[M4IF Technotes] Question on inverse Quantization

Jani Huoponen Jani.Huoponen hantro.com
Thu Sep 13 20:05:41 EDT 2001


Dear Gene Mar,
Divisor is 16. The correction can be found from N336, Technical Correigendum
2. It is also updated to N4350, the 2001 edition of 14496-2 Visual standard.
BR, Jani 
> -----Original Message-----
> From: Eugene Mar [mailto:gmar   cvs.agilent.com]
> Sent: 13. syyskuuta 2001 1:08
> To: technotes   lists.m4if.org
> Subject: [M4IF Technotes] Question on inverse Quantization
> 
> 
> Hello,
> 
> I have a question concerning an apparent discrepency between 
> the ISO/IEC
> 14496-2 Standards Document, the ISO/IEC JTC1/SC29/WG11 N3908 VM
> Document, and the reference software from MoMuSys and Microsoft. 
> 
> On section 7.4.4.1.2 of the ISO/IEC 14496-2 Standards Document, the
> equation for F''[v][u] is shown with a divisor of 16 if 
> QF[v][u] != 0. 
> 
> On section 7.4.4.6 of the ISO/IEC 14496-2 Standards Document, 
> the sample
> c-code shows F''[v][u] with a divisor of 32.
> 
> On section 3.4.5.4.1 of the N3908 VM Document, the equation for
> F''[v][u] is shown to be divided by 32.
> 
> On section 3.4.5.7 of the N3908 VM Document, the sample c-code shows
> F''[v][u] with a divisor of 32.
> 
> Both reference c-code from MoMySys and Microsoft shows 
> F''[v][u] with a
> divisor of 16. On MoMuSys the code comes from
> vm_common\src\text_quant.c. On Microsoft the code comes from
> sys\encoder\blkenc.cpp.
> 
> My question is, is the divisor for F''[v][u] 16 or 32?
> 
> 
> Sincerely,
> 
> Gene Mar
> _______________________________________________
> Technotes mailing list
> Technotes   lists.m4if.org
> http://lists.m4if.org/mailman/listinfo/technotes
> 



More information about the Mp4-tech mailing list