Symmetric motion vector difference coding

1. A device for video decoding, comprising:

a processor configured to:

determine that bi-directional optical flow (BDOF) is enabled for a picture comprising a video block;

obtain a symmetric motion vector difference (SMVD) indication associated with the video block, wherein the SMVD indication indicates whether SMVD is used in a determination of a motion vector (MV) for the video block;

determine to bypass BDOF for the video block based on the SMVD indication indicating that SMVD is used for the video block; and

decode the picture based on the determination to bypass BDOF for the video block.

2. The device of claim 1 , wherein SMVD uses a motion vector difference (MVD) for the video block, and the MVD indicates a difference between a motion vector predictor (MVP) for the video block and the MV for the video block.

3. The device of claim 2 , wherein the MVP for the video block is determined based on an MV of a spatial neighboring block of the video block or an MV of a temporal neighboring block of the video block.

4. The device of claim 1 , wherein, based on a determination that the SMVD indication indicates SMVD is used in the determination of the MV for the video block, the processor is configured to:

obtain a first motion vector difference (MVD) associated with a reference picture of the video block; and

determine a second MVD based on the first MVD associated with the reference picture, wherein the second MVD is symmetric to the first MVD.

5. The device of claim 1 , wherein, based on a determination that the SMVD indication indicates SMVD is used in the determination of the MV for the video block, the processor is configured to:

parse a first motion vector difference (MVD) associated with a first reference picture list in video data; and

determine a second MVD associated with a second reference picture list based on the first MVD, wherein the first MVD and the second MVD are symmetric to each other.

6. The device of claim 1 , wherein the processor is configured to receive a sequence-level SMVD indication indicating whether SMVD is enabled for a sequence of pictures, wherein the picture comprising the video block is in the sequence, and wherein, based on a determination that SMVD is enabled for the sequence, the processor is configured to obtain the SMVD indication associated with the video block.

7. The device of claim 1 , wherein decoding the picture based on the determination to bypass BDOF for the video block comprises reconstructing the video block without performing BDOF based on the determination to bypass BDOF.

8. The device of claim 1 , wherein the determination to bypass BDOF comprises disabling BDOF for the video block.

9. A method for video decoding, comprising:

determining that bi-directional optical flow (BDOF) is enabled for a picture comprising a video block;

obtaining a symmetric motion vector difference (SMVD) indication associated with the video block, wherein the SMVD indication indicates whether SMVD is used in a determination of a motion vector (MV) for the video block;

determining to bypass BDOF for the video block based on the SMVD indication indicating that SMVD is used for the video block; and

decoding the picture based on the determination to bypass BDOF for the video block.

10. The method of claim 9 , wherein SMVD uses a motion vector difference (MVD) for the video block, and the MVD indicates a difference between a motion vector predictor (MVP) for the video block and the MV for the video block.

11. The method of claim 10 , wherein the MVP for the video block is determined based on an MV of a spatial neighboring block of the video block or an MV of a temporal neighboring block of the video block.

12. The method of claim 9 , wherein, based on a determination that the SMVD indication indicates SMVD is used for the determination of the MV for the video block, the method comprises:

obtaining a first motion vector difference (MVD) associated with a reference picture of the video block; and

determining a second MVD based on the first MVD associated with the reference picture, wherein the second MVD is symmetric to the first MVD.

13. The method of claim 9 , wherein, based on a determination that the SMVD indication indicates SMVD is used in the determination of the MV for the video block, the method comprises:

parsing a first motion vector difference (MVD) associated with a first reference picture list in video data; and

determining a second MVD associated with a second reference picture list based on the first MVD, wherein the first MVD and the second MVD are symmetric to each other.

14. A device for video encoding, comprising:

a processor configured to:

determine that bi-directional optical flow (BDOF) is enabled for a picture comprising a video block;

determine whether symmetric motion vector difference (SMVD) is enabled for the video block;

determine to bypass BDOF for the video block based on a determination that SMVD is enabled for the video block; and

encode the picture based on the determination to bypass BDOF for the video block.

15. The device of claim 14 , wherein, based on a determination that SMVD is enabled for the video block, the processor is configured to:

obtain a first motion vector difference (MVD) associated with a first reference picture list;

determine a second MVD that is symmetric to the first MVD associated with the first reference picture list; and

include an indication of the first MVD associated with the first reference picture list in video data, wherein the video data does not include an indication of the second MVD.

16. The device of claim 14 , wherein a motion vector difference (MVD) for the video block indicates a difference between a motion vector predictor (MVP) for the video block and the MV for the video block, and wherein the MVP for the video block is determined based on an MV of a spatial neighboring block of the video block or an MV of a temporal neighboring block of the video block.

17. The device of claim 14 , wherein the determination to bypass BDOF comprises disabling BDOF for the video block.

18. A method for video encoding, comprising:

determining that bi-directional optical flow (BDOF) is enabled for a picture comprising a video block;

determining whether symmetric motion vector difference (SMVD) is enabled for the video block;

determining to bypass BDOF for the video block based on a determination that SMVD is enabled for the video block; and

encoding the picture based on the determination of whether to bypass BDOF for the video block.

19. The method of claim 18 , wherein, based on a determination that SMVD is enabled for the video block, the method comprises:

obtaining a first motion vector difference (MVD) associated with a first reference picture list;

determining a second MVD that is symmetric to the first MVD associated with the first reference picture list; and

including an indication of the first MVD associated with the first reference picture list in video data, wherein the video data does not include an indication of the second MVD.

20. The method of claim 18 , wherein a motion vector difference (MVD) for the video block indicates a difference between a motion vector predictor (MVP) for the video block and the MV for the video block, and wherein the MVP for the video block is determined based on an MV of a spatial neighboring block of the video block or an MV of a temporal neighboring block of the video block.