Three simplified variants of the context model initialization method of JVET-O0191 are proposed. The first one reduces the computational complexity by simplifying the computation of initialization values. The second variant reduces the number of bits required per initialization value from 6 to 4 bit. This allows to store the three initialization values and the 4 bit window size parameter in 16 bit per context model instead of 22 bit as necessary for the 6 bit variant. For the 6 bit variant, the luma BD rates over VTM-5.0 are 0.00% for AI, -0.01% for RA, 0.05% for LB, -0.01% for LP, 0.19% for RA-HighQP. For the 4 bit variant, the luma BD rates over VTM-5.0 are 0.07% for AI, 0.10 % for RA, 0.32% for LB, 0.27% for LP, and 0.48% for RA-HighQP. The third variant only changes the clipping operation with virtually no BD rate change for common test conditions but it improves the BD rate from 0.11% to -0.01% for RA-HighQP.