Menu
▾
▴
jffmpeg-devel
You can subscribe to this list here.
| 2002 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
(5) |
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2003 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(1) |
Nov
(2) |
Dec
(4) |
| 2004 |
Jan
(26) |
Feb
(18) |
Mar
(12) |
Apr
(3) |
May
|
Jun
(1) |
Jul
(12) |
Aug
(5) |
Sep
(1) |
Oct
(10) |
Nov
(3) |
Dec
(3) |
| 2005 |
Jan
|
Feb
(6) |
Mar
(7) |
Apr
(10) |
May
|
Jun
|
Jul
|
Aug
(1) |
Sep
(19) |
Oct
(4) |
Nov
(2) |
Dec
(3) |
| 2006 |
Jan
(19) |
Feb
(32) |
Mar
(7) |
Apr
(1) |
May
(1) |
Jun
(8) |
Jul
(4) |
Aug
(3) |
Sep
(8) |
Oct
(3) |
Nov
(10) |
Dec
(8) |
| 2007 |
Jan
(3) |
Feb
(10) |
Mar
(1) |
Apr
|
May
|
Jun
(2) |
Jul
(2) |
Aug
(3) |
Sep
|
Oct
|
Nov
|
Dec
|
| 2012 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
(1) |
Sep
|
Oct
|
Nov
|
Dec
|
| 2013 |
Jan
|
Feb
(1) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2014 |
Jan
|
Feb
|
Mar
|
Apr
|
May
(1) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2015 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
(1) |
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: Philip F. <fr...@st...> - 2006-02-06 08:57:38
|
Hello! Does the MPEG-4 Codec implemented in jffmpeg support fine granular scalability (FGS)? Is there any place besides the source code where I can get more detailed informations about the various codecs implemented in this project? Thank you, Phil |
|
From: Philip F. <fr...@st...> - 2006-02-05 19:01:50
|
>> > The killer, however, is the requirement that each stream is coherent >> > (I assume you would want to have three or four streams and "stop >> > sending" data if the bandwidth is too low). This means that each >> > stream independently must produce some kind of satisfactory video >> > content. >> >> Actually, this is wrong: only the 1st (lowest quality) stream is >> self-contained. Other streams depend on the sum of all streams with lower >> quality in order to produce a picture with superior quality. That is the difference between MDC and SDC. In theory MDC should be possible (use ANY stream to watch video) but I have not found any implementation of it... Thought you guys might have. Thanx for the answers! Phil Jonathan Hueber wrote: > > I stand corrected :) > > JH > > >> Hi, >> >> > I am not aware of codecs that work this way. >> >> Actually, this corresponds to some optional mode of H.263+. >> I am not aware of any implementation beside the prototype used for >> standardization by ITU. >> >> > The killer, however, is the requirement that each stream is coherent >> > (I assume you would want to have three or four streams and "stop >> > sending" data if the bandwidth is too low). This means that each >> > stream independently must produce some kind of satisfactory video >> > content. >> >> Actually, this is wrong: only the 1st (lowest quality) stream is >> self-contained. Other streams depend on the sum of all streams with lower >> quality in order to produce a picture with superior quality. >> >> > At the end of the day, you would probably find that you were better >> > off encoding your video a number of times, targeting a number of >> > bit-rates, and choosing on the fly. >> >> This is how Windows Media Server works, among other products. >> >> Regards, >> Guilhem. > > > _________________________________________________________________ > Express yourself instantly with MSN Messenger! Download today it's FREE! > http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ > > > > ------------------------------------------------------- > This SF.net email is sponsored by: Splunk Inc. Do you grep through log > files > for problems? Stop! Download the new AJAX search engine that makes > searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642 > _______________________________________________ > Jffmpeg-devel mailing list > Jff...@li... > https://lists.sourceforge.net/lists/listinfo/jffmpeg-devel |
|
From: Jonathan H. <jon...@ho...> - 2006-02-05 12:42:28
|
I stand corrected :) JH >Hi, > > > I am not aware of codecs that work this way. > >Actually, this corresponds to some optional mode of H.263+. >I am not aware of any implementation beside the prototype used for >standardization by ITU. > > > The killer, however, is the requirement that each stream is coherent > > (I assume you would want to have three or four streams and "stop > > sending" data if the bandwidth is too low). This means that each > > stream independently must produce some kind of satisfactory video > > content. > >Actually, this is wrong: only the 1st (lowest quality) stream is >self-contained. Other streams depend on the sum of all streams with lower >quality in order to produce a picture with superior quality. > > > At the end of the day, you would probably find that you were better > > off encoding your video a number of times, targeting a number of > > bit-rates, and choosing on the fly. > >This is how Windows Media Server works, among other products. > >Regards, >Guilhem. _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ |
|
From: Guilhem T. <gra...@ya...> - 2006-02-04 10:00:51
|
