Description:-As Attached- , Claims:
1. Video decoder for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video (16) encoded thereinto in a manner according to which sub-pictures (24) into which the pictures are spatially sub-divided are encoded independently from each other, the video decoder comprising a DPB (48) and configured to

decode from the data stream, for each sub-picture of each picture of the video, a sub-picture related information (40) which, when assuming g a first state, identifies the respective sub-picture as being coded in a random access point manner and, when assuming a second state, identifies the respective sub-picture as not being coded in a random access point manner,

flush the DPB (48) responsive Co the sub-picture related information (40) assuming, for a predetermined picture of the video, the first state for all sub-pictures of the predetermined picture.

2. Video decoder of claim 1, configured to

perform the flushing after parsing or after decoding all sub-pictures of the predetermined picture.

3. Video decoder of any of claims 1 to 2, configured to, responsive to a signaling (44) in the data stream,

flush the DPB responsive to the sub-picture related information assuming, for a predetermined picture of the set of one or more pictures, the first state for a first sub-picture of the predetermined picture, being first in a sub-picture decoding order defined among the sub-pictures of the predetermined picture.

4. Video decoder of claim 3, configured to

perform the flushing the DPB responsive to the sub-picture related information assuming the first state for the first sub-picture after parsing or after decoding the first sub-picture of the currently decoded picture.

5. Video decoder of any of claims 1 to 4, configured to

mark (60) pictures in the DPB which are to remain in the DPB to serve as a prediction reference,

remove (62) pictures from the DPB which are not marked and which are not needed for future display,

decode (64) from the data stream, for a currently decoded picture, a reference picture buffer description (40’) for each sub-picture of the currently decoded picture, and

identify (66) the pictures in the DPB to be marked by, for the currently decoded picture, forming a union of sets of reference pictures indicated by the reference picture buffer descriptions for the sub-pictures of the currently decoded picture.

6. Video decoder for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video encoded thereinto in a manner according to which sub-pictures (24) into which the pictures are spatially sub-divided are encoded independently from each other, the video decoder comprising a DPB (48) and configured to

mark (60) pictures in the DPB which are to remain in the DPB to serve as a prediction reference,

remove (62) pictures from the DPB which are not marked and which are not needed for future display,

decode (64) from the data stream, for a currently decoded picture, a reference picture buffer description for each sub-picture of the currently decoded picture, and

identify (66) the pictures in the DPB to be marked by, for the currently decoded picture, forming a union of sets of reference pictures indicated by the reference picture buffer description for the sub-pictures of the currently decoded picture.

7. Video decoder of claim 6, configured to

perform the marking of the pictures in the DPB which are to remain in the DPB to serve as a prediction reference, and the removing of the pictures from the DPB which are not marked and which are not needed for future display, picture-wise cyclically, and

perform the removing after parsing or after decoding all sub-pictures of the currently decoded picture.

8. Video decoder of claim 6 or 7, configured to, responsive to a signaling (44) in the data stream,

identify the pictures in the DPB to be marked for the currently decoded picture depending on the reference picture buffer description for a first sub-picture of the currently decoded picture.

9. Video decoder of claim 8, configured to be responsive to a signaling in the data stream, identify the data stream as a non-conforming data stream in case of

a set of reference pictures indicated by the reference picture buffer description for any sub-picture of the currently decoded picture other than the first sub-picture, comprising a reference picture in the DPB not comprised by a set of reference pictures indicated by the reference picture buffer description for the first sub-picture of the currently decoded picture.

10. Video decoder for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video (16) encoded thereinto in a manner according to which the pictures are spatially sub-divided sub-pictures (24) in a manner constant over a sequence of pictures so that sub-videos (28) are coded into the sequence of pictures and in each picture of the sequence of pictures one sub-picture belongs to one of the sub-videos and wherein each sub-picture is encoded independently from sub-pictures of the pictures which belong to a different sub-video, the video decoder comprising a DPB (48) and configured to

perform emptying the DPB in units of sub-pictures.

11. Video decoder of claim 10, configured to

decode from the data stream, for a currently decoded picture, a reference picture buffer description for each sub-picture of the currently decoded picture, and

for each sub-video, use the reference picture buffer description (40’) for a sub-picture of the currently decoded picture which belongs to the respective sub-video to identify among sub-pictures in the DPB which belong to the respective sub-video ones ought to remain in the DPB.

