Reply by ast September 14, 20052005-09-14
"Davy" <zhushenli@gmail.com> a &#2013265929;crit dans le message de news: 1126612343.391773.51440@f14g2000cwb.googlegroups.com...
| Hi ast,
|
| Aha, I understand.
|
| I read the table and find another related problem ;-)
|
| VLC_length = 3 => cbp = 60;
| VLC_length = 4 => cbp = 4,8,16,32;
| VLC_length = 5 => cbp = 12,48,20,40...;
|
| The probability of "4 block per microblock" is intuitively larger than
| others.
|
| Do you think the probability of "1 block per microblock" is larger
| than "2 block per microblock" ?

The smaller the vlc length, the higher the probability. So the probability of "1 block per
macroblock" is probably larger than "2 block per macroblock" even if it is not intuitive.

|
| Any suggestions will be appreciated!
|
| Best regards,
| Davy
| 

Reply by Davy September 13, 20052005-09-13
Hi ast,

Aha, I understand.

I read the table and find another related problem ;-)

VLC_length = 3 => cbp = 60;
VLC_length = 4 => cbp = 4,8,16,32;
VLC_length = 5 => cbp = 12,48,20,40...;

The probability of "4 block per microblock" is intuitively larger than
others.

Do you think the probability of "1 block per microblock" is larger
than "2 block per microblock" ?

Any suggestions will be appreciated!

Best regards,
Davy

Reply by ast September 13, 20052005-09-13
"Davy" <zhushenli@gmail.com> a &#2013265929;crit dans le message de news: 1126592823.103230.103840@o13g2000cwo.googlegroups.com...
| And There is a further problem: I read the coded_block_pattern VLC
| table.
| The shortest VLC is 111 and cbp=60.
|
| I can make an intelligent guess.
| Usually cbp represents a six bit number were each bit stands for each
| block(since this is a 420 ) thus a cbp = 60 ( => 1 1 1 1 0 0) thus it
| may be found from many sequences that the chroma component (Cr Cb)
| usually do not have any coefficients for coding (which I feel is
| intutive) and only luma component are present. Further loss of chroma
| would also not have much artifacts compared to loss of luma (Y).
|
| Is it right?

yes

|
| Any suggestions will be appreciated!
|
| Best regards,
| Davy
| 

Reply by Davy September 13, 20052005-09-13
And There is a further problem: I read the coded_block_pattern VLC
table.
The shortest VLC is 111 and cbp=60.

I can make an intelligent guess.
Usually cbp represents a six bit number were each bit stands for each
block(since this is a 420 ) thus a cbp = 60 ( => 1 1 1 1 0 0) thus it
may be found from many sequences that the chroma component (Cr Cb)
usually do not have any coefficients for coding (which I feel is
intutive) and only luma component are present. Further loss of chroma
would also not have much artifacts compared to loss of luma (Y).

Is it right?

Any suggestions will be appreciated!

Best regards,
Davy

Reply by Davy September 11, 20052005-09-11
Hi all,

I am reading MPEG-2's spec. It has a VLC coded_block_pattern_420. The
spec said we can get cbp from coded_block_pattern_420. But what's cbp's
function? Is it a important parameter?

Any suggestions will be appreciated!
Best regards,
Davy