Hi, > I am not aware of codecs that work this way. Actually, this corresponds to some optional mode of H.263+. I am not aware of any implementation beside the prototype used for standardization by ITU. > The killer, however, is the requirement that each stream is coherent > (I assume you would want to have three or four streams and "stop > sending" data if the bandwidth is too low). This means that each > stream independently must produce some kind of satisfactory video > content. Actually, this is wrong: only the 1st (lowest quality) stream is self-contained. Other streams depend on the sum of all streams with lower quality in order to produce a picture with superior quality. > At the end of the day, you would probably find that you were better > off encoding your video a number of times, targeting a number of > bit-rates, and choosing on the fly. This is how Windows Media Server works, among other products. Regards, Guilhem. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com |
|
From: Jonathan H. <jon...@ho...> - 2006-02-03 22:16:27
|
Hi, I am not aware of codecs that work this way. Internally, this is exactly what the DCT compression does, but the different components are encoded together for the best compression at a target bit-rate . The model you are describing would split the higher frequency components into separate streams. Even this simple approach would result in a loss of compression (the algorithms at this level are designed for extreme compactness, and you would end up with some quite significant overheads). The killer, however, is the requirement that each stream is coherent (I assume you would want to have three or four streams and "stop sending" data if the bandwidth is too low). This means that each stream independently must produce some kind of satisfactory video content. While this would work OK for still screens, you must remember that video decoding depends heavily on prediction from previous frame. In your model, there would be four or so different quality videos, *all* of which would need separate motion compensation. This means (1) your compression would be hit (very, very hard) (2) your encoder would have to monitor each video quality encoding - effectively equivalent to encoding the stream multiple times. At the end of the day, you would probably find that you were better off encoding your video a number of times, targeting a number of bit-rates, and choosing on the fly. The upshot is, the only way to reduce bandwidth is the either hard way... effectively equivalent to decoding and reencoding the video with different parameters. Alternatively, the easy way is to reduce frame-rate. This can be done quite easily with codecs that include B-frames (these can be stripped out without affecting the stream decoding). On the other hand, I may be completely wrong and it may be possible to get your model to work. It would take a number of months to research fully ;) Good luck, JH >Hey, > >is there a codec that allows splitting of a (a/v) stream into several >stripes? The idea is that each stripe contains the stream (audio or >video) in poor quality and several stripes can be combined to get better >quality. > >Thanks, > Phil _________________________________________________________________ Don't just search. Find. Check out the new MSN Search! http://search.msn.com/ |
|
From: Guilhem T. <gra...@ya...> - 2006-02-03 06:40:36
|
Hi, > I am tying to send a H263/RTP stream using jffmpeg. When de JMF > processor starts I get this error: > > [h263 @ 0xb0370300]a vbv buffer size is needed, for encoding with a > maximum bitrate could not open codec=20 This looks like the native version of jffmpeg, so I will answer: This error should never happen with H.263/RTP because vbv buffer size refers to MPEG-x kinds of codecs. If you changed the source code base, please try to fix your own bugs. Otherwise, you should reinstall Jffmpeg and make sure that all libraries are used correctly (no version mismatch). Regards, Guilhem. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com |
|
From: Philip F. <ph...@bo...> - 2006-02-01 13:46:16
|
Hey, is there a codec that allows splitting of a (a/v) stream into several stripes? The idea is that each stripe contains the stream (audio or video) in poor quality and several stripes can be combined to get better quality. Thanks, Phil |
|
From: Matias_Villanueva <mat...@ce...> - 2006-02-01 12:49:31
|
I am tying to send a H263/RTP stream using jffmpeg. When de JMF
processor starts I get this error:
[h263 @ 0xb0370300]a vbv buffer size is needed, for encoding with a
maximum bitrate could not open codec=20
Failed to realize: com.sun.media.ProcessEngine@f593af
Cannot build a flow graph with the customized options:
Unable to transcode format: YUV Video Format: Size =3D
java.awt.Dimension[width=3D352,height=3D288] MaxDataLength =3D 152064 =
DataType
=3D class [B yuvType =3D 2 StrideY =3D 352 StrideUV =3D 176 OffsetY =3D =
0 OffsetU
=3D 101376 OffsetV =3D 126720
to: H263/RTP, 352x288, FrameRate=3D10.0, Length=3D101376
outputting to: RAW/RTP
Error: Unable to realize com.sun.media.ProcessEngine@f593af
Am I doing something wrong?