12. Video decoder of claim 10, configured to

split the DPB (48) into sub-picture DPBs (481-6), which comprise, at least, one sub-picture DPB for each sub-picture into which the pictures are spatially sub-divided,

decode from the data stream, for a currently decoded picture, information (40’) a reference picture buffer description (40’) and a sub-picture identifier (80) for each sub-picture of the currently decoded picture, and

buffer the sub-pictures of the pictures in the DPB (48) by using each sub-picture DPB for buffering of sub-pictures of a predetermined sub-picture identifier associated with the respective sub-picture DPB,

for each sub-picture DPB, inspect whether the currently decoded picture contains a corresponding sub-picture the sub-picture identifier (80) of which equals the predetermined sub-picture identifier (A1 - 6) associated with the respective sub-picture DPB and, if so, use the reference picture buffer description (40’) for the corresponding sub-picture to identify sub-pictures in the respective sub-picture DPB ought to remain in the DPB.

13. Video decoder of claim 12, configured to

remove sub-pictures in the respective sub-picture DPB not belonging to the sub-pictures in the respective sub-picture DPB ought to remain in the DPB, after parsing or after decoding the corresponding sub-picture of the currently decoded picture and before decoding a next sub-picture of the currently decoded picture.

14. Video decoder of any of claims 12 to 13, configured to

if the currently decoded picture does not contains a corresponding sub-picture (24) the sub-picture identifier (80) of which equals the predetermined sub-picture identifier (A1to 6) associated with the respective sub-picture DPB, flush the respective sub-picture DPB and re-assign the respective sub-picture DPB to another predetermined sub-picture identifier of a non-corresponding sub-picture of the currently decoded picture, unequal to the predetermined sub-picture identifier of all sub-picture DPBs.

15. Video decoder of any of claims 10 to 14, configured to, responsive to a signaling (44) in the data stream,

perform the emptying the DPB in units of pictures.

16. Video decoder of any of claims 10 to 15, configured to, responsive to a signaling (44) in the data stream,

perform the emptying the DPB at each picture depending on a reference picture buffer description (40’) comprised by the data stream for a firstly encountered sub-picture of the respective picture.

17. Video decoder of any of claims 10 to 16, configured to, responsive to a signaling (44) in the data stream,

perform the emptying the DPB at each picture depending on a reference picture buffer description (40’) comprised by the data stream for a firstly encountered sub-picture of the respective picture after parsing or after decoding the first encountered sub-picture.

18. Video decoder for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video (16) encoded thereinto in a manner according to which the pictures are spatially sub-divided sub-pictures (24) in a manner constant over a sequence of pictures so that sub-videos (28) are coded into the sequence of pictures and in each picture of the sequence of pictures one sub-picture belongs to one of the sub-videos and wherein each sub-picture is encoded independently from sub-pictures of the pictures which belong to a different sub-video, the video decoder configured to

decode from the data stream, for each sub-picture of each picture of the video, a sub-picture related information (40) which, when assuming a first state, identifies the respective sub-picture as being coded in a random access point manner and, when assuming a second state, identifies the respective sub-picture as not being coded in the random access point manner,

decode from the data stream, for each sub-picture of each picture of the video, a first POC value (90),

manage, for each sub-video, one or more inter-sub-picture POC compensation parameters (92) which parametrize a parametrizable function (94) which, when applied to the first POC value decoded for a sub-picture of a currently decoded picture belonging to the respective sub-video, yields a final POC value (96) so that the final POC values of the sub-pictures of the currently decoded picture are equal to each other.

19. Video decoder of claim 18, wherein for each sub-video, the one or more inter-sub-picture POC compensation parameters (92) comprise a POC compensation offset which, by the parametrizable function (94), offsets the first POC value decoded for sub-pictures of the respective sub-video, and the video decoder is configured to

reset the POC compensation offset for a predetermined sub-video of a predetermined picture upon encountering a first sub-picture belonging to the predetermined sub-video for which the sub-picture related information (40) assumes the first state, using a difference (98) between a first POC value (90) decoded for the first sub-picture and a final POC value (96) of a further sub-picture for which the sub-picture related information (40) assumes the second state and which is comprised by a picture, which comprises the first sub-picture.

20. Video decoder of claim 18 or 19, wherein for each sub-video, the one or more inter-sub-picture POC compensation parameters (92) comprise a POC compensation offset which, by the parametrizable function (94), offsets the first POC value decoded for sub-pictures of the respective sub-video, and the video decoder is configured to

if the sub-picture related information (40) assumes the first state for all sub-pictures of a predetermined picture, set the POC compensation offset for all sub-videos to which the sub-pictures of the predetermined picture belong equally to a default POC value.

