[XviD-devel] [Fwd: [M4IF Technotes] resync_marker emulation]
Chandra Sekhar Reddy
gchandra at tataelxsi.co.in
Sat Jul 26 21:02:04 CEST 2003
Hi,
One of my team-mate raised this issue on M4IF-technotes, but got no responces.
What is XVID codec doing to take care of this situation?
Regards,
Chandra
Dhinesh wrote:
> Dear experts,
>
> Reference : Section "6.2.7 Block - ISO/IEC 14496-2:2001(E)"
> "In Block decoding if the dct_dc_size_chrominance > 8 a marker_bit
> is inserted to avoid the resync_marker emulation."
>
> If the encoders follow the above statement,
> it will still lead to resync_marker emulation in some cases as explained
> below.
>
> We fould atleast two cases,
> in case the first part of the emulated resync_marker starts at
> byte-boundary.
>
> Should the encoder take care of this emulation problem by
> re-encoding the probematic-macroblock/block with changed parameters
> to avoid emulation?
>
> or
>
> Replace the text in the Section "6.2.7 Block - ISO/IEC 14496-2:2001(E)"
> ****************************************************************************
> *****
> "In Block decoding if the dct_dc_size_chrominance > 8
> a marker_bit is inserted to avoid the resync_marker emulation."
> ****************************************************************************
> *****
> with
> ****************************************************************************
> *****
> " dct_dc_size_chrominance >= 8 "
> ****************************************************************************
> *****
>
> ****************************************************************************
> *****
> EMULATION COMBINATION - 1
> ****************************************************************************
> *****
> First part of the emulated resync_marker
> -----------------------------------------
> Table B-16 -- VLC Table for Intra Luminance and Chrominance TCOEF
> VLC-CODE LAST RUN LEVEL
> 0010 000s 1 4 1
>
> Second part of the emulated resync_marker
> -----------------------------------------
> Table B-14 -- Variable length codes for dct_dc_size_chrominance
> Variable-length-code dct_dc_size_chrominance
> 0000 0000 0001 12
>
> TCOEF dct_dc_size_chrominance
> (s = 0)
> 0010 0000 0000 0000 0001
> ****************************************************************************
> *****
>
> ****************************************************************************
> *****
> EMULATION COMBINATION - 2
> ****************************************************************************
> *****
> First part of the emulated resync_marker
> -----------------------------------------------------------
> Table B-15 -- Differential DC additional codes
> Additional-code Differential-DC Size
> 00000000 to 01111111 -255 to -128 8
>
> Second part of the emulated resync_marker
> -----------------------------------------------------------
> Table B-16 -- VLC Table for Intra Luminance and Chrominance TCOEF
> VLC-CODE LAST RUN LEVEL
> 0000 0000 111s 0 0 21
> or
> 0000 0000 110s 0 0 22
> or
> 0000 0000 101s 1 0 6
> or
> 0000 0000 100s 1 0 7
>
> Differential-DC TCOEF
> (-255)
> 00000000 0000 0000 111s
> ****************************************************************************
> *****
>
> Waiting for comments .
>
> Best Regards
> Dhinesh
>
> _______________________________________________
> Technotes mailing list
> Technotes at lists.m4if.org
> http://lists.m4if.org/mailman/listinfo/technotes
More information about the XviD-devel
mailing list