Does anyone know how to fix the problem?
Thanks in advance.
=20
Mat=EDas Villanueva
|
|
From: Matias_Villanueva <mat...@ce...> - 2006-02-01 12:38:40
|
I am tying to send a H263/RTP stream using jffmpeg. When de JMF
processor starts I get this error:
[h263 @ 0xb0370300]a vbv buffer size is needed, for encoding with a
maximum bitrate could not open codec=20
Failed to realize: com.sun.media.ProcessEngine@f593af
Cannot build a flow graph with the customized options:
Unable to transcode format: YUV Video Format: Size =3D
java.awt.Dimension[width=3D352,height=3D288] MaxDataLength =3D 152064 =
DataType
=3D class [B yuvType =3D 2 StrideY =3D 352 StrideUV =3D 176 OffsetY =3D =
0 OffsetU
=3D 101376 OffsetV =3D 126720
to: H263/RTP, 352x288, FrameRate=3D10.0, Length=3D101376
outputting to: RAW/RTP
Error: Unable to realize com.sun.media.ProcessEngine@f593af
Am I doing something wrong?
Does anyone know how to fix the problem?
Thanks in advance.
=20
Mat=EDas Villanueva
|
|
From: Jonathan H. <jon...@ho...> - 2006-01-31 23:35:05
|
Hi, This is probably best discussed with the other developers here, but my recommendation is to look at the native decoder/encoder in JFFMPEG for H263. H263 is the technical name for the MPEG4-like video compression. The original purpose of JFFMPEG was to create a high-speed H263 solution using JNI. There is actually a pure Java H263 encoder/decoder shipped by default with JMF (but it is too slow/not as good). I can't say much about it, as I only tried it once... For audio, you could try MP3 or OGG Vorbis (although JFFMPEG doesn't have encoders for those). Other things to look at (not JFFMPEG) are the H323 video telephony standards. Basically, you have started on the quest for the Atlantis: everyone wants high quality audio/video with low bandwidth! JH >Hi, > >I am looking for a good audio and video codec to do live broadcasting over >the internet. The quality should be as good as possible but it is even more >important that the bandwidth requirements are pretty low. The software for >broadcasting is going to be written in Java using JMF and Jffmpeg. > >What codecs would you recommend? A live encoding on ordinary pc's has to be >possible. > >Thanx for your advice! > Phil _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ |
|
From: Jonathan H. <jon...@ho...> - 2006-01-31 23:24:11
|
Hi, Your MP3 problem is either going to be very easy to fix, or very hard... Easy: As your MP3 plays locally but fails remotely, there is clearly a difference between what is being passed into the JFFMPEG MP3 Decoder in the two cases. I would modify the code to dump the bytes to a file to check whether you are losing or gaining packets that cause your issue. A straight diff should be identical for the playback to be the same. Hard: The pure Java MP3 code hasn't been designed or tested for RTP streaming. If you lose packets (or the packet order changes), very bad things will happen! Buona fortuna, JH >Good evening, > >using jffmpeg in JMF I can't play an RTPstream received form a server = >which transmits an mp3. > >It raises net.sourceforge.jffmpeg.codecs.utils.FFMpegException: Buffer = >underflow=20 >at net.sourceforge.jffmpeg.codecs.utils.BitStream.getVLC(Unknown Source) > >then JMF creates a new chain of plugins: >com.sun.media.parser.RawBuffer -> = >com.sun.media.codec.audio.mpa.Depacketizer -> = >com.sun.media.codec.audio.mpa.NativeDecoder (it's an MPEG-2 decoder) -> = >com.sun.media.renderer.audio.JavaSoundRenderer > > >and plays a noise like TV channel where the signal is disturbed (an = >hiss) till last 5 seconds of the song which plays correctly. > >I tried to setup the buffer of JMF with the setBufferLength() and = >setMinimumThreshold() methods, I tried also to insert a chain (an = >RTPParser, a buffer of frame, and a Multiplexer) using the DataSource = >exiting by the Multiplexr as input for createRealizedPlayer() but the = >result is the same (also if the buffer contains 1000 frames which = >presents more of a minute of the song, the player is realized only after = >the buffer is fulled). > >If I play directly the mp3 song without streaming it, all it's ok. > >I've also tried with the mp3 plugin of Sun and setting up the buffer of = >JMF, or putting the chain with my buffer, it renders the stream = >correctly, so I ask you if can be an incompatibility with the mpeg = >depacketizer, or if jffmpeg has a bug with RTPstream or if i miss = >something in the code (I tried both with jmf solution as AVReceive and = >with a simple receiver of RTPstream, not a lot different from jfm = >solution, on which I insert the my buffer). > >Last thing, if I register Sun codec plugin between jffmpeg codec and = >NativeDecoder it don't raise the exception ad creates the chain with the = >Sun codec plugin instead of jffmpeg, but if Sun codec is registered = >under NativeDecoder I've the problem I'd described over. > >Someone could help me about this? > > >Thaks for help and answer, i'm sorry for my poor English > > > >Sergio V. > > _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ |
