2 v). The sum of these force density yields asymmetric forces
fields f = −∇p + ρg + µ∇2 v determines the change of mo- vj 2
= µ∇2 v(ra ) = µ ∑ m j
viscosity
mentum ρ Dv fi ∇ W (ri − r j , h). (13)
Dt of the particles on the left hand side. For the ρj
j
acceleration of particle i we, thus, get:
dvi f because the velocity field varies from particle to particle.
ai = = i, (8) Since viscosity forces are only dependent on velocity differ-
dt ρi
ences and not on absolute velocities, there is a natural way
where vi is the velocity of particle i and fi and ρi are the force to symmetrize the viscosity forces by using velocity differ-
density field and the density field evaluated at the location of ences:
particle i, repectively. We will now describe how we model
v j − vi 2
= µ∑ m j
viscosity
the force density terms using SPH. fi ∇ W (ri − r j , h). (14)
j ρj

3.1. Pressure A possible interpretation of Eqn. (14) is to look at the neigh-

bors of particle i from i’s own moving frame of reference.
Application of the SPH rule described in Eqn. (1) to the pres- Then particle i is accelerated in the direction of the relative
sure term −∇p yields speed of its environment.
pj
= −∇p(ri ) = − ∑ m j ∇W (ri − r j , h). (9)
pressure
fi
j ρj
3.3. Surface Tension
Unfortunately, this force is not symmetric as can be seen
We model surface tension forces (not present in Eqn. (7))
when only two particles interact. Since the gradient of the
explicitly based on ideas of Morris12 . Molecules in a fluid
kernel is zero at its center, particle i only uses the pressure of
are subject to attractive forces from neighboring molecules.
particle j to compute its pressure force and vice versa. Be-
Inside the fluid these intermolecular forces are equal in all
cause the pressures at the locations of the two particles are
directions and balance each other. In contrast, the forces act-
not equal in general, the pressure forces will not be symmet-
ing on molecules at the free surface are unbalanced. The net
ric. Different ways of symmetrization of Eqn. (9) have been
forces (i.e. surface tension forces) act in the direction of the
proposed in the literature. We suggest a very simple solution
surface normal towards the fluid. They also tend to mini-
which we found to be best suited for our purposes of speed
mize the curvature of the surface. The larger the curvature,
and stability
the higher the force. Surface tension also depends on a ten-
pi + p j sion coefficient σ which depends on the two fluids that form
= −∑mj
pressure
fi ∇W (ri − r j , h).. (10)
j 2ρ j the surface.
The so computed pressure force is symmetric because it uses The surface of the fluid can be found by using an addi-
the arithmetic mean of the pressures of interacting particles. tional field quantity which is 1 at particle locations and 0

c The Eurographics Association 2003.

⃝
 Müller et al / Particle-Based Fluid Simulation for Interactive Applications
2

everywhere else. This field is called color field in the litera- 14

1.8

1.6

ture. For the smoothed color field we get: 12

10
1.4

1.2

1 8

cS (r) = ∑ m j W (r − r j , h).
0.8

(15)
2
6
0.6

ρj
4
0.4
1

j 2 0.2

–1 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 –1 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 –1 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1

–2
–1

The gradient field of the smoothed color field –2

n = ∇cs (16)
Figure 2: The three smoothing kernels Wpoly6 , Wspiky and
yields the surface normal field pointing into the fluid and the Wviscosity (from left to right) we use in our simulations. The
divergence of n measures the curvature of the surface thick lines show the kernels, the thin lines their gradients
−∇2 cs in the direction towards the center and the dashed lines the
κ= . (17) Laplacian. Note that the diagrams are differently scaled. The
|n|
curves show 3-d kernels along one axis through the center
The minus is necessary to get positive curvature for con- for smoothing length h = 1.
vex fluid volumes. Putting it all together, we get for the sur-
face traction:
n
tsurface = σκ (18)
|n|