21. Video decoder of any of claims 18 to 20, wherein for each sub-video, the one or more inter-sub-picture POC compensation parameters (92) comprise a POC step size scaling factor which, by the parametrizable function (94), scales the final POC value.

22. Video decoder of claim 21, configured to derive, for each sub-video, the POC step size scaling factor from the data stream.

23. Video decoder of any of claims 18 to 22, configured to

use the final POC value for picture output and/or referencing of buffered reference pictures.

24. Video decoder of any of claims 18 to 23, configured to

use, for each sub-video, the POC values decoded for the sub-pictures which belong to the respective sub-video for scaling motion vector prediction candidates used in decoding the sub-pictures which belong to the respective sub-video.

25. Video decoder of any of claims 18 to 24, wherein for each sub-video, the one or more inter-sub-picture POC compensation parameters (92) comprise a POC step size scaling factor which, by the parametrizable function (94), scales the final POC value, and the video decoder is configured to

perform, for each sub-video, motion vector prediction candidate scaling used in decoding the sub-pictures which belong to the respective sub-video, by inversely scaling the final POC values so as to obtain inversely scaled POC values and use the inversely scaled POC values for scaling motion vector prediction candidates used in decoding the sub-pictures which belong to the respective sub-video.

26. Video decoder for decoding a video of pictures from a data stream, the video decoder configured to

log, for each of reference pictures of the video, subdivision of the respective reference picture into a refreshed picture region (118) and non-refreshed picture region (116),

decode from the data stream (10) an information on a refresh renewal region (110) within a currently decoded picture, and decode the refresh renewal region (110) from the data stream using intra prediction,

determine a first region (120) of the currently decoded picture, disjoint to the refresh renewal region (110) and to be coded independent from the non-refreshed picture region (118) of the reference pictures,

decode the first region (120) from the data stream by deriving, based on coding parameters signaled in the data stream for the first region (120), a prediction for the first region (120) from the reference pictures in a manner dependent on a location of the refreshed picture region (118) of the reference pictures so that the prediction is independent from the non-refreshed picture region (116) of the reference pictures, and

determine a refreshed picture region of the currently decoded picture as a union of the first region and the refresh renewal region.

27. Video decoder of claim 26, wherein the first region abuts the refresh renewal region.

28. Video decoder of any of claims 26 to 27, configured to determine the first region of the currently decoded picture to be a region of the currently decoded picture which is co-located to a refreshed picture region of an immediately previously decoded picture.

29. Video decoder of any of claims 26 to 28, wherein the coding parameters comprise motion vectors and/or motion vector indices into motion vector candidate lists.

30. Video decoder of any of claims 26 to 29, configured to perform the deriving the prediction by performing one or more of

clipping motion vectors so as to not reference portions of the reference pictures overlaying the non-refreshed picture region (116) of the reference pictures,

padding portions of the reference pictures referenced by motion vectors and overlaying the non-refreshed picture region (116) of the reference pictures from the refreshed picture region (118),

excluding motion vectors of the non-refreshed picture region (116) of the reference pictures from motion vector predictor candidate list into which motion indices signaled for the first region point.

31. Video decoder of any of claims 26 to 30, configured to

decode a second region (116) which is disjoint to the first region and the refresh renewal region by predicting the second region from the refreshed picture region as well as the non-refreshed picture region of the reference pictures.

32. Video decoder of any of claims 26 to 31, configured to

decode, for a predetermined picture, a gradual decoding refresh signalization and, responsive thereto,

start the logging the subdivision of the respective reference picture into the refreshed picture region (118) and the non-refreshed picture region (116) by letting each reference picture be completely composed of the non-refreshed picture region (116),

decode a refresh renewal region (110) of the predetermined picture from the data stream using intra prediction, and

decode a second region of the predetermined picture by predicting the second region from the non-refreshed picture region of the reference pictures, and

determine a refreshed picture region of the predetermined picture to be the refresh renewal region of the predetermined picture.

33. Video decoder of claim 32, configured to

derive from the gradual decoding refresh signalization an information on a refresh renewal region (110) of a sequence of pictures starting at the predetermined picture wherein the refresh renewal regions (110) of the sequence of pictures are mutually non-overlapping and completely cover a picture area of the video.