|
From: Jonathan H. <jon...@ho...> - 2006-01-31 23:17:32
|
Hi Jean, As Guilhem says, I believe that jffmpeg is not currently in sync with ffmpeg CVS. I normally pull the source from the ffmpeg release page tarball (version 0.4.9-pre1 from July 2004). That said, I can't imagine that fixing the native code to work with CVS ffmpeg would be too hard. JH > >--=-7gf5AXGw6GrL+HwrC0cC >Content-Type: text/plain; charset=ISO-8859-15 >Content-Transfer-Encoding: quoted-printable > >Hi everyone, > I'm trying to build jffmpeg via the config.mak / Makefile. Here are >the steps that I've done. > >I've downloaded ffmpeg version 0.47.0 via CVS with the command: cvs >-r"release_0_47_0" -d:pserver:ano...@mp...:/cvsroot/ffmpeg co >ffmpeg >I've downloaded the jffmpeg-1.1.0.src.tar.gz and decompress it. > >Let's say we are rootDir/ffmpeg and rootDir/jffmpeg. ffmpeg build just >fine. For jffmpeg, I've edited config.mak to reflect my configuration. I >didn't touch to the Makefile. When I try to do "make", in the >rootDir/jffmpeg, I have the following error: > >gcc -I. "-I/usr/java/j2sdk1.4.2_10/include" >"-I/usr/java/j2sdk1.4.2_10/include/linux" >"-I/home/jfcrotea/views/jfcrotea_PEP_1/vobs/FELIN1_COTS/ffmpeg" >-Ibuild/cpp -Wall -D_FILE_OFFSET_BITS=3D64 -D_LARGEFILE_SOURCE >-D_GNU_SOURCE -c -o build/cpp/decoder.o src/cpp/decoder.c >In file included from src/cpp/decoder.c:37: >/home/jfcrotea/views/jfcrotea_PEP_1/vobs/FELIN1_COTS/ffmpeg/libavcodec/avco= >dec.h:14:20: error: avutil.h: No such file or directory >In file included from src/cpp/decoder.c:37: >/home/jfcrotea/views/jfcrotea_PEP_1/vobs/FELIN1_COTS/ffmpeg/libavcodec/avco= >dec.h:687: error: syntax error before 'uint8_t' > >............ and more and more errors. > >I didn't apply the 0.47.0 patch. > >Any idea? > >Thank you in advance _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ |
|
From: Sergio V. <ge...@ti...> - 2006-01-31 22:58:36
|
Hello, thanks for your answer, I'm sorry but I didn't explained very well the question. Jffmpeg codec is the first of the list of registered codecs (it's at the top) in both of case. If Sun codec mp3-plugin (com.sun.media.codec.audio.mp3.JavaDecoder) is between jffmpeg and com.sun.media.codec.audio.mpa.NativeDecoder, jffmpeg doesn't raise any exception and I find Sun mp3-plugin after codec depacketizer and renderer (and mp3 stream is correctly played). Else, if Sun codec mp3-plugin is under com.sun.media.codec.audio.mpa.NativeDecoder (which is also under jffmpeg codec plugin) jffmpeg plugin raises the exception. In both case jffmpeg plugin does't appear in chain but in one of them it raises the exception of buffer underflow; I don't understand this = different behavior. Thanks again Sergio V. ----- Original Message ----- From: "Guilhem Tardy" <gra...@ya...> To: <jff...@li...> Sent: Tuesday, January 31, 2006 5:44 AM Subject: [Jffmpeg-devel] Re: Buffer underflow on rendering = mp3-RTP-stream (Sergio Valisi) > Hi, > > Hannes knows better about the MP3 stuff (it must be pure Java), but I = can at > least answer this point: > > > if I register Sun codec plugin between jffmpeg > > codec and NativeDecoder it don't raise the exception ad creates > > the chain with the Sun codec plugin instead of jffmpeg, but if > > Sun codec is registered under NativeDecoder I've the problem > > I'd described over. > > This is normal, if you register Sun codecs before jffmpeg (i.e. higher = in the > list), than the latter is not used at all. Order matters. > > Regards, > Guilhem. |
|
From: Philip F. <ph...@bo...> - 2006-01-31 12:33:46
|
Hi, I am looking for a good audio and video codec to do live broadcasting over the internet. The quality should be as good as possible but it is even more important that the bandwidth requirements are pretty low. The software for broadcasting is going to be written in Java using JMF and Jffmpeg. What codecs would you recommend? A live encoding on ordinary pc's has to be possible. Thanx for your advice! Phil |
|
From: Guilhem T. <gra...@ya...> - 2006-01-31 04:48:16
|
Hi,
> I'm trying to build jffmpeg via the config.mak / Makefile. Here
> are the steps that I've done.
