Two modifications have been proposed with respect to the GALF in the BMS. First one is disabling predefined fixed filter predictors, and second one is using logarithmic quantization (in base 2) to represent the outlying filter coefficients in such a way that multiplication operation is replaced by bit-shifting. The proposed method reportedly allows simplified hardware implementations due to the fact that 14 of the 21 multiplication operations in the filtering operation of a pixel sample is replaced by simple bit-shift operations. Experimental results of the combined method are as follows: