Embodiments of the present disclosure provide a method that comprises, based at least on two frames of a plurality of frames of a video stream, performing interpolation according to an interpolation mode to generate one or more interpolation frames. The method also comprises determining a merit of the interpolation based on one or more of (i) a measure of errors of the interpolation, and (ii) a measure of accuracy of one or more occlusion or reveal regions of the one or more interpolation frames. The method further comprises dynamically changing the interpolation mode based at least on the determined merit of the interpolation.