||In this thesis I present physically-based techniques for building real-time digital painting tools. In particular, I focus on two components that are most needed in existing paint systems: 3D brush modeling and water-based paint simulation. A 3D deformable brush model is crucial in generating organic brush strokes. When paired with a proper input device, such a brush model also allows true-to-life control of stroke creation because users can see how the brush deforms in response to their manipulation while painting. Existing paint programs do simulate various paint media including charcoal, oil, and watercolor. However, water-based media like Eastern ink and Western watercolor are the most unsatisfactory due to their inability to produce natural-looking marks, especially those with flow effects. In the second part of my thesis, I present a technique based on a relatively new computational fluid dynamics method called Lattice Boltzmann to simulate water-based paint media. Many of the existing real-world artistic effects as well as novel ones are achievable with this approach.