“…This is useful since more complex CTUs tend to have larger depths to achieve better motion compensation. Numerous information sources are exploited here from directional gradients [79], pyramidal motion divergence [60,61], RD cost of encoding CU [23], etc.However, the most common approach is to determine the texture complexity of the CTU using the variance of pixels since variance is strongly correlated with the texture complexity of the CTU [62,63,65,67,76,80]. Moreover, motion vectors are key factors in determining texture complexity as well, and they are also exploited commonly in these methods [61,73,74,77].…”