hi keith,
thanks a lot.I think whatever u have suggested should definitely
work,the problem is that libmpeg2 is not properly documented and it takes
some time to understand the functions.
hope i have not disturbed u
by my chain of mails. this is my last mail for today .....any more doubts i
will definitely route it to u :)
.wish u a happy
easter .
thanks & have a nice day
byeeee
regards
anand.
On 4/5/07, Keith Winstein <[EMAIL PROTECTED]> wrote:
Hi Anand,
I'm sorry, I was mistaken. mpeg2_getpos() returns the amount REMAINING in
libmpeg2's buffer -- not the absolute position within the stream, as I
wrongly said earlier.
That buffer is replenished when you call mpeg2_buffer() to load more bytes
into the decoder.
So if you want to track position in the stream, you have to save,
yourself, a running count of the number of bytes you've fed into the
decoder. You're probably calling mpeg2_buffer() in the demultiplexer, so
that's where you'd have to keep track of this running total.
Check out mpeg2dec.c for an example of how to do this.
##############
// It defines:
static int total_offset = 0;
// Every time we load more bytes into the buffer, we add that to the
// running total:
mpeg2_buffer (mpeg2dec, current, end);
total_offset += end - current;
// to get the absolute position within the stream at a certain point:
total_offset - mpeg2_getpos (mpeg2dec)
##############
Also, keep in mind that this method will give you the absolute position in
the elemental stream -- after demultiplexing. None of the systems-layer
headers (TS, PES, etc.) will be included in the count.
Sorry to have confused you,
Keith
On Thu, 5 Apr 2007, anand meher wrote:
> Hello,
> thanks keith for your quick reply .ya,whatever you say makes a
> lot of sense. Actually i am using a transport stream as a input.i am
using
> demultiplex functions of mpeg2dec.c and then the decode functions.
>
> I dont know much about mpeg2 and other codecs, so u need to clarify me
on
> one thing. will the no of bytes of definition of pictureheader differs
for
> I,B,P frames?If its the same,then my code makes sense.
>
> What i observed using mpeg2_getpos() is that it gives the position in
the
> current 188 byte packet which is being decoded (in decode_mpeg2).
>
> the other difficulty i am facing is that the definition of a frame
extends
> over to the next 188 byte packet. so when i actaully take the difference
of
> mpeg2_getpos() in STATE_SLICE and STATE_PICTURE, i have started getting
> negative frame sizes :) .
>
> so what i am doing now is (return value of mpeg2_getpos() in
> STATE_SLICE)+(188-(return value of mpeg2_getpos() in STATE_PICTURE).
hope u
> have understood my problem. i can calrify you on this a bit more if
> required.
>
> so i need calrification whether what
i
> am doing is correct or not..
> I can use the secondary method u suggested but i am using,libmpeg2
already
> for getting the other info like chroam_width etc......
>
> thanks in advance
>
> regards
>
> anand
>
>
>
>
> On 4/5/07, Keith Winstein <[EMAIL PROTECTED]> wrote:
>>
>> Hi Anand,
>>
>> For most real-world streams, that will give you something very close to
>> the exact answer, sure. So yes, I think that should work fine.
>>
>> I hesitate for two reasons:
>>
>> (1) A picture can have more than one slice, I think, even in the Main
>> Profile. I don't think I've ever seen this in a real stream,
though.
>>
>> (2) Technically, you're calling mpeg2_getpos() AFTER libmpeg2 has
parsed
>> the header. So when you're in STATE_PICTURE, your position will be
at
>> the end of the picture header, not the beginning of it. This only
>> makes a tiny difference but if you need the "exact" frame size,
>> you might care about this.
>>
>> If you want to do this absolutely perfectly correct, don't use libmpeg2
at
>> all. Just search the MPEG-2 elemental stream for the four-byte string
>> {0x0, 0x0, 0x1, 0x0}. That's the start of a picture header. Then search
>> for the next picture start code (0x0 0x0 0x1 0x0). That's the start of
the
>> next picture. Done -- the difference in absolute position between these
>> two start codes is the number of bytes required to code the picture.
>>
>> If you want to have a more refined definition of what the "exact frame
>> size" means, you might want to also stop counting when you get to a new
>> sequence header (0x0 0x0 0x1 0xB3) or an end-of-sequence header (0x0
0x0
>> 0x1 0xB7), and then don't start up the count for the next frame until
you
>> hit a picture start (0x0 0x0 0x1 0x0).
>>
>> But unless you care about being absolutely precise down to the last few
>> tens of bytes, your method will work fine.
>>
>> Hope this helps,
>> Keith
>>
>> On Thu, 5 Apr 2007, anand meher wrote:
>>
>> > Hi,
>> > i am trying to find the frame sizes (in bytes ) using decode_mpeg2.
>> >
>> > Can i take the difference of
>> > mpeg2_getpos() return values when it hits the STATE_PICTURE(when the
>> picture
>> > header is found) and STATE_SLICE(when the last slice of a picture has
>> been
>> > found) to get the exact frame size.
>> > Thanks in advance
>> >
>> > regards
>> >
>> > anand
>> >
>> >
>> >
>> > On 4/5/07, Keith Winstein <[EMAIL PROTECTED]> wrote:
>> >>
>> >> Hi Anand,
>> >>
>> >> Please reply to the list in case somebody else wants to chime in.
