Ok, but I still don't get how the final recipient gets to know the
*original* presentation time (which correspond to audio) if I generate a
new one in the transcoder... Is this done via RTCP?! If yes, could you
shortly explain it, please?
Julian
___
liv
Hi!
I've developed a class to transcode video streams. I measure the time
between "doGetNextFrame()" and "afterGetting(this)" with the
gettimeofday() function. The results show that it takes in average 40ms
to complete one frame (according to the frame rate of 25Hz). But
irregularily one frame
> Because they're not useful to you.
>
> I'll say this yet again (and hopefully for the last time): Our
> RTP/RTCP implementation automatically computes synchronized
> presentation times using RTP timestamps and RTCP reports. Code that
> receives a RTP stream (using our library) should never h
Hi!
As RTCP SR packets carry NTP ans RTP timestamps I would like to know how
they are used to sync to streams. I found out that the RTP timestamp in
the RTCP SR corresponds to the timevalue which the NTP timestamp indicates.
But why can't I find the RTP timestamp in the packets of my videostrea
Ross Finlayson schrieb:
>> How is the timeval structure mapped onto the 32bit presentation time in
>> the RTP packet, what interval do I have to add to fPresentationTime to
>> make RTP timestamp increase by 3600 (i.e. one frame duration @25Hz)?
>>
>
> Receiving code should not need to care abo
> Note that In your original MPEG-2 stream, with B-frames, the frames
> will be in decoding order, which is different from the display order,
> and thus different from the order of frames in the resulting MPEG-4
> stream (because that doesn't have B frames). In particular, the
> presentation
I've set up an RTCPInstance for my Sink exactly as shown in the
testprogs, but there are no RTCP packets sent.
Also the vobStreamer just sends one single RTCP packet.
Why is that? Shouldn't there be more RTCP traffic?
Julian
smime.p7s
Description: S/MIME Cryptographic Signature
___
Hi!
I'm transcoding MPEG-2 (with B-Frames) into MPEG-4 (without B-Frames,
just I and P). Currently I'm just transcoding video, but later the
transcoded streams should by able to be synchronised with the source's
audio stream:
> Audio
|
Another interesting effect is that the average total processing time for
one frame (measured from the start of doGetNextFrame() until after
afterGetting(this)) is always close to 40ms, no matter how long my
transcoder needs to process one frame (assuming that this time stays
below ~35ms, which
No. Assuming that your "afterGettingFrame()" function was passed as
a parameter to "getNextFrame()", then it will be called by
"FramedSource::afterGetting()" (which your "doGetNextFrame()"
implementation should have called once it completed delivery of
incoming data).
So is my code correc
Am I right assuming that my afterGettingFrame() function is called by
the TaskScheduler? Sometimes it happens that afterGettingFrame() is
called before the first call has completed. Could anyone explain, why
sometimes I have a delay of >10ms between two calls to
afterGettingFrame()?
smime.p7
OK, I've investigated the problem further by adding timestamp at the
beginning and the end of doGetNextFrame(), afterGettingFrame() and
afterGettingFrame1(). As you can see from the code snippet in this
thread, everytime my transcoder needs more data to complete the frame it
calls fInputSource-
If your "Transcoder" filter delivers discrete MPEG-4 frames, and sets
"fPresentationTime" and "fDurationInMicroseconds" properly, then you
don't need to insert a "MPEG4VideoStreamDiscreteFramer".
It does, but if I leave out the framer, the sink doesn't even call
doGetNextFrame() of my clas
Hi!
I've implemented a FramedFilter subclass that transcoders video from
MPEG-2 to MPEG-4 using the libavcodec library.
Therefore I use a live-"chain" that looks like:
MPEG1or2VideoRTPSource -> Transcoder (my class) ->
MPEG4VideoStreamDiscreteFramer -> MPEG4ESVideoRTPSink
The Transcoder's
Hi Christian ;)
I am having a few problems while parsing PES stripped from DVB-T
MPEG-TS to the MPEGDemux Class.
PES coming from DVB-T are indeed very peculiar. First of all, they can
be very big. Some channels had PESs up to 150KB... so first question:
is this correct? Can PES really be unbo
Hi!
I would like to know how vobStreamer respectively MPEG1or2VideoRTP
Source encapsulate the VideoStream from a DVD into the RTP Packets. Does
it extract the Elementary Streams from the Program stream and pack the
Elementary streams according to RFC2250?
Why does Wireshark identify the RTP
Note that the error about "atempting to read more than once" occurs
if you call "getNextFrame()" on the *same* object a second time,
before the 'after getting' function from the first call has been
invoked. I.e., it's OK to do:
source->getNextFrame(..., afterGettingFunc, ...);
> Are you sure that your "Transcoder" code is always correctly setting
"fFrameSize" before calling "afterGetting(this)"?
OK, I found the problem I think, but now I need support:
I made a call to afterGetting(this) even if there was no encoded frame
available. I changed that now, but there is a
Hi!
