Method and apparatus for video coding
An apparatus for video decoding includes processing circuitry configured to determine an affine model for a current block coded with an interweaved affine mode. Based on the affine model, a first prediction block corresponding to a first pattern for partitioning the current block into first sub-blocks and a second prediction block corresponding to a second pattern for partitioning the current block into second sub-blocks can be generated. The first and second prediction blocks include interpolated samples having an intermediate bit-depth larger than an input bit-depth of the current block. Co-located first and samples in the first and second prediction blocks with a precision corresponding to the intermediate bit-depth are weighted averaged to obtain averaged samples. The averaged samples are rounded to the input bit-depth to obtain corresponding third samples in a final prediction block of the current block.
1. A method of video decoding at a video decoder, comprising:
determining an affine model for a current block coded with an interweaved affine mode;
generating, based on the affine model, a first prediction block corresponding to a first pattern for partitioning the current block into first sub-blocks and a second prediction block corresponding to a second pattern for partitioning the current block into second sub-blocks, wherein the first prediction block includes first samples and the second prediction block includes second samples, the first samples and the second samples have a precision of an intermediate bit-depth larger than an input bit-depth of the current block, the second pattern is obtained by shifting the first pattern, and the second sub-blocks include whole sub-blocks and fractional sub-blocks;
generating a third prediction block based on the first and second prediction blocks, wherein the first samples in the first prediction block and the corresponding second samples in the second prediction block each with the precision of the intermediate bit-depth larger than the input bit-depth of the current block are combined by performing a weighted average operation over a region corresponding to the whole sub-blocks of the second sub-blocks to obtain averaged samples, the first samples and the corresponding second samples are predicted from a same reference picture, and the averaged samples are rounded to the input bit-depth to obtain corresponding third samples in the third prediction block; and
when interweaved affine prediction is applied to the region corresponding to the whole sub-blocks of the second sub-blocks and not applied to a region corresponding to the fractional sub-blocks of the second sub-blocks, disabling deblocking for the region corresponding to the whole sub-blocks of the second blocks, and performing deblocking for the region corresponding to the fractional sub-blocks of the second sub-blocks.
2. The method of claim 1 , wherein the weighted average operation includes adding a rounding offset to a weighted sum of the first sample and the corresponding second sample.
3. The method of claim 1 , wherein
when the first and second samples are located within the region corresponding to the fractional sub-blocks of the second sub-blocks, then the second samples are given a zero weight in the weighted average operation.
4. The method of claim 3 , wherein when the first and second samples are located within the region corresponding to the whole sub-blocks of the second sub-blocks, then the first and second samples are given an equal weight in the weighted average operation.
5. The method of claim 1 , wherein the third samples are each constrained to be within a range from 0 to a maximum possible value corresponding to the input bit-depth.
6. The method of claim 1 , wherein deblocking is disabled for the current block.
7. The method of claim 1 , wherein
deblocking is not applied to the region corresponding to the fractional sub-blocks of the second sub-blocks.
8. An apparatus of video decoding, comprising circuitry configured to:
determine an affine model for a current block coded with an interweaved affine mode;
generate, based on the affine model, a first prediction block corresponding to a first pattern for partitioning the current block into first sub-blocks and a second prediction block corresponding to a second pattern for partitioning the current block into second sub-blocks, wherein the first prediction block includes first samples and the second prediction block includes second samples, the first samples and the second samples have a precision of an intermediate bit-depth larger than an input bit-depth of the current block, the second pattern is obtained by shifting the first pattern, and the second sub-blocks include whole sub-blocks and fractional sub-blocks;
generate a third prediction block based on the first and second prediction blocks, wherein the first samples in the first prediction block and the corresponding second samples in the second prediction block each with the precision of the intermediate bit-depth larger than the input bit-depth of the current block are combined by performing a weighted average operation over a region corresponding to the whole sub-blocks of the second sub-blocks to obtain averaged samples, the first samples and the corresponding second samples are predicted from a same reference picture, and the averaged samples are rounded to the input bit-depth to obtain corresponding third samples in the third prediction block; and
when interweaved affine prediction is applied to the region corresponding to the whole sub-blocks of the second sub-blocks and not applied to a region corresponding to the fractional sub-blocks of the second sub-blocks, disable deblocking for the region corresponding to the whole sub-blocks of the second blocks, and perform deblocking for the region corresponding to the fractional sub-blocks of the second sub-blocks.
9. The apparatus of claim 8 , wherein the weighted average operation includes adding a rounding offset to a weighted sum of the first sample and the corresponding second sample.
10. The apparatus of claim 8 , wherein
when the first and second samples are located within the region corresponding to the fractional sub-blocks of the second sub-blocks, then the second samples are given a zero weight in the weighted average operation.
11. The apparatus of claim 10 , wherein when the first and second samples are located within the region corresponding to the whole sub-blocks of the second sub-blocks, then the first and second samples are given an equal weight in the weighted average operation.
12. The apparatus of claim 8 , wherein the third samples are each constrained to be within a range from 0 to a maximum possible value corresponding to the input bit-depth.
13. The apparatus of claim 8 , wherein deblocking is disabled for the current block.
14. The apparatus of claim 8 , wherein
deblocking is not applied to the region corresponding to the fractional sub-blocks of the second sub-blocks.
15. A non-transitory computer-readable medium storing instructions that, when executed by a computer for video decoding, cause the computer to perform a method of video decoding, the method comprising:
determining an affine model for a current block coded with an interweaved affine mode;
generating, based on the affine model, a first prediction block corresponding to a first pattern for partitioning the current block into first sub-blocks and a second prediction block corresponding to a second pattern for partitioning the current block into second sub-blocks, wherein the first prediction block includes first samples and the second prediction block includes second samples, the first samples and the second samples have a precision of an intermediate bit-depth larger than an input bit-depth of the current block, the second pattern is obtained by shifting the first pattern, and the second sub-blocks include whole sub-blocks and fractional sub-blocks;
generating a third prediction block based on the first and second prediction blocks, wherein the first samples in the first prediction block and the corresponding second samples in the second prediction block each with the precision of the intermediate bit-depth larger than the input bit-depth of the current block are combined by performing a weighted average operation over a region corresponding to the whole sub-blocks of the second sub-blocks to obtain averaged samples, the first samples and the corresponding second samples are predicted from a same reference picture, and the averaged samples are rounded to the input bit-depth to obtain corresponding third samples in the third prediction block; and
when interweaved affine prediction is applied to the region corresponding to the whole sub-blocks of the second sub-blocks and not applied to a region corresponding to the fractional sub-blocks of the second sub-blocks, disabling deblocking for the region corresponding to the whole sub-blocks of the second blocks, and performing deblocking for the region corresponding to the fractional sub-blocks of the second sub-blocks.
16. The non-transitory computer-readable medium of claim 15 , wherein
when the first and second samples are located within the region corresponding to the fractional sub-blocks of the second sub-blocks, then the second samples are given a zero weight in the weighted average operation.
17. The non-transitory computer-readable medium of claim 16 , wherein when the first and second samples are located within the region corresponding to the whole sub-blocks of the second sub-blocks, then the first and second samples are given an equal weight in the weighted average operation.
18. The non-transitory computer-readable medium of claim 15 , wherein the third samples are each constrained to be within a range from 0 to a maximum possible value corresponding to the input bit-depth.
19. The non-transitory computer-readable medium of claim 15 , wherein deblocking is disabled for the current block.
20. The non-transitory computer-readable medium of claim 15 , wherein
deblocking is not applied to the region corresponding to the fractional sub-blocks of the second sub-blocks.