[M4IF Technotes] question about intra_dc_vlc_thr2

[Kumar, Gautam]

Kosuge-San,
> What does it(mcbpc and cbpy in subclause 6.3.6 are changed 
>accordingly)  represents ? It is  that (CBP==1) -> (CBP==0)
> or (CBP==1) -> (CBP==0),or (Intra MB) -> (Inter MB) ? 

generally:
1. for inter mb we consider all 64 values(including dc value) in a block to
calculate cbp ((cbpy<<2) | (mcbpc & 3)).
2. for intra mb we consider only 63 values(excluding dc value) in a block to
calculate cbp ((cbpy<<2) | (mcbpc & 3)).
When the intra AC VLC is turned on, case 1 should be applicable for intra
mb.
Thanks and Regards,
Gautam
>-----Original Message-----
>From: tetsuo kosuge [mailto:kosuge   td.idc.lsi.sanyo.co.jp]
>Sent: Monday, February 24, 2024 6:39 AM
>To: Kumar, Gautam
>Cc: Technotes   lists.m4if.org
>Subject: Re: [M4IF Technotes] question about intra_dc_vlc_thr2
>
>
>Dear Mr.Gautam and other MPEG-4 Experts :
>
>Thanks to answer my question again, Mr.Gautam
>
>Your thouhgt is reasonable,I think so.
>If (Qp >= intra_dc_vlc_thr && pattern_code==0),usually think 
>dc_coeff exist.
>
>But I checked this sequence in microsft_vfdis_v10-990812 
>codec,seem the bitstream doesn't obey that sequence.
>In this bitstream, <pattern_code == 0> has priority over Intra_MB.
>When (Qp >= intra_dc_vlc_thr && pattern_code==0), the block 
>doesn't have dc_coeff(say has 0 value in all coeff).
>The other bitstream which does't make microsft codec obeys the 
>same sequence.
>
>And I read 14496-2:2001,this represents as follows:
>
>	7.4.1.4 Intra dc coefficient decoding for the case of 
>switched encoding
>	
>	At the VOP layer,using quantizer value as the 
>threshold,a 3 bit code(intra_dc_vlc_thr) allows switching 
>	between 2VLCs(DC Intra VLC and AC Intra VLC)when 
>decoding DC coefficients of Intra macroblocks,see Table 
>	6-21.
>
>	NOTE	When the intra AC VLC is turned on,IntraDC 
>coefficients are not handled separately any more,but 
>	treated the same as all other coefficients.That means 
>that a zero Intra DC coefficient will not be coded 
>	but will simply increase the run for the following AC 
>coefficients.The definitions of mcbpc and cbpy in 
>	subclause 6.3.6 are changed accordingly.
>
>
> What does it(mcbpc and cbpy in subclause 6.3.6 are changed 
>accordingly)  represents ? It is  that (CBP==1) -> (CBP==0)
> or (CBP==1) -> (CBP==0),or (Intra MB) -> (Inter MB) ? I can't 
>judge correct sequence(for my lack of English kowledge? 
>(TT) ).
> Which is sequence correct?
>
>Bye, Tetsuo Kosuge
>	 
>
>
>>Kosuge-San,
>>
>>intra_vlc_format is function of intra_dc_vlc_thr and qp.
>>if intra_vlc_format = 1 and mbintra = 1 ==> dc value is 
>encoded using ac
>>table (table B-16).
>>last need not to be zero. run is always zero for this, as 
>this is the first
>>element.
>>i think, escape code should be used if it falls outside the 
>table range.
>>
>>Warm Regards,
>>Gautam
>>
>>
>>>-----Original Message-----
>>>From: tetsuo kosuge [mailto:kosuge   td.idc.lsi.sanyo.co.jp]
>>>Sent: Friday, February 21, 2024 10:14 AM
>>>To: Technotes   lists.m4if.org
>>>Subject: [M4IF Technotes] question about intra_dc_vlc_thr2
>>>
>>>
>>>Dear: MPEG-4 Experts
>>>
>>>Thanks to answer my previous question, Mr.Gautam.
>>>I can understand.
>>>
>>>I have one more question.
>>>If Intra MB which over Intra_dc_thr and block pattern_code == 
>>>0, in this case dc_coeff is also derived from Intra VLC ?
>>>If derive from Intra VLC, is always LAST and RUN 0 ?  And do 
>>>exist ESC ?
>>>
>>>bye
>>>	Tetsuo Kosuge
>>>_______________________________________________
>>>Technotes mailing list
>>>Technotes   lists.m4if.org
>>>http://lists.m4if.org/mailman/listinfo/technotes
>>>
>>_______________________________________________
>>Technotes mailing list
>>Technotes   lists.m4if.org
>>http://lists.m4if.org/mailman/listinfo/technotes
>>
>