The error messages indicate that another version of ffmpeg is used.
avutil.h: No such file or directory
This file did not exist in ffmpeg 0.4.7. Please make sure of your -I and -L
paths in the command to gcc, or suffer the consequences.
Regards,
Guilhem.
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
|
|
From: Guilhem T. <gra...@ya...> - 2006-01-31 04:44:55
|
Hi, Hannes knows better about the MP3 stuff (it must be pure Java), but I can at least answer this point: > if I register Sun codec plugin between jffmpeg > codec and NativeDecoder it don't raise the exception ad creates > the chain with the Sun codec plugin instead of jffmpeg, but if > Sun codec is registered under NativeDecoder I've the problem > I'd described over. This is normal, if you register Sun codecs before jffmpeg (i.e. higher in the list), than the latter is not used at all. Order matters. Regards, Guilhem. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com |
|
From: Jean-Francois C. <JFC...@oe...> - 2006-01-30 21:41:45
|
Hi everyone,
I'm trying to build jffmpeg via the config.mak / Makefile. Here are
the steps that I've done.
I've downloaded ffmpeg version 0.47.0 via CVS with the command: cvs
-r"release_0_47_0" -d:pserver:ano...@mp...:/cvsroot/ffmpeg co
ffmpeg
I've downloaded the jffmpeg-1.1.0.src.tar.gz and decompress it.
Let's say we are rootDir/ffmpeg and rootDir/jffmpeg. ffmpeg build just
fine. For jffmpeg, I've edited config.mak to reflect my configuration. I
didn't touch to the Makefile. When I try to do "make", in the
rootDir/jffmpeg, I have the following error:
gcc -I. "-I/usr/java/j2sdk1.4.2_10/include"
"-I/usr/java/j2sdk1.4.2_10/include/linux"
"-I/home/jfcrotea/views/jfcrotea_PEP_1/vobs/FELIN1_COTS/ffmpeg"
-Ibuild/cpp -Wall -D_FILE_OFFSET_BITS=3D64 -D_LARGEFILE_SOURCE
-D_GNU_SOURCE -c -o build/cpp/decoder.o src/cpp/decoder.c
In file included from src/cpp/decoder.c:37:
/home/jfcrotea/views/jfcrotea_PEP_1/vobs/FELIN1_COTS/ffmpeg/libavcodec/avco=
dec.h:14:20: error: avutil.h: No such file or directory
In file included from src/cpp/decoder.c:37:
/home/jfcrotea/views/jfcrotea_PEP_1/vobs/FELIN1_COTS/ffmpeg/libavcodec/avco=
dec.h:687: error: syntax error before 'uint8_t'
............ and more and more errors.
I didn't apply the 0.47.0 patch.
Any idea?
Thank you in advance
--=20
Jean-Fran=E7ois Croteau
Programmeur
Oerlikon Contraves
http://www.oerlikon.ca
courriel: jfc...@oe...
