TMVP derivation

1. A video processing method, comprising:

deriving multiple temporal motion vector prediction (TMVP) candidates for a video block in a current picture based on multiple blocks associated with a second block in one or more pictures that are temporally before or after the current picture, and the second block is collocated with the video block, wherein the second block has a same size as the video block, and wherein a relative position of the second block to a top-left corner of a second picture of the one or more pictures is same as that of the video block to a top-left corner of the current picture;

adding the multiple TMVP candidates to a motion candidate list associated with the video block; and

performing a conversion between the video block and a bitstream,

wherein the multiple blocks are located inside a coding tree block covering the second block, and

wherein the second block is identified by a non-zero motion vector that is derived from a spatial merge candidate of the second block,

the second block is identified by a motion vector derived from a merge candidate of the video block, or

the second block is identified by a motion vector derived from motion information of a spatial neighboring block of the video block.

2. The method of claim 1 , wherein the one or more pictures includes a single picture co-located with the current picture.

3. The method of claim 1 , wherein the multiple blocks are located inside the second block.

4. The method of claim 1 , wherein the multiple blocks are located outside the second block.

5. The method of claim 1 , wherein the non-zero motion vector is derived by scaling a motion vector based on one of the one or more pictures.

6. The method of claim 1 , further comprising:

adjusting one of the one or more pictures to be a reference picture associated with a spatial merge candidate.

7. The method of claim 1 , further comprising:

comparing a new TMVP candidate against all existing TMVP candidates;

determining that the new TMVP candidate is identical to an existing TMVP candidate; and

refraining from adding the new TMVP candidate to the multiple TMVP candidates.

8. The method of claim 1 , further comprising:

comparing a new TMVP candidate against a subset of existing TMVP candidates;

determining that the new TMVP candidate is identical to an existing TMVP candidate; and

refraining from adding the new TMVP candidate to the multiple TMVP candidate.

9. A video encoding apparatus comprising a processor configured to implement a method of video processing, comprising:

deriving multiple temporal motion vector prediction (TMVP) candidates for a video block in a current picture based on multiple blocks associated with a second block in one or more pictures that are temporally before or after the current picture, and the second block is temporally collocated with the video block, wherein the second block has a same size as the video block, and wherein a relative position of the second block to a top-left corner of a second picture of the one or more pictures is same as that of the video block to a top-left corner of the current picture;

adding the multiple TMVP candidates to a motion candidate list associated with the video block; and

performing a conversion between the video block and a bitstream,

wherein the multiple blocks are located inside a coding tree block covering the second block, and

wherein the second block is identified by a non-zero motion vector that is derived from a spatial merge candidate of the second block,

the second block is identified by a motion vector derived from a merge candidate of the video block, or

the second block is identified by a motion vector derived from motion information of a spatial neighboring block of the video block.

10. The apparatus of claim 9 , wherein the one or more pictures includes a single picture co-located with the current picture.

11. The apparatus of claim 9 , wherein the multiple blocks are located inside the second block.

12. The apparatus of claim 9 , wherein the multiple blocks are located outside the second block.

13. The apparatus of claim 9 , wherein the non-zero motion vector is derived by scaling a motion vector based on one of the one or more pictures.

14. A video decoding apparatus comprising a processor configured to implement a method of video processing, comprising:

deriving multiple temporal motion vector prediction (TMVP) candidates for a video block in a current picture based on multiple blocks associated with a second block in one or more pictures that are temporally before or after the current picture, and the second block is collocated with the video block, wherein the second block has a same size as the video block, and wherein a relative position of the second block to a top-left corner of a second picture of the one or more pictures is same as that of the video block to a top-left corner of the current picture;

adding the multiple TMVP candidates to a motion candidate list associated with the video block; and

performing a conversion between the video block and a bitstream,

wherein the multiple blocks are located inside a coding tree block covering the second block, and

wherein the second block is identified by a non-zero motion vector that is derived from a spatial merge candidate of the second block,

the second block is identified by a motion vector derived from a merge candidate of the video block, or

the second block is identified by a motion vector derived from motion information of a spatial neighboring block of the video block.

15. The apparatus of claim 14 , wherein the one or more pictures includes a single picture co-located with the current picture.

16. The apparatus of claim 14 , wherein the multiple blocks are located inside the second block.

17. The apparatus of claim 14 , wherein the multiple blocks are located outside the second block.

18. A non-transitory computer-readable program medium having code stored thereupon, the code comprising instructions that, when executed by a processor, causing the processor to implement a method of video processing, comprising:

deriving multiple temporal motion vector prediction (TMVP) candidates for a video block in a current picture based on multiple blocks associated with a second block in one or more pictures that are temporally before or after the current picture, and the second block is collocated with the video block, wherein the second block has a same size as the video block, and wherein a relative position of the second block to a top-left corner of a second picture of the one or more pictures is same as that of the video block to a top-left corner of the current picture;

adding the multiple TMVP candidates to a motion candidate list associated with the video block; and

performing a conversion between the video block and a bitstream,

wherein the multiple blocks are located inside a coding tree block covering the second block, and

wherein the second block is identified by a non-zero motion vector that is derived from a spatial merge candidate of the second block,

the second block is identified by a motion vector derived from a merge candidate of the video block, or

the second block is identified by a motion vector derived from motion information of a spatial neighboring block of the video block.

19. The non-transitory computer-readable program medium of claim 18 , wherein the one or more pictures includes a single picture co-located with the current picture.

20. The non-transitory computer-readable program medium of claim 18 , wherein the multiple blocks are located inside the second block.