August 2007 - ffmpeg-user - lists.ffmpeg.org

real-time encoding
by tkeinonen＠gmail.com 27 Aug '07

27 Aug '07

>Message: 8 >Date: Sun, 26 Aug 2007 14:52:44 -0700 (PDT) >From: Amin Zia <aminzia at yahoo.com> >Subject: [Ffmpeg-user] real-time encoding >To: ffmpeg-user at mplayerhq.hu >Message-ID: <719837.70417.qm at web51805.mail.re2.yahoo.com> >Content-Type: text/plain; charset=iso-8859-1 >Hi everybody, >I'm working on implementation of a real-time encoder >that reads from another process output and encodes it >into a file. I'm wondering if you have seen any >similar source that employs ffmpeg that do something >similar. >Thank you >Amin Hi, I did something similar in AGISim. (http://sourceforge.net/projects/agisim/) I implemented very simple system for encoding video data rendered in AGISIM server and sending it over socket to client system to decode and view. In other context i've used ffmpeg to encode video data using std streams. - Teemu Keinonen

1 0

need some hints to encode h264 level 1b video for my cell
by stefan_ell＠hotmail.com 27 Aug '07

27 Aug '07

hey there ffmpeg users, Since I own one of these new audio-video-whatsoever capabale Sony Ericsson phones (K610i) I'm trying to encode a video and make it play on the cell - which it constantely refuses to do no matter what i try. The aim is to create a 3pg container file with aac audio and h2.64 "level 1b" video compression. I used some info i found on the h264 wiki page (http://en.wikipedia.org/wiki/H.264#Levels) So here's what i used as command line: ffmpeg -i my_inputfile.avi -acodec aac -vcodec h264 -s qcif -r 15 the_outputfile.3gp -s qcif resizes it to 176x144 which is actually even smaller than the K610i is capable of according to the manual (176x220). -r 15 reduced the framerate to 15fps like in the "level 1b" standart. Now comes the tricky part with macroblock quantity per second and macroblock size. I tried some of the video advanced quantizer options (qsacle, mblmax) but none of them actually made any sense to me and it wouldn't make the video playable on the cell. >From what I read here (http://forum.doom9.org/archive/index.php/t-118081.html) the K610i IS "level 1b" capable. What's left to do? I also tried the wma/wmv codecs of linux but it refused to play them as well. It's all kind of strange, since the videos ALWAYS plays under linux and not a single error messages is shown during the encoding process... So anyone has some adjustments to my modeline to make it play on the cell? What is for instance about this "vbv-maxrate=128 kbit/s, vbv-bufsize=350 kbit." thingy they are talking about on the doom9 page? Is this the missing piece in the puzzle? Thanks for any hints since I'm so out of ideas... Lou _________________________________________________________________ JETZT die neueste Version des Windows Live Messenger downloaden! http://get.live.com/de-ch/messenger/overview

3 2

real-time encoding
by aminzia＠yahoo.com 26 Aug '07

26 Aug '07

Hi everybody, I'm working on implementation of a real-time encoder that reads from another process output and encodes it into a file. I'm wondering if you have seen any similar source that employs ffmpeg that do something similar. Thank you Amin

