Physics-based animation can lead to systems with very complex and realistic behavior,

for example scenes involving pieces of cloth, fires, ropes, liquids, etc. However, providing

user control over physics-based animations is a daunting task. Among physical

models, fluid simulation brings up particularly difficult problems to the control process

because of the complexity and non linearity of the Navier-Stokes equations. Although

many methods and tools have been developed to convincingly simulate fluids, too few

methods provide efficient and intuitive ways to give a targeted behavior to a fluid. However

and in many cases, being able to control a fluid in order to make it behave according

to an intention is crucial to enhance the artistic process.

This document describes the PhD research we have done so far, whose main axis is

to introduce new artist-friendly ways of controlling liquid simulations. As part of our

first project, we have developed a system enabling animators to control particle-based

liquid simulations using precomputed patches of animation stored in a database. In this

system, liquid located in a control area is subjected to forces modifying its behavior in

order to locally reproduce the template instantiated by the animator. Temporary particles

can efficiently correct the density error between liquid and template regardless

of the surrounding environment. A 2D proof of concept showing some promising interactive

results has been implemented on CPU. Even if our control scheme leads to a

relatively small overhead, interactive frame rate can be maintained only for coarse or

small-scale simulations. Indeed and for high resolutions, the simulation itself becomes

too expensive. After extending the method to 3D liquids and as part of a second project,

we would like to develop a method to efficiently constrain a high-resolution liquid from

a low-resolution one in order to complete our first project. A generic optimization-based

formulation of particle-based liquid control applicable to several use cases will also be

investigated as part of a third project.

In the end, we hope that our contributions to this difficult, yet crucial field will steer

new fundamental and applied research to help animators to create even more astonishing

liquid animated visual effects.