To distribute the surface traction among particles near the that is can be evaluated without computing square roots in
surface and to get a force density we multiply by a normal- distance computations. However, if this kernel is used for the
ized scalar field δs = |n| which is non-zero only near the computation of the pressure forces, particles tend to build
surface. For the force density acting near the surface we get clusters under high pressure. As particles get very close to
n each other, the repulsion force vanishes because the gradient
fsurface = σκn = −σ∇2 cS (19)
|n| of the kernel approaches zero at the center. Desbrun2 solves
Evaluating n/|n| at locations where |n| is small causes nu- this problem by using a spiky kernel with a non vanishing
merical problems. We only evaluate the force if |n| exceeds gradient near the center. For pressure computations we use
a certain threshold. Debrun’s spiky kernel
$
15 (h − r)3 0≤r≤h
3.4. External Forces Wspiky (r, h) = (21)
πh6 0 otherwise,
Our simulator supports external forces such as gravity, col-
lision forces and forces caused by user interaction. These that generates the necessary repulsion forces. At the bound-
forces are applied directly to the particles without the use of ary where it vanishes it also has zero first and second deriva-
SPH. When particles collide with solid objects such as the tives.
glass in our examples, we simply push them out of the ob-
ject and reflect the velocity component that is perpendicular Viscosity is a phenomenon that is caused by friction and,
to the object’s surface. thus, decreases the fluid’s kinetic energy by converting it into
heat. Therefore, viscosity should only have a smoothing ef-
fect on the velocity field. However, if a standard kernel is
3.5. Smoothing Kernels
used for viscosity, the resulting viscosity forces do not al-
Stability, accuracy and speed of the SPH method highly de- ways have this property. For two particles that get close to
pend on the choice of the smoothing kernels. The kernels each other, the Laplacian of the smoothed velocity field (on
we use have second order interpolation errors because they which viscosity forces depend) can get negative resulting in
are all even and normalized (see Fig. 2). In addition, kernels forces that increase their relative velocity. The artifact ap-
that are zero with vanishing derivatives at the boundary are pears in coarsely sampled velocity fields. In real-time appli-
conducive to stability. Apart from those constraints, one is cations where the number of particles is relatively low, this
free to design kernels for special purposes. We designed the effect can cause stability problems. For the computation of
following kernel viscosity forces we, thus, designed a third kernel:
$
315 (h2 − r2 )3 0 ≤ r ≤ h $ 3
r r2 h
Wpoly6 (r, h) = (20) 15 0≤r≤h − 2h 3 + h2 + 2r − 1
9
64πh 0 otherwise Wviscosity (r, h) =
2πh3 otherwise. 0
and use it in all but two cases. An important feature of this (22)
simple kernel is that r only appears squared which means whose Laplacian is positive everywhere with the following

c The Eurographics Association 2003.

