Abstract: Due to shortage of the realistic fog effect with the empirical model and the delay problem with the physical-based model,a combined fog simulation algorithm is proposed to improve the reality and real-time performance.First,the empirical model is adopted to simulate the basic fog,and an optimized fog formula is unified to compute the advection fog,the radiation fog,the layer fog and the volume fog by introducing the visibility factor and temperature coefficient.The unified formula enriched the diversity of fog effects.Secondly,the Simplex noise and the wind speed are introduced to improve the boundary noise model.The method overcomes the regularity issues of the traditional effect and optimizes the dynamic fog effect.Thirdly,the physical model is used to compensate the details.The Lagrangian method is used to simulate the details of the fog,and the wind field is established by the harmonic superposition method to improve the movement path of fog particles and to enhance the realistic effect.And also,the improved timing axis algorithm solves the unchangeable fog color problem to obtain gradually changeable fog effect with time.Finally,the approximate algorithm of the Mie scattering is calculated according to the Henyey-Greenstein phase function,and the fog particles are quickly rendered with the texture and billBoard technology.Then,the base fog and the detail fog are blended to generate realistic fog effort.Different simulation scenes are built on GPU.The comparison between the blended fog effect and the traditional effect is judged by the subjective effect and objective frame rates.Experimental results show that the algorithm simulates a variety of fog effect with outstanding details on the basis of the frame rates no more than 40.