>> >>
>> >> mpeg2_getpos() returns the total number of bytes (octets) since the
>> start
>> >> of the stream. So yes, you will have to take the difference between
>> >> hitting one PICTURE header and the next header (probably PICTURE or
>> >> SEQUENCE) in order to find the coded length of the picture (in
bytes).
>> >>
>> >> There's basically no way to determine the coded length of the
picture
>> >> until you hit the next header.
>> >>
>> >> Best,
>> >> Keith
>> >>
>> >> On Wed, 4 Apr 2007, anand meher wrote:
>> >>
>> >> > Hi kieth,
>> >> > As i mentioned earlier i am looking for the frame
>> >> > sizes(i.ethe exact no of bytes ) required for coding,
>> >> > I am calling mpeg2_getpos() in STATE_PICTURE. of
>> >> decode_mpeg2
>> >> > function .I am getting an integer number in return.Can u please
>> explain
>> >> > further to get the exact frame sizes . do we need to take the
>> >> > diference(i.ethe difference of the return value pf
mpeg2_getpos()when
>> >> > 1st picture header
>> >> > was hit and the second return value from mpeg2_getpos() when the
next
>> >> > picture header is hit.... OR
>> >> >
>> >> > will the reurn value of mpeg2_getpos() directly gives the frame
>> size...?
>> >>
>>
> looking
>> >> > for ur reply and thanks in advance
>> >> >
>> >> > regards
>> >> >
>> >> > anand.
>> >> >
>> >> >
>> >> >
>> >> > i hope i am making sense..please correct me if i am wrong ) .
>> >> >
>> >> >
>> >> > On 4/4/07, anand meher < [EMAIL PROTECTED]> wrote:
>> >> >>
>> >> >> Hi keith,
>> >> >> i am glad to recieve a mail from you.I am working on
a
>> >> >> project which requires the exact frame sizes(i.e the no of bytes)
>> >> >> required to code. I am using mpeg2dec.c program in libmpeg2 .
There
>> is
>> >> a
>> >> >> function called as decode_mpeg2 which is a kind of event handler.
>> >> >>
>> >> >> The point 1 in ur reply is the exact one i was asking
about.
>> I
>> >> >> think in the decode_mpeg2
>> >> >> function and in the case STATE_PICTURE ,we need to call
>> >> mpeg2_getpos().I
>> >> >> will try for that.
>> >> >> anyways i am really glad for ur quick reply.
>> >> >> If i get any more doubts..i will
>> try
>> >> to
>> >> >> mail you.
>> >> >>
>> >> >> Thanks & regards
>> >> >>
>> >> >> Anand.
>> >> >>
>> >> >>
>> >> >>
>> >> >>
>> >> >> On 4/4/07, Keith Winstein < [EMAIL PROTECTED]> wrote:
>> >> >> >
>> >> >> > Hi Anand,
>> >> >> >
>> >> >> > Not sure exactly what you mean. Are you asking about:
>> >> >> >
>> >> >> > (1) The number of bits required to code each picture ( e.g.,
>> "911376
>> >> bits
>> >> >> > for the I frame, 115936 for the B frame after it, etc.")? For
>> this,
>> >> >> > you'll
>> >> >> > have to call mpeg2_getpos() to get the position in the stream
as
>> you
>> >> hit
>> >> >> > PICTURE headers.
>> >> >> >
>> >> >> > (2) The horizontal and vertical sizes of the luma and chroma
>> >> matrices.
>> >> >> > These are in the sequence header -- see the code to
dump_state.c
>> for
>> >> an
>> >> >> > example. E.g., "1920x1088 for luma, 960x544 for chroma".
>> >> >> >
>> >> >> > (3) The intended displayed area of the frame (e.g.,
"1920x1080").
>> >> This
>> >> >> > is
>> >> >> > in the sequence display extension header, also printed out by
>> >> >> > dump_state.c
>> >> >> > ("seq->display_width", etc.).
>> >> >> >
>> >> >> > Sizes #2 and #3 are not allowed to vary on a picture-by-picture
>> >> basis.
>> >> >> > Only by ending the sequence and then starting a new sequence
can
>> you
>> >> >> > change the size of the image. (A picture display extension
header
>> can
>> >> >> > vary
>> >> >> > the location of the "display rectangle" within the full luma
and
>> >> chroma
>> >> >> > matrices, but it can't change the dimensions of the "display
>> >> >> > rectangle.")
>> >> >> >
>> >> >> > Hope that helps. If you can be more specific I'd be happy to
help
>> >> >> > further.
>> >> >> >
>> >> >> > Best,
>> >> >> > Keith
>> >> >> >
>> >> >> > On Wed, 4 Apr 2007, anand meher wrote:
>> >> >> >
>> >> >> > > Hi,
>> >> >> > > i am trying to find the frame sizes of the individual
i,b,p
>> >> frame
>> >> >> > > sizes using libmpeg2.the current picture data structure does
not
>> >> >> > provide
>> >> >> > > with the frame size.so is there any other way to find the
frame
>> >> sizes
>> >> >> > > usiiiing libmpeg2. i am using the decode function in
>> >> mpeg2dec.cprogram
>> >> >> of
>> >> >> > > libmpeg2.
>> >> >> > >
>> >> >> > > thanks in advance
>> >> >> > >
>> >> >> >
>> >> >>
>> >> >>
>> >> >
>> >>
>> >
>>
>
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Libmpeg2-devel mailing list
Libmpeg2-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmpeg2-devel
