Efficient Digital Film Grain Simulation.

We propose an efficient digital simulation of film grain noise that is based on the random-dot model, applies to a wide range of imaging conditions (magnification factors, optical blur), and accounts for the signal-dependence and spatial correlation of film grain. Our analysis and experimental results demonstrate that the proposed approach produces sharper images with similar film grain compared to Monte Carlo simulations proposed by Newson et al., offering superior rendering efficiency across a wide range of parameter settings including a case where Monte Carlo simulations fail (low magnification with low blur). While recently proposed statistical approximations of film grain by Zhang et al. are significantly faster, they cannot be used for high magnification or low blur.