(450) 358-7596
|
|
From: Sergio V. <ge...@ti...> - 2006-01-30 20:56:41
|
Good evening, using jffmpeg in JMF I can't play an RTPstream received form a server = which transmits an mp3. It raises net.sourceforge.jffmpeg.codecs.utils.FFMpegException: Buffer = underflow=20 at net.sourceforge.jffmpeg.codecs.utils.BitStream.getVLC(Unknown Source) then JMF creates a new chain of plugins: com.sun.media.parser.RawBuffer -> = com.sun.media.codec.audio.mpa.Depacketizer -> = com.sun.media.codec.audio.mpa.NativeDecoder (it's an MPEG-2 decoder) -> = com.sun.media.renderer.audio.JavaSoundRenderer and plays a noise like TV channel where the signal is disturbed (an = hiss) till last 5 seconds of the song which plays correctly. I tried to setup the buffer of JMF with the setBufferLength() and = setMinimumThreshold() methods, I tried also to insert a chain (an = RTPParser, a buffer of frame, and a Multiplexer) using the DataSource = exiting by the Multiplexr as input for createRealizedPlayer() but the = result is the same (also if the buffer contains 1000 frames which = presents more of a minute of the song, the player is realized only after = the buffer is fulled). If I play directly the mp3 song without streaming it, all it's ok. I've also tried with the mp3 plugin of Sun and setting up the buffer of = JMF, or putting the chain with my buffer, it renders the stream = correctly, so I ask you if can be an incompatibility with the mpeg = depacketizer, or if jffmpeg has a bug with RTPstream or if i miss = something in the code (I tried both with jmf solution as AVReceive and = with a simple receiver of RTPstream, not a lot different from jfm = solution, on which I insert the my buffer). Last thing, if I register Sun codec plugin between jffmpeg codec and = NativeDecoder it don't raise the exception ad creates the chain with the = Sun codec plugin instead of jffmpeg, but if Sun codec is registered = under NativeDecoder I've the problem I'd described over. Someone could help me about this? Thaks for help and answer, i'm sorry for my poor English Sergio V. |
|
From: Jean-Francois C. <JFC...@oe...> - 2006-01-23 17:40:51
|
Hi everyone,
Is there anyone that successfully uses the latest ffmpeg version
from CVS with jffmpeg? If yes, can you send me an email, please?
Thank you
--=20
Jean-Fran=E7ois Croteau
Programmeur
Oerlikon Contraves
http://www.oerlikon.ca
courriel: jfc...@oe...
|
|
From: Jonathan H. <jon...@ho...> - 2006-01-21 09:32:32
|
Yup, that's pretty much how I did it.
The problem is that looping over width and height can get a bit painful when
dealing with a 640x480 video. There are a couple of tricks I used:
1) Define variables before enering for loops. This should be an
optimisation made by the compiler, but it isn't.
2) Use integer arithmetic rather than floating point, and by using powers
of two, right shifts instead of division.
3) Use comparisons with zero rather than with constants or variables - as
the status flags are automatically. So:
for ( int x = 0; x < 100; x++ ) changes to for ( int x = 99; x >= 0; x--
)
4) Don't assume that the compiler will optimise r = red[ x + y * screenX ]
b = blue[ x + y * screenX]
As Guilem says - there is a YUV to RGB converter written in C which is much
faster (it takes out the array bound checks and indexing), but that would
kind of defeats the purpose of not using libjffmpeg.so!
JH
>From: Stormboy :
>
>Great JH. I'll try it out.
>
>Just out of interest... there is an efficient way to convert YUV to RGB
>using OpenGL and 3D texture maps. I have written code to do this for
>HLS (hue, luminance and saturation) to RGB using JOGL - it turns out to
>be extremely quick. I imagine it should also work for YUV.
>
>It involves creating a 2 x 8 x 64 texture map where the dimension of the
>map represent the S x H x L values, and the pixels represent the RGB
>mapping of the HLS coordinates. For each pixel in the texture map, an
>RGB entry for the HLS coordinates is placed.
>
>To render RGB values, you may draw QUADS onto the GL context for a
>particular point in the texture map representing the HLS value to draw.
>OpenGL will then interpolate to render the actual RGB value.
>
>Here's an example of drawing a buffer of HLS pixels to GL, where HLS
>values are between 0.0 and 1.0.