1 0

libfaad enabled but unsupported coded id 86018
by antonius.ng＠mbit.tv 26 Aug '07

26 Aug '07

Hi, I compiled ffmpeg with libfaac and libfaad enabled. But I am getting unsupported codec when trying to convert mp4 file. I have been googling for solution but could not find any. So I really appreciate any help from you. :) FFmpeg version SVN-r9358, Copyright (c) 2000-2007 Fabrice Bellard, et al. configuration: --prefix=/home/mbit/temp/ffmpegtemp/ --enable-libamr-nb --enable-libamr-wb --enable-libmp3lame --enable-libfaac --enable-libfaad --enable-libx264 --enable-libxvid --enable-shared --enable-gpl libavutil version: 49.4.0 libavcodec version: 51.40.4 libavformat version: 51.12.1 built on Aug 23 2007 17:11:14, gcc: 4.1.1 20070105 (Red Hat 4.1.1-52) Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '21082007009.mp4': Duration: 00:00:05.1, start: 0.000000, bitrate: 1192 kb/s Stream #0.0(und): Video: mpeg4, yuv420p, 640x480, 30000.00 fps(r) Stream #0.1(und): Audio: mp4a / 0x6134706D, 48000 Hz, stereo Output #0, 3gp, to 'kodok.3gp': Stream #0.0: Video: h263, yuv420p, 176x144, q=2-31, 128 kb/s, 100.00 fps(c) Stream #0.1: Audio: libamr_nb, 8000 Hz, mono, 10 kb/s Stream mapping: Stream #0.0 -> #0.0 Stream #0.1 -> #0.1 Unsupported codec (id=86018) for input stream #0.1 # ldconfig -p | grep faad* libmpeg4ip_faad.so.0 (libc6) => /usr/local/lib/libmpeg4ip_faad.so.0 libmpeg4ip_faad.so (libc6) => /usr/local/lib/libmpeg4ip_faad.so libfaad.so.0 (libc6) => /usr/local/lib/libfaad.so.0 libfaad.so (libc6) => /usr/local/lib/libfaad.so libfaac.so.0 (libc6) => /usr/local/lib/libfaac.so.0 libfaac.so (libc6) => /usr/local/lib/libfaac.so # ldd ./ffmpeg linux-gate.so.1 => (0x007b6000) libavformat.so.51 => /usr/local/lib/libavformat.so.51 (0x00579000) libavcodec.so.51 => /usr/local/lib/libavcodec.so.51 (0x00830000) libavutil.so.49 => /usr/local/lib/libavutil.so.49 (0x00110000) libm.so.6 => /lib/libm.so.6 (0x4c8f5000) libc.so.6 => /lib/libc.so.6 (0x4c7b6000) libdl.so.2 => /lib/libdl.so.2 (0x4c91e000) libamrnb.so.2 => /usr/local/lib/libamrnb.so.2 (0x006e8000) libamrwb.so.2 => /usr/local/lib/libamrwb.so.2 (0x0011a000) libmp3lame.so.0 => /usr/local/lib/libmp3lame.so.0 (0x003fc000) libx264.so.55 => /usr/local/lib/libx264.so.55 (0x0019c000) /lib/ld-linux.so.2 (0x4bde7000) libpthread.so.0 => /lib/libpthread.so.0 (0x4c924000) Regards, Antonius Ng -- Antonius Ng Technology Manager mBit Pte Ltd 245 Beach Road #02-01 Singapore 189755 Tel: +65 6223 1004 Fax: +65 6396 8179