34. Video decoder of claim 32 or 33, configured to

derive information on a refresh renewal region (110) of a sequence of pictures starting at the predetermined picture from a first parameter set of the data stream having larger scope than a second parameter set comprising the gradual decoding refresh signalization.

35. Video decoder of claim 34, wherein the first parameter set is a sequence parameter set and the second parameter set is a picture parameter set.

36. Video composition apparatus configured to compose a composed data stream out of a plurality of sub-streams,

the composed video data stream having a composed video of composed pictures encoded thereinto which are subdivided into sub-pictures,

wherein the sub-streams have encoded thereinto the sub-pictures of the composed pictures of the composed video in a manner so that for each composed picture, each sub-picture of the respective composed picture, is encoded into one of the sub-streams independent from other sub-pictures of the respective composed video which are encoded into the other of the sub-streams,

the video composition apparatus configured to

compose the composed data stream by putting together the sub-streams with, for each sub-stream, taking over from the respective sub-stream into the composed data stream, for each sub-picture encoded into the respective sub-stream, information on a set of reference pictures for the respective sub-picture.

37. Video composition apparatus configured to compose a composed data stream (10) out of a plurality of sub-streams (32),

the composed video data stream having a composed video (16) of composed pictures (18) encoded thereinto which are subdivided into sub-pictures (24),

wherein the sub-streams have encoded thereinto the sub-pictures of the composed pictures of the composed video in a manner so that for each composed picture, each sub-picture of the respective composed picture, is encoded into one of the sub-streams independent from other sub-pictures of the respective composed video which are encoded into the other of the sub-streams,

the video composition apparatus configured to

compose the composed data stream by putting together the sub-streams with, for each sub-stream, taking over from the respective sub-stream into the composed data stream, for each sub-picture encoded into the respective sub-stream, a POC value (90).

38. Video composition of claim 37, configured to

in composing the composed data stream by putting together the sub-streams, for each sub-stream, write into the composed data stream (10), one or more inter-sub-picture POC compensation parameters (92) for parametrizing a parametrizable function (94) so that same, when applied to the POC values of sub-pictures encoded into the respective sub-stream, yields a final POC value (96) which is equal for sub-pictures in one composed picture.

39. Data stream having a video (16) of pictures (18) encoded thereinto in a manner according to which the pictures are spatially sub-divided sub-pictures (24) in a manner constant over a sequence of pictures so that sub-videos (28) are coded into the sequence of pictures and in each picture of the sequence of pictures one sub-picture belongs to one of the sub-videos and wherein each sub-picture is encoded independently from sub-pictures of the pictures which belong to a different sub-video, the data stream comprising

for each sub-picture of each picture of the video, a sub-picture related information (40) which, when assuming a first state, identifies the respective sub-picture as being coded in a random access point manner and, when assuming a second state, identifies the respective sub-picture as not being coded in the random access point manner, and

one or more inter-sub-picture POC compensation parameters (92) for parametrizing a parametrizable function (94) so that same, when applied to the POC values of sub-pictures encoded into the respective sub-stream, yields a final POC value (96) which is equal for sub-pictures in one composed picture.

?40. Video encoder for encoding a video of pictures into a data stream, the video encoder configured to

log, for each of reference pictures of the video, subdivision of the respective reference picture into a refreshed picture region (118) and non-refreshed picture region (116),

encode into the data stream (10) an information on a refresh renewal region (110) within a currently decoded picture, and encode the refresh renewal region (110) into the data stream using intra prediction,

determine a first region (120) of the currently decoded picture, disjoint to the refresh renewal region (110) and to be coded independent from the non-refreshed picture region (118) of the reference pictures,

encode the first region (120) into the data stream by deriving, based on coding parameters signaled in the data stream for the first region (120), a prediction for the first region (120) from the reference pictures in a manner dependent on a location of the refreshed picture region (118) of the reference pictures so that the prediction is independent from the non-refreshed picture region (116) of the reference pictures, and

determine a refreshed picture region of the currently decoded picture as a union of the first region and the refresh renewal region.

41. Method for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video (16) encoded thereinto in a manner according to which sub-pictures (24) into which the pictures are spatially sub-divided are encoded independently from each other, the method comprising

decode from the data stream, for each sub-picture of each picture of the video, a sub-picture related information (40) which, when assuming a first state, identifies the respective sub-picture as being coded in a random access point manner and, when assuming a second state, identifies the respective sub-picture as not being coded in a random access point manner,

flush a DPB (48) used in the decoding the video responsive to the sub-picture related information (40) assuming, for a predetermined picture of the video, the first state for all sub-pictures of the predetermined picture.

