Can anyone please comment on this seeming discrepancy between SMPTE 381M-2005 (
Mapping MPEG Streams into MXF Containers) and SMPTE RP224.9-2008 (
SMPTE Labels Register)?
381M Table 5 defines the SMPTE label for MPEG essence container identification. Byte 15 of this UL is defined as the value of bits 6..0 of the MPEG stream ID, as defined by ISO 13818-1. The explanation to the table (Sec. 5) clarifies that
if Byte 14 is set to 08h or 09h, the value of Byte 15 shall be set to the same value as Byte 14. A value of 08h or 09h in Byte 14 indicates MPEG PS or MPEG TS, respectively, in which case the notion of stream ID is not defined, so I suppose that in this case the value in Byte 15 is set to an arbitrary default.
On the other hand, RP224 gives a list of such ULs with Byte 14 being 08h or 09h and Byte 15 different from Byte 14 (In the spreadsheet rows 1978-2118). The description for those ULs is, for example, "Identifier for MXF-GC, Clip-wrapped MPEG-TS with AudioStream-12 SID value" (for 06.0E.2B.34.04.01.01.02.0D.01.03.01.02.09.4C.02), which does not make much sense since the same MPEG-TS usually contains several elementary streams with different stream IDs.
I wonder how MXF applications should handle Byte 15 if Byte 14 is 08h or 09h. In the spirit of interoperability I suggest that:
- MXF encoders should use the definition of 381M and set Byte 15 to be the same as Byte 14, first because it is a standard while RP224 is only a recommended practice, and second because it makes more sense.
- MXF decoders should ignore the value of Byte 15 unless they act as a consistency checkers, in which case they should flag any deviation from 381M (Byte 15 different from Byte 14).
- Applications which pass or process MXF files should leave Byte 15 untouched, unless they add or modify a UL for MPEG essence container identification, in which case they should set the UL according to 381M.
Razi