2 1

swscale
by aviadr1＠gmail.com 26 Aug '07

26 Aug '07

Hi all, I am trying to write an image format converter (RGB32 to YUV420 for instance). I know that swscale is very powerful and fast and can do the job, but I am finding a little hard to use since there is no documentation. attached below is sample code, loosely based on imgconvert.c which tries to use swscale. basically this code works when resampling RGB32 to (a bigger) RGB32 but format conversions do not work correctly, for instance with RGB32 to RGB24 I get 4 interleaved copies of the src image in the dst buffer. my questions are: 1) are my stride calculations correct in the case where I want to convert the complete src image? 2) are my offset calculations correct in the case where I want to convert the complete src image? 3) what is the meaning of the srcSliceY, srcSliceH parameters in the sws_scale (...) function? thanks Aviad Rozenhek /// code based on ffmpeg/libavcodec/imgconvert.c static void computeComplexStride (int width, int height, int x_chroma_shift, int y_chroma_shift, int outStride [], int outOffset []) { int size = width * height; int w2 = (width + (1 << x_chroma_shift) - 1) >> x_chroma_shift; int h2 = (height + (1 << y_chroma_shift) - 1) >> y_chroma_shift; int size2 = w2 * h2; outStride[0] = width; outStride[1] = w2; outStride[2] = w2; outOffset[0] = 0; outOffset[1] = size; outOffset[2] = size + size2; } static HRESULT mediaType2ffmpegPixFmt (const CMediaType& mt, int& outPixFmt, int outStride[], int outOffset[]) { ZeroMemory (outStride, 3 * sizeof (int)); ZeroMemory (outOffset, 3 * sizeof (int)); if (mt.subtype == MEDIASUBTYPE_YV12) { outPixFmt = PIX_FMT_YUV420P; computeComplexStride ( getWidth (mt), getHeight (mt), 1, 1, outStride, outOffset ); } else if (mt.subtype == MEDIASUBTYPE_RGB32) { outPixFmt = PIX_FMT_RGB32; outStride [0] = getWidth (mt) * 4; } else if (mt.subtype == MEDIASUBTYPE_RGB24) { outPixFmt = PIX_FMT_RGB24; outStride[0] = getWidth (mt) * 3; } else if (mt.subtype == MEDIASUBTYPE_YUYV) { outPixFmt = PIX_FMT_YUYV422; outStride[0] = getWidth (mt) * 2; } else if (mt.subtype == MEDIASUBTYPE_IYUV) { outPixFmt = PIX_FMT_YUV444P; computeComplexStride ( getWidth (mt), getHeight (mt), 0, 0, outStride, outOffset ); } else if (mt.subtype == MEDIASUBTYPE_Y411) { outPixFmt = PIX_FMT_YUV411P; computeComplexStride ( getWidth (mt), getHeight (mt), 2, 0, outStride, outOffset ); } else if (mt.subtype == MEDIASUBTYPE_UYVY) { outPixFmt = PIX_FMT_UYVY422; outStride[0] = getWidth (mt) * 2; } else { return E_FAIL; } return S_OK; } int main (int argc, char* argv) { CMediaType _inMedia = /// ... CMediaType _outMedia = /// ... int inPixFmt; int outPixFmt; int inStride [3], outStride[3]; int inOffset [3], outOffset[3]; mediaType2ffmpegPixFmt (*_inMedia, inPixFmt, inStride, inOffsets); mediaType2ffmpegPixFmt (*_outMedia, outPixFmt, outStride, outOffsets); SwsContext* swsContext = sws_getContext ( getWidth (*_inMedia), getHeight(*_inMedia), inPixFmt, getWidth (*_outMedia) , getHeight(*_outMedia), outPixFmt, SWS_FAST_BILINEAR, NULL, NULL, NULL ); uint8_t* inBuffer = /// uint8_t* outBuffer = /// uint8_t* src[3] = { inBuffer + _inOffsets [0], inBuffer + _inOffsets [1], inBuffer + _inOffsets [2]}; uint8_t* dst[3] = { outBuffer + _inOffsets [0], outBuffer + _inOffsets [1], outBuffer + _inOffsets [2]}; sws_scale (_swsContext, src, inStride, 0, getHeight (*inMedia), dst, _outStride); }

1 0

How to join FLV files?&In-Reply-To=
by Shussain＠al-faris.com 25 Aug '07

25 Aug '07

Hi Johan, With referring to your posting at the URL : http://lists.mplayerhq.hu/pipermail/ffmpeg-user/2006-August/003877.html if you have found any solution please let me know. Thanks & Regards Syed S. Hussain Software Engineer SHussain at al-faris.com

1 0

recompile -fPIC error message with subversion build
by harijay＠gmail.com 24 Aug '07