42. Method for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video encoded thereinto in a manner according to which sub-pictures (24) into which the pictures are spatially sub-divided are encoded independently from each other, the method comprising

mark (60) pictures in a DPB used in the decoding the video which are to remain in the DPB to serve as a prediction reference,

remove (62) pictures from the DPB which are not marked and which are not needed for future display,

decode (64) from the data stream, for a currently decoded picture, a reference picture buffer description for each sub-picture of the currently decoded picture, and

identify (66) the pictures in the DPB to be marked by, for the currently decoded picture, forming a union of sets of reference pictures indicated by the reference picture buffer description for the sub-pictures of the currently decoded picture.

43. Method for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video (16) encoded thereinto in a manner according to which the pictures are spatially sub-divided sub-pictures (24) in a manner constant over a sequence of pictures so that sub-videos (28) are coded into the sequence of pictures and in each picture of the sequence of pictures one sub-picture belongs to one of the sub-videos and wherein each sub-picture is encoded independently from sub-pictures of the pictures which belong to a different sub-video, the method comprising

perform emptying a DPB used in the decoding the video in units of sub-pictures.

44. Method for decoding a video (16) of pictures (18) from a data stream (10), the data stream having the video (16) encoded thereinto in a manner according to which the pictures are spatially sub-divided sub-pictures (24) in a manner constant over a sequence of pictures so that sub-videos (28) are coded into the sequence of pictures and in each picture of the sequence of pictures one sub-picture belongs to one of the sub-videos and wherein each sub-picture is encoded independently from sub-pictures of the pictures which belong to a different sub-video, the method comprising

decode from the data stream, for each sub-picture of each picture of the video, a sub-picture related information (40) which, when assuming a first state, identifies the respective sub-picture as being coded in a random access point manner and, when assuming a second state, identifies the respective sub-picture as not being coded in the random access point manner,

decode from the data stream, for each sub-picture of each picture of the video, a first POC value (90),

manage, for each sub-video, one or more inter-sub-picture POC compensation parameters (92) which parametrize a parametrizable function (94) which, when applied to the first POC value decoded for a sub-picture of a currently decoded picture belonging to the respective sub-video, yields a final POC value (96) so that the final POC values of the sub-pictures of the currently decoded picture are equal to each other.

45. Method for decoding a video of pictures from a data stream, the method comprising

log, for each of reference pictures of the video, subdivision of the respective reference picture into a refreshed picture region (118) and non-refreshed picture region (116),

decode from the data stream (10) an information on a refresh renewal region (110) within a currently decoded picture, and decode the refresh renewal region (110) from the data stream using intra prediction,

determine a first region (120) of the currently decoded picture, disjoint to the refresh renewal region (110) and to be decoded independent from the non-refreshed picture region (118) of the reference pictures,

decode the first region (120) from the data stream by deriving, based on coding parameters signaled in the data stream for the first region (120), a prediction for the first region (120) from the reference pictures in a manner dependent on a location of the refreshed picture region (118) of the reference pictures so that the prediction is independent from the non-refreshed picture region (116) of the reference pictures, and

determine a refreshed picture region of the currently decoded picture as a union of the first region and the refresh renewal region.

46. Method for encoding a video of pictures into a data stream, method comprising

log, for each of reference pictures of the video, subdivision of the respective reference picture into a refreshed picture region (118) and non-refreshed picture region (116),

encode into the data stream (10) an information on a refresh renewal region (110) within a currently decoded picture, and encode the refresh renewal region (110) into the data stream using intra prediction,

determine a first region (120) of the currently decoded picture, disjoint to the refresh renewal region (110) and to be coded independent from the non-refreshed picture region (118) of the reference pictures,

encode the first region (120) into the data stream by deriving, based on coding parameters signaled in the data stream for the first region (120), a prediction for the first region (120) from the reference pictures in a manner dependent on a location of the refreshed picture region (118) of the reference pictures so that the prediction is independent from the non-refreshed picture region (116) of the reference pictures, and

determine a refreshed picture region of the currently decoded picture as a union of the first region and the refresh renewal region.

47. Data stream encoded by a method according to claim 46.