⃝
 Müller et al / Particle-Based Fluid Simulation for Interactive Applications

additional properties: We start searches from all the cells that contain surface par-
45 ticles and from there recursively traverse the grid along the
∇2W (r, h) = (h − r) surface. With the use of a hash table we make sure that the
πh6
cells are not visited more than once. For each cell identified
W (|r| = h, h) = 0
to contain the surface, the triangles are generated via a fast
∇W (|r| = h, h) = 0 table lookup.
The use of this kernel for viscosity computations increased
the stability of the simulation significantly allowing to omit 5. Implementation
any kind of additional damping.
Since the smoothing kernels used in SPH have finite support
h, a common way to reduce the computational complexity is
3.6. Simulation to use a grid of cells of size h. Then potentially interacting
For the integration of the Eqn. (8) we use the Leap-Frog partners of a particle i only need to be searched in i’s own
scheme16 . As a second order scheme for which the forces cell and all the neighboring cells. This technique reduces the
need to be evaluation only once, it best fits our purposes and time complexity of the force computation step from O(n2 )
in our examples allows time steps up to 10 milliseconds. For to O(nm), m being the average number of particles per grid
the examples we used constant time steps. We expect even cell.
better performance if adaptive time steps are used based on With a simple additional trick we were able to speed
the Courant-Friedrichs-Lewy condition2 . up the simulation by an additional factor of 10. Instead of
storing references to particles in the grid, we store copies
4. Surface Tracking and Visualization of the particle objects in the grid cells (doubling memory
consumption). The reason for the speed up is the prox-
The color field cS and its gradient field n = ∇cS defined in imity in memory of the information needed for interpola-
section 3.3 can be used to identify surface particles and to tion which dramatically increases the cash hit rate. Further
compute surface normals. We identify a particle i as a surface speedup might be possible through even better clustering us-
particle if ing Hilbert space filling curves11 . The data structure for fast
|n(ri )| > l, (23) neighbor searches is also used for surface tracking and ren-
dering.
where l is a threshold parameter. The direction of the surface
normal at the location of particle i is given by
6. Results
−n(ri ). (24)
The water in the glass shown in Fig. 3 is sampled with 2200
particles. An external rotational force field causes the fluid
4.1. Point Splatting to swirl. The first image (a) shows the individual particles.
We now have a set of points with normals but without For the second image (b), point splatting was used to ren-
connectivity information. This is exactly the type of infor- der the free surface only. In both modes, the animation runs
mation needed for point splatting techniques24 . However, at 20 frames per second on a 1.8 GHz Pentium IV PC with
these methods are designed to work with point clouds ob- a GForce 4 graphics card. The most convincing results are
tained from scanners that typically contain at least 10, 000 produced when the iso surface of the color field is visual-
to 100, 000 points. We only use a few thousand particles a ized using the marching cubes algorithm as in image (c).
fraction of which are identified as being on the surface. Still However, in this mode the frame rate drops to 5 frames per
surface splatting yields plausible results as shown in the re- second. Still this frame rate is significantly higher than the
sults section. one of most off-line fluid simulation techniques and with the
next generation of graphics hardware, real-time performance
We are currently working on ways to upsample the sur- will be possible.
face of the fluid. Hereby, the color field information of sur-
face particles is interpolated to find locations for additional The image sequence shown in Fig. 4 demonstrates inter-
particles on the surface only used for rendering. action with the fluid. Through mouse motion, the user gen-
erates an external force field that cause the water to splash.
The free surface is rendered using point splatting while iso-
4.2. Marching Cubes lated particles are drawn as single droplets. The simulation
with 1300 particles runs at 25 frames per second.
Another way to visualize the free surface is by rendering an
iso surface of the color field cS . We use the marching cubes For the animation shown in Fig. 5 we used 3000 particles
algorithm8 to triangulate the iso surface. In a grid fixed in and rendered the surface with the marching cubes technique
space the cells that contain the surface are first identified. at 5 frames per second.

c The Eurographics Association 2003.