24 Aug '07

Hi I just tried to build the latest subverison (10214.) build with gcc 3.2 which is only gcc version on my shared internet host. The error I get is /usr/bin/ld: /home/bioscree/lib/libz.a(compress.o): relocation R_X86_64_32 against `a local symbol' can not be used when making a shared object; recompile with -fPIC /home/bioscree/lib/libz.a: could not read symbols: Bad value collect2: ld returned 1 exit status make[1]: *** [libavcodec.so.51] Error 1 make[1]: Leaving directory `/home/bioscree/alt_test3/ffmpeg/libavcodec' make: *** [lib] Error 2 The configure options I had mentioned are ./configure --prefix=/home/bioscree --enable-memalign-hack --disable-vhook --disable-ffplay --disable-ffserver --cross-compile --enable-shared --enable-libmp3lame --extra-cflags=-I/home/bioscree/include --extra-ldflags=-L/home/bioscree/lib --enable-libfaac --enable-libfaad --enable-gpl --enable-libamr_nb --enable-libamr_wb --enable-libxvid --enable-libx264 --enable-libogg --enable-libvorbis Please can someone tell me how to get around this Thanks Hari The last few lines from make process are... make[1]: Entering directory `/home/bioscree/alt_test3/ffmpeg/libavcodec' gcc -shared -Wl,-soname,libavcodec.so.51 -L/home/bioscree/lib -Wl,--warn-common -Wl,--as-needed -Wl,-rpath-link,"/home/bioscree/alt_test3/ffmpeg"/libavcodec -Wl,-rpath-link,"/home/bioscree/alt_test3/ffmpeg"/libavformat -Wl,-rpath-link,"/home/bioscree/alt_test3/ffmpeg"/libavutil -o libavcodec.so.51 bitstream.o utils.o allcodecs.o mpegvideo.o jrevdct.o jfdctfst.o jfdctint.o resample.o resample2.o dsputil.o imgconvert.o mpeg12.o mpeg12data.o simple_idct.o eval.o error_resilience.o raw.o faandct.o parser.o opt.o bitstream_filter.o audioconvert.o imgresample.o i386/fdct_mmx.o i386/cputest.o i386/dsputil_mmx.o i386/mpegvideo_mmx.o i386/motion_est_mmx.o i386/simple_idct_mmx.o i386/idct_mmx_xvid.o i386/fft_sse.o i386/fft_3dn.o i386/fft_3dn2.o aasc.o ac3dec.o ac3tab.o ac3.o mdct.o fft.o ac3enc.o alac.o asv1.o atrac3.o avs.o bethsoftvideo.o bmp.o bmpenc.o c93.o cavs.o cavsdec.o cavsdsp.o golomb.o cinepak.o cljr.o cook.o cscd.o cyuv.o dca.o dnxhddec.o dnxhddata.o dsicinav.o dvbsubdec.o dvbsub.o dvdsubdec.o dvdsubenc.o dv.o dxa.o 8bps.o ffv1.o rangecoder.o huffyuv.o flac.o flacenc.o flashsv.o flashsvenc.o flicvideo.o h263dec.o h263.o mpegvideo_enc.o motion_est.o ratecontrol.o 4xm.o fraps.o gifdec.o lzw.o gif.o h261dec.o h261.o h261enc.o h264.o h264idct.o cabac.o idcinvideo.o imc.o indeo2.o indeo3.o interplayvideo.o dpcm.o jpeglsdec.o jpegls.o mjpegdec.o mjpeg.o jpeglsenc.o kmvc.o ljpegenc.o mjpegenc.o loco.o mace.o mjpegbdec.o mmvideo.o mpegaudiodec.o mpegaudiodecheader.o mpegaudio.o mpegaudiodata.o mpegaudioenc.o mpc.o mpeg12enc.o msmpeg4.o msmpeg4data.o msrle.o msvideo1.o lcldec.o nuv.o rtjpeg.o pnmenc.o pnm.o png.o pngdec.o pngenc.o ptx.o qdm2.o qdrw.o qpeg.o qtrle.o qtrleenc.o ra144.o ra288.o rawdec.o rawenc.o