This paper presents a method that exploits an optical illusion due to the motion of video sequences for improving the coding efficiency of sequences with large motion. The employed optical illusion is a special case of the motion sharpening, which we call “quasi-motion sharpening”. Under this effect, comparable subjective quality is perceived when comparing an original video against the processed same video where some images between two frames are being blurred. A quantitative model based on this phenomenon has been constructed for deriving the level of blurriness of selected frames without sacrificing the subjective quality of the whole sequence. Subjective experiments on decoded video of H.264 show that, using our model for pre-processing video sequences with fast motion, we can improved the coding efficiency by 9.3∼14.0% while maintaining the subjective quality.