>
> int i = 0;
> for (int y=0; y<height; y++) {
>
> gl.glBegin(GL.GL_QUAD_STRIP);
>
> for (int x=0; x<width; x++) {
>
> float L = buf[i++];
> float H = buf[i++];
> float S = buf[i++];
>
> // map values to bounds of 3D texture
> L = (float) (( L * 63.0 + 0.5) / 64.0);
> H = (float) (( H * 6.0 + 0.5) / 8.0);
> S = (float) (( S * 1.0 + 0.5) / 2.0);
>
> // bottom-left
> gl.glTexCoord3f(S, H, L);
>
> gl.glVertex2d(x, -y);
>
> // top-left
> gl.glTexCoord3f(S, H, L);
> gl.glVertex2d(x, -y-1);
>
> // you may optionally draw the other 2
>points of the quad
> }
>
> gl.glEnd();
> }
>
>You may even repopulate the RGB values in the texture map for different
>contrast/brightness/colour-map settings.
>
>Stormboy
>
>Jonathan Hueber wrote:
> > Hi,
> >
> > I thought it was about time to roll up the fixes I had made over the
> > past months. There are quite a few fixes for MPEG4, XVID and DIVX.
> >
> > There has also been quite a lot of work on performance and profiling.
> > Rather embarassingly, most of the CPU was actually being used
> > converting YUV to RGB. I should have known this - the floating point
> > unit in i386 is extrememly slow in propagating error codes (Java is
> > designed for SPARC processors). Similarly, I have learnt not to rely
> > on the compiler to optimise my expressions.
> >
> > Ah well... Enjoy.
> >
> > JH
> >
> > _________________________________________________________________
> > FREE pop-up blocking with the new MSN Toolbar - get it now!
> > http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/
> >
> >
> >
> > -------------------------------------------------------
> > This SF.net email is sponsored by: Splunk Inc. Do you grep through log
> > files
> > for problems? Stop! Download the new AJAX search engine that makes
> > searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
> > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642
> > _______________________________________________
> > Jffmpeg-devel mailing list
> > Jff...@li...
> > https://lists.sourceforge.net/lists/listinfo/jffmpeg-devel
>
_________________________________________________________________
Don't just search. Find. Check out the new MSN Search!
http://search.msn.com/
|
|
From: Guilhem T. <gra...@ya...> - 2006-01-21 07:22:41
|
Jonathan, > I thought it was about time to roll up the fixes I had made > over the past months. There are quite a few fixes for MPEG4, > XVID and DIVX. > > There has also been quite a lot of work on performance and > profiling. Rather embarassingly, most of the CPU was actually > being used converting YUV to RGB. Thanks for this great contribution. Please note that some code to the same effect (i.e. convert YUV to RGB) is in the native files on JFFMPEG's CVS, too. If you have any question about it, please feel free to email me privately. Best regards, Guilhem. __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com |
|
From: Stormboy <sto...@st...> - 2006-01-21 01:11:15
|
Great JH. I'll try it out.
Just out of interest... there is an efficient way to convert YUV to RGB
using OpenGL and 3D texture maps. I have written code to do this for
HLS (hue, luminance and saturation) to RGB using JOGL - it turns out to
be extremely quick. I imagine it should also work for YUV.
It involves creating a 2 x 8 x 64 texture map where the dimension of the
map represent the S x H x L values, and the pixels represent the RGB
mapping of the HLS coordinates. For each pixel in the texture map, an
RGB entry for the HLS coordinates is placed.
To render RGB values, you may draw QUADS onto the GL context for a
particular point in the texture map representing the HLS value to draw.
OpenGL will then interpolate to render the actual RGB value.
Here's an example of drawing a buffer of HLS pixels to GL, where HLS
values are between 0.0 and 1.0.
int i = 0;
for (int y=0; y<height; y++) {
gl.glBegin(GL.GL_QUAD_STRIP);
for (int x=0; x<width; x++) {
float L = buf[i++];
float H = buf[i++];
float S = buf[i++];
// map values to bounds of 3D texture
L = (float) (( L * 63.0 + 0.5) / 64.0);
H = (float) (( H * 6.0 + 0.5) / 8.0);
S = (float) (( S * 1.0 + 0.5) / 2.0);
// bottom-left
gl.glTexCoord3f(S, H, L);
gl.glVertex2d(x, -y);
// top-left
gl.glTexCoord3f(S, H, L);
gl.glVertex2d(x, -y-1);
// you may optionally draw the other 2
points of the quad
}
gl.glEnd();
}
You may even repopulate the RGB values in the texture map for different
contrast/brightness/colour-map settings.
Stormboy
Jonathan Hueber wrote:
> Hi,
>
> I thought it was about time to roll up the fixes I had made over the
> past months. There are quite a few fixes for MPEG4, XVID and DIVX.
>
> There has also been quite a lot of work on performance and profiling.
