JVET-Q0276 AHG9: On EOB NAL unit out of band [S. Wenger, B. Choi (Tencent)]
Discussed Saturday 11 January at 0900 (chaired by GJS).
The contribution proposes a) the option to receive the End of Bitstream (EOB-) NAL unit out of band, b) when receiving the EOB NAL unit out of band to cease decoding (by ignoring incoming NAL units) and start decoding only with a well-formed new bitstream. It is argued that such language is required for certain application scenarios.
The contribution proposes to add a statement that says “If an end of bitstream NAL unit becomes available, the decoder shall ignore all future NAL units up to the beginning of a new video bitstream.”
It was commented that the spec is written such that anything in a channel after the EOB is outside the scope of our standard; such data is part of a system rather than part of our bitstream. The bitstream has ended and there is no additional data, as far as our standard is concerned. There are no future NAL units in the bitstream.
It was further commented that multibitstream handling may be a practical need of system design but is outside the “spec domain” of the video coding standard.
No action was taken on this.