⃝
 Müller et al / Particle-Based Fluid Simulation for Interactive Applications

7. Conclusions and Future Work 8. William E. Lorensen and Harvey E. Cline. Marching cubes:
A high resolution 3d surface construction algorithm. In Pro-
We have presented a particle-based method for interactive ceedings of the 14th annual conference on Computer graphics
fluid simulation and rendering. The physical model is based and interactive techniques, pages 163–169. ACM Press, 1987.
on Smoothed Particle Hydrodynamics and uses special pur-
9. L. B. Lucy. A numerical approach to the testing of the fission
pose kernels to increase stability and speed. We have pre-
hypothesis. The Astronomical Journal, 82:1013–1024, 1977.
sented techniques to track and render the free surface of flu-
ids. The results are not as photorealistic yet as animations 10. J. J. Monaghan. Smoothed particle hydrodynamics. Annual
computed off-line. However, given that the simulation runs Review of Astronomy and Astrophysics, 30:543–574, 1992.
at interactive rates instead of taking minutes or hours per 11. B. Moon, H. V. Jagadish, C. Faloutsos, and J. H. Saltz. Analy-
frame as in today’s off-line methods, the results are quite sis of the clustering properties of hilbert space-filling curve.
promising. IEEE Transactions on Knowledge and Data Engineering,
13(1):124–141, 2001.
While we are quite content with the physical model, track-
ing and rendering of the fluid surface in real time certainly 12. J. P. Morris. Simulating surface tension with smoothed particle
remains an open research problem. In the future we will in- hydrodynamics. International Journal for Numerical Methods
in Fluids, 33(3):333–353, 2000.
vestigate upsampling techniques as well as ways to increase
the performance of the marching cubes-based algorithm. 13. S. A. Münzel. Smoothed Particle Hydrodynamics und ihre An-
wendung auf Akkretionsscheiben. PhD thesis, Eberhard-Karls-
Universität Thübingen, 1996.
Acknowledgements 14. D. Nixon and R. Lobb. A fluid-based soft-object model.
The authors would like to thank Simone Hieber for her help- IEEE Computer Graphics and Applications, pages 68–75,
July/August 2002.
ful comments and Rolf Bruderer, Simon Schirm and Thomas
Rusterholz for their contributions to the real-time system. 15. D. Pnueli and C. Gutfinger. Fluid Mechanics. Cambridge
Univ. Press, NY, 1992.
16. C. Pozrikidis. Numerical Computation in Science and Engi-
References neering. Oxford Univ. Press, NY, 1998.
1. Mark Carlson, Peter J. Mucha, III R. Brooks Van Horn, and 17. W. T. Reeves. Particle systems — a technique for modeling a
Greg Turk. Melting and flowing. In Proceedings of the ACM class of fuzzy objects. ACM Transactions on Graphics 2(2),
SIGGRAPH symposium on Computer animation, pages 167– pages 91–108, 1983.
174. ACM Press, 2002.
18. Jos Stam. Stable fluids. In Proceedings of the 26th annual
2. M. Desbrun and M. P. Cani. Smoothed particles: A new conference on Computer graphics and interactive techniques,
paradigm for animating highly deformable bodies. In pages 121–128. ACM Press/Addison-Wesley Publishing Co.,
Computer Animation and Simulation ’96 (Proceedings of 1999.
EG Workshop on Animation and Simulation), pages 61–76.
Springer-Verlag, Aug 1996. 19. Jos Stam and Eugene Fiume. Depicting fire and other gaseous
phenomena using diffusion processes. Computer Graphics,
3. D. Enright, S. Marschner, and R. Fedkiw. Animation and ren- 29(Annual Conference Series):129–136, 1995.
dering of complex water surfaces. In Proceedings of the 29th
annual conference on Computer graphics and interactive tech- 20. Dan Stora, Pierre-Olivier Agliati, Marie-Paule Cani, Fabrice
niques, pages 736–744. ACM Press, 2002. Neyret, and Jean-Dominique Gascuel. Animating lava flows.
In Graphics Interface, pages 203–210, 1999.
4. N. Foster and R. Fedkiw. Practical animation of liquids.
21. T. Takahashi, U. Heihachi, A. Kunimatsu, and H. Fujii. The
In Proceedings of the 28th annual conference on Computer
simulation of fluid-rigid body interaction. ACM Siggraph
graphics and interactive techniques, pages 23–30. ACM Press,
Sketches & Applications, July 2002.
2001.
22. D. Tonnesen. Dynamically Coupled Particle Systems for Geo-
5. R. A. Gingold and J. J. Monaghan. Smoothed particle hy-
metric Modeling, Reconstruction, and Animation. PhD thesis,
drodynamics: theory and application to non-spherical stars.
University of Toronto, November 1998.
Monthly Notices of the Royal Astronomical Society, 181:375–
398, 1977. 23. Andrew Witkin and Paul Heckbert. Using particles to sample
and control implicit surfaces. In Computer Graphics (Proc.
6. Damien Hinsinger, Fabrice Neyret, and Marie-Paule Cani. In- SIGGRAPH ’94), volume 28, 1994.
teractive animation of ocean waves. In Proceedings of the
ACM SIGGRAPH symposium on Computer animation, pages 24. Matthias Zwicker, Hanspeter Pfister, Jeroen van Baar, and
161–166. ACM Press, 2002. Markus Gross. Surface splatting. In Proceedings of the 28th
annual conference on Computer graphics and interactive tech-
7. Jean-Christophe Lombardo and Claude Puech. Oriented parti- niques, pages 371–378. ACM Press, 2001.
cles: A tool for shape memory objects modelling. In Graphics
Interface’95, pages 255–262, mai 1995. Quebec city, Canada.

c The Eurographics Association 2003.

⃝
 Müller et al / Particle-Based Fluid Simulation for Interactive Applications

(a) (b) (c)

Figure 3: A swirl in a glass induced by a rotational force field. Image (a) shows the particles, (b) the surface using point
splatting and (c) the iso-surface triangulated via marching cubes.

Figure 4: The user interacts with the fluid causing it to splash.

Figure 5: Pouring water into a glass at 5 frames per second.

c The Eurographics Association 2003.

⃝