> Rather embarassingly, most of the CPU was actually being used
> converting YUV to RGB. I should have known this - the floating point
> unit in i386 is extrememly slow in propagating error codes (Java is
> designed for SPARC processors). Similarly, I have learnt not to rely
> on the compiler to optimise my expressions.
>
> Ah well... Enjoy.
>
> JH
>
> _________________________________________________________________
> FREE pop-up blocking with the new MSN Toolbar - get it now!
> http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through log
> files
> for problems? Stop! Download the new AJAX search engine that makes
> searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642
> _______________________________________________
> Jffmpeg-devel mailing list
> Jff...@li...
> https://lists.sourceforge.net/lists/listinfo/jffmpeg-devel
|
|
From: Jonathan H. <jon...@ho...> - 2006-01-21 00:31:37
|
Hi, I thought it was about time to roll up the fixes I had made over the past months. There are quite a few fixes for MPEG4, XVID and DIVX. There has also been quite a lot of work on performance and profiling. Rather embarassingly, most of the CPU was actually being used converting YUV to RGB. I should have known this - the floating point unit in i386 is extrememly slow in propagating error codes (Java is designed for SPARC processors). Similarly, I have learnt not to rely on the compiler to optimise my expressions. Ah well... Enjoy. JH _________________________________________________________________ FREE pop-up blocking with the new MSN Toolbar - get it now! http://toolbar.msn.click-url.com/go/onm00200415ave/direct/01/ |
|
From: Jonathan H. <jon...@ho...> - 2006-01-15 12:32:30
|
Hi, The current Java code needs an immensely power computer to run at full speed - (Dual processor or core). Also, there are still a fair few bugs when dealing with the different types of MPEG4 encoders (although the code in CVS has a few more fixes). Your best bet is to get the JNI version of JFFMPEG working (libjffmpeg.so). This involves setting the java.library.path variable: java -Djava.library.path=path/to/directory JMStudio Unfortunately, video in Java a little hit-and-miss - this is mainly due to the poor support Sun have for JMF. Maybe one day, they'll include it as part of the default Java install... Good luck, JH >Hi, > >Personally I've not tried the MPEG4 decoder in jffmpeg yet, but I think >Jonathan may have added it. How's the CPU usage on your machine while >all this is going on? > >On Sat, 2006-07-01 at 13:46 -0500, Track God wrote: > > > Hello, > > > > I am trying to write a Java application (to run on the Linux platform) > > to play back MPEG4 movies. I am using transcode to encode a movie, and > > was playing around with the xvid4 options. I can get Linux (mplayer) > > and Windows (Media Player) to play the movie just fine, but when I try > > to use JMStudio (with jffmpeg installed), the video is extremely > > choppy (a frame every 5 seconds?) and the audio plays fine. What are > > my options for this? Even playing the VOB directly is still skippy > > (much better than xvid4) in JMStudio. I want to be able to play these > > movies in Java. I'd like to avoid programming my application in C, but > > I cannot seem to find a way to play DVD-like video within Java. > > Anyone's input on how to go about this would be appreciated. > > > > - Matt > > > > > _________________________________________________________________ Express yourself instantly with MSN Messenger! Download today it's FREE! http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/ |
|
From: David S. <da...@si...> - 2006-01-10 16:31:42
|
Hi, Personally I've not tried the MPEG4 decoder in jffmpeg yet, but I think Jonathan may have added it. How's the CPU usage on your machine while all this is going on? On Sat, 2006-07-01 at 13:46 -0500, Track God wrote: > Hello, > > I am trying to write a Java application (to run on the Linux platform) > to play back MPEG4 movies. I am using transcode to encode a movie, and > was playing around with the xvid4 options. I can get Linux (mplayer) > and Windows (Media Player) to play the movie just fine, but when I try > to use JMStudio (with jffmpeg installed), the video is extremely > choppy (a frame every 5 seconds?) and the audio plays fine. What are > my options for this? Even playing the VOB directly is still skippy > (much better than xvid4) in JMStudio. I want to be able to play these > movies in Java. I'd like to avoid programming my application in C, but > I cannot seem to find a way to play DVD-like video within Java. > Anyone's input on how to go about this would be appreciated. > > - Matt > > Dave -- David Stuart, SIPquest Email: dave (at) sipquest (dot) com Phone: 254-8886 x234 Web: http://www.sipquest.com/ Address: 300 - 350 Terry Fox Drive, Kanata Ontario, K2K 2P5 |
2 messages has been excluded from this view by a project administrator.