I would like to know how I can
1. Request more data from a source (MPEG1or2VideoRTPSource)
2. Tell a sink (MPEG4VideoStreamDiscreteFramer) that data is completely
delivered to buffer (this should be afterGetting(this), right?)
Right now I have a code structure like:
void Transcoder::doGe
Hi!
I've added some stdout messages to MPEG4VideoStreamDiscreteFramer and I
can see that there are many calls to doGetNextFrame() and
afterGettingFrame1() even if my source did not deliver one frame:
MPEG4VideoStreamDiscreteFramer::doGetNextFrame() passed
MPEG4VideoStreamDiscreteFramer::after
Thanks for your reply, I changed my code.
Nevertheless, enc_bytes did never exceed fMaxSize in my program before
and thus the problem still exists. VLC, which I use to play the stream,
reports loads of errors regarding late pictures and damaged headers...
:( Any more ideas?
Thank you!
Julian
Hi Severin!
How did you do that, I'm trying exactly the same. But when I use
MPEG4VideoStreamFramer instead of the discrete one I get errors that say:
StreamParser::afterGettingBytes() warning: read 23329 bytes; expected no
more than 10026
MultiFramedRTPSink::afterGettingFrame1(): The input f
If I dump the stream at the receiver with openRTSP I also get a
corrupted stream, so it's not a problem with the player I'm using (VLC
with live555 support).
smime.p7s
Description: S/MIME Cryptographic Signature
___
live-devel mailing list
live-devel@
Hi!
I've a ffmpeg code that writes MPEG4 frames into a buffer. I want to
stream these frames over RTP/RTSP. At the moment I pass the buffer to an
"MPEG4VideoStreamDiscreteFramer" that sends it to an "MPEG4ESVideoRTPSink".
But the stream I receive is totally corrupted. Using wireshark I can
see
Hi!
I have an encoder that writes MPEG4 frames to a buffer which I want to
stream out with MPEG4ESVideoRTPSink.
I'm using the MPEG4VideoStreamDiscreteFramer class in between.
The problem is that the stream is corrupted, but if I write the buffer
to a file like "test.m4e" it can be played cor
As you already guessed, I did that ;)
I took MP3ADUTranscoder as an example where presentation time and
duration are just passed through. I don't get why my code does not work.
When I write one complete frame to fTo, MPEG4VideoStreamDiscreteFramer
should be able to pass them correctly to MPEG4
Hi!
I've some questions related to the FramedFilter class. I subclassed it
to transcode a MPEG2 Stream to MPEG4. So the structure of my program
looks like:
MPEG1or2VideoRTPSource -> Transcoder (my class) ->
MPEG4VideoStreamDiscreteFramer -> MPEG4ESVideoRTPSink
My Transcoder class reads dat
Sorry, my fault ;) You should not run wireshark on the same computer
that sends the packets...
smime.p7s
Description: S/MIME Cryptographic Signature
___
live-devel mailing list
live-devel@lists.live555.com
http://lists.live555.com/mailman/listinfo/live
Hi!
Since my transcoder now transcodes I have a new problem:
I deliver complete MPEG4 Frames to MPEG4VideoStreamDiscreteFramer
followed by MPEG4ESVideoRTPSink.
But all the packets sent have a wrong UDP checksum. Wireshark reports
that a lot of subsequent UDP packets have the SAME checksum. For
OK, I found the solution to *this* problem ;)
I'm now using MPEG4VideoStreamDiscreteFramer and get no more errors
reported.
Thank you anyway!
smime.p7s
Description: S/MIME Cryptographic Signature
___
live-devel mailing list
live-devel@lists.live555.c
Humm, actually it should not:
if(enc_bytes > fMaxSize)
{
fFrameSize = fMaxSize;
fNumTruncatedBytes = enc_bytes - fMaxSize;
}
else
{
fFrameSize = enc_bytes;
fNumTruncatedBytes = 0;
}
smime.p7s
Description: S/MIME Cryptographic Signature
___
Hi!
When I try to stream out MPEG4 content using MPEG4VideoStreamFramer and
MPEG4ESVideoRTPSink I frequently get errors that say:
StreamParser::afterGettingBytes() warning: read X bytes; expected no
more than Y
(where X is fFrameSize).
What am I doing wrong?
smime.p7s
Description: S/MIME
Therefore, if you are feeding input from a "MPEG1or2VideoRTPSource"
into a decoder, and your decoder is not smart enough to decode one
slice at a time, then you must aggregate the input data into complete
video frames before feeding them to your decoder.
Can this be done with live555 stuff?
J
Hi!
I've subclassed FramedFilter to transcode a MPEG2 Stream (from
vobStreamer) to MPEG4.
My "live-chain" is: MPEG1or2VideoRTPSource -> Transcoder -> FileSink
I use the following code:
void Transcoder::doGetNextFrame() {
fInputSource->getNextFrame(fOrg, 4096, afterGettingFrame, this,
han
34 matches
Mail list logo
