Prediction refinement for affine merge and affine motion vector prediction mode

An apparatus for processing video data comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to:

• determine, for a conversion between a current video block of a video and a bitstream of the video, a size of a prediction block corresponding to the current video block according to a rule;
• and
• perform the conversion based on the determining,
• wherein an affine merge mode is enabled for the current video block,
• wherein the rule specifies that a first size of the prediction block is determined responsive to whether a prediction refinement using optical flow technique is enabled for the current video block, and
• wherein the current video block has a second size;
• wherein a second width and a second height of the second size of the current video block are indicated by M and N, respectively, and M and N are integers greater than or equal to 0,
• wherein a prediction sample of the prediction block is present as predSamplesLX[xL][yL],
• wherein xL is between 0 and M+1 inclusively, and yL is between 0 and N+1 inclusively, and
• wherein the prediction sample predSamplesLX[xL][yL] is derived by invoking a luma integer sample fetching process for one or more of conditions are true:
• xL is equal to 0, xL is equal to M+1, yL is equal to 0 and yL is equal to N+1, and the prediction sample predSamplesLX[xL][yL] is derived by invoking a luma sample 8-tap interpolation filtering process for all conditions are false.

A non-transitory computer-readable storage medium storing instructions that cause a processor to:

• determine, for a conversion between a current video block of a video and a bitstream of the video, a size of a prediction block corresponding to the current video block according to a rule;
• and
• perform the conversion based on the determining,
• wherein an affine merge mode is enabled for the current video block, and
• wherein the rule specifies that a first size of the prediction block is determined responsive to whether a prediction refinement using optical flow technique is enabled for the current video block, and
• wherein the current video block has a second size;
• wherein a second width and a second height of the second size of the current video block are indicated by M and N, respectively, and M and N are integers greater than or equal to 0,
• wherein a prediction sample of the prediction block is present as predSamplesLX[xL][yL],
• wherein xL is between 0 and M+1 inclusively, and yL is between 0 and N+1 inclusively,
• wherein the prediction sample predSamplesLX[xL][yL] is derived by invoking a luma integer sample fetching process for one or more of conditions are true:
• xL is equal to 0, xL is equal to M+1, yL is equal to 0 and yL is equal to N+1, and the prediction sample predSamplesLX[xL][yL] is derived by invoking a luma sample 8-tap interpolation filtering process for all conditions are false.

A non-transitory computer-readable recording medium storing a bitstream of a video which is generated by a method performed by a video processing apparatus, wherein the method comprises:

• determining, for a current video block of the video, a size of a prediction block corresponding to the current video block according to a rule;
• and
• generating the bitstream based on the determining,
• wherein an affine merge mode is enabled for the current video block, and
• wherein the rule specifies that a first size of the prediction block is determined responsive to whether a prediction refinement using optical flow technique is enabled for the current video block, and
• wherein the current video block has a second size;
• wherein a second width and a second height of the second size of the current video block are indicated by M and N, respectively, and M and N are integers greater than or equal to 0,
• wherein a prediction sample of the prediction block is present as predSamplesLX[xL][yL],
• wherein xL is between 0 and M+1 inclusively, and yL is between 0 and N+1 inclusively, and
• wherein the prediction sample predSamplesLX[xL][yL] is derived by invoking a luma integer sample fetching process for one or more of conditions are true:
• xL is equal to 0, xL is equal to M+1, yL is equal to 0 and yL is equal to N+1, and the prediction sample predSamplesLX[xL][yL] is derived by invoking a luma sample 8-tap interpolation filtering process for all conditions are false.