For projection formats composed of a plurality of faces, no matter what kind of compact frame packing arrangement is used, discontinuities appear between two or more adjacent faces in the frame packed picture. For example, considering the 3×2 frame packing configuration depicted in , the top and bottom halves of the frame packed picture are discontinuous in the 3D geometry. If in-loop filtering operations are performed across this discontinuity, face seam artifacts may become visible in the reconstructed video.
Figure 56 – An example of 3x2 frame packing To alleviate face seam artifacts, in-loop filtering operations cloud be disabled across discontinuities in the frame-packed picture. This technique improved the subjective quality of 360-degree video. In VVC, vertical and/or horizontal virtual boundaries, across which the in-loop filtering operations are disabled, are introduced and the positions of those boundaries are signalled in either SPS or Picture Header. Compared to using two tiles, one for each set of continuous faces, and to disable in-loop filtering operations across tiles, this technique is more flexible as it does not require the face size to be a multiple of the CTU size. The signaling is designed to be general purpose, and applicable to other non-360-degree video use cases. The maximum number of vertical virtual boundaries is 3 and the maximum number of horizontal virtual boundaries is 3. The distance between two virtual boundaries is greater than or equal to the CTU size and the virtual boundary granularity is 8 luma samples.
The virtual boundary could also be used in Gradual Decoding Refresh (GDR) or Progressive Intra Refresh (PIR) which is a technique to limit the large bitrate variations between Intra (I) frames and Inter (P or B) frames while maintaining the same random access period. For this application, the locations of virtual boundaries are signalled in Picture Header since the positions are changed frame by frame.
