CPC H04N 19/513 (2014.11) [H04N 19/119 (2014.11); H04N 19/184 (2014.11); H04N 19/44 (2014.11); H04N 19/543 (2014.11)] | 20 Claims |
1. A method of coding video data, comprising:
deriving motion information for a current video block;
dividing, based on a size of the current video block, the current video block into one or multiple sub-blocks;
calculating a motion offset for each sub-block of the one or multiple sub-blocks by using only luma component of each sub-block;
deriving, based on the motion offset and a signaled luma motion vector, a first refined motion information for luma component of each sub-block of the one or multiple sub-blocks and a second refined motion information for chroma component of each sub-block of the one or multiple sub-blocks using a decoder motion vector refinement (DMVR) tool; and
performing a conversion between the current video block and a bitstream using the first refined motion information and the second refined motion information,
wherein deriving the first refined motion information or the second refined motion information for each sub-block further comprises:
determining, based on the motion information of the current video block, at least one region in a reference picture for the each sub-block;
using one row of every N rows in each of the at least one region to calculate at least one matching cost;
determining the first refined motion information or the second refined motion information for the each sub-block based on the at least one matching cost.
|