Piano is arguably one of the most important instruments in Western music due to its complexity and versatility. The size, weight, and price of grand pianos, and the relatively simple control surface (keyboard) have lead to the development of digital counterparts aiming to mimic the sound of the acoustic piano as closely as possible. While most commercial digital pianos are based on sample playback, it is also possible to reproduce the sound of the piano by modeling the physics of the instrument. The process of physical modeling starts with first understanding the physical principles, then creating accurate numerical models, and finally finding numerically optimized signal processing models that allow sound synthesis in real time by neglecting inaudible phenomena, and adding some perceptually important features by signal processing tricks. Accurate numerical models can be used by physicists and engineers to understand the functioning of the instrument, or to help piano makers in instrument development. On the other hand, efficient real-time models are aimed at composers and musicians performing at home or at stage. This paper will overview physics-based piano synthesis starting from the computationally heavy, physically accurate approaches and then discusses the ones that are aimed at best possible sound quality in real-time synthesis.