roqvideodec.o roqvideo.o roqvideoenc.o elbg.o roqaudioenc.o rpza.o rv10.o sgidec.o sgienc.o rle.o shorten.o smacker.o smc.o snow.o sonic.o sp5xdec.o svq1dec.o svq1.o svq1enc.o targa.o targaenc.o vp3.o xiph.o vp3dsp.o tiertexseqv.o tiff.o tiffenc.o lzwenc.o truemotion1.o truemotion2.o truespeech.o tscc.o tta.o txd.o s3tc.o ulti.o vc1.o vc1data.o vc1dsp.o vcr1.o vmdav.o vmnc.o vorbis_dec.o vorbis.o vorbis_data.o vorbis_enc.o vp5.o vp56.o vp56data.o vp6.o vqavideo.o wavpack.o wmadec.o wma.o wmaenc.o wnv1.o ws-snd1.o xan.o xl.o xsubdec.o lclenc.o zmbv.o zmbvenc.o pcm.o adpcm.o adx.o g726.o libamr.o libfaac.o libfaad.o libmp3lame.o libvorbis.o libx264.o libxvidff.o libxvid_rc.o aac_parser.o aac_ac3_parser.o ac3_parser.o cavs_parser.o dca_parser.o dvbsub_parser.o dvdsub_parser.o h261_parser.o h263_parser.o h264_parser.o mjpeg_parser.o mpeg4video_parser.o mpegaudio_parser.o mpegvideo_parser.o pnm_parser.o vc1_parser.o dump_extradata_bsf.o remove_extradata_bsf.o noise_bsf.o mp3_header_compress_bsf.o mp3_header_decompress_bsf.o mjpega_dump_header_bsf.o imx_dump_header_bsf.o i386/idct_mmx.o i386/cavsdsp_mmx.o i386/snowdsp_mmx.o i386/vp3dsp_mmx.o i386/vp3dsp_sse2.o -L"/home/bioscree/alt_test3/ffmpeg"/libavutil -lavutil -lm -lz -lamrnb -lm -lamrwb -lm -lmp3lame -lm -lvorbis -lvorbisenc -logg -logg -lx264 -lxvidcore -lfaac -lfaad /usr/bin/ld: /home/bioscree/lib/libz.a(compress.o): relocation R_X86_64_32 against `a local symbol' can not be used when making a shared object; recompile with -fPIC /home/bioscree/lib/libz.a: could not read symbols: Bad value collect2: ld returned 1 exit status make[1]: *** [libavcodec.so.51] Error 1 make[1]: Leaving directory `/home/bioscree/alt_test3/ffmpeg/libavcodec' make: *** [lib] Error 2

1 1

ffmpeg is not compiling mpg videos
by spam＠robolo.de 24 Aug '07

24 Aug '07

Hello, How which packages do I have to enable with ffmpeg to convert the following videos: .mov and .mpg. Because it's not possible with the default configuration. Sebastian

2 3

AV Sync using Libavcodec
by dranger＠gmail.com 24 Aug '07

24 Aug '07

> Date: Thu, 9 Aug 2007 13:06:29 +0100 > From: "Charles Mason" <charlie.mas at gmail.com> > Subject: [Ffmpeg-user] AV Sync using Libavcodec > To: ffmpeg-user at mplayerhq.hu > Message-ID: > <1960fb140708090506w2b651888q4e07c17a7fe61c7 at mail.gmail.com> > Content-Type: text/plain; charset=ISO-8859-1 > > I am currently writing an app to do video playback using libavformat > and libavcodec. I have both audio and video streams decoding nicely. > Now I want to sync the audio and video. As I can easily find out what > position the audio is currently playing, I thought I could then drop / > extend the video frame presentation to match. > > Of course to be able to do that I need to know at what time each video > frame should shown. I found a tutorial at the below address, which > shows how to get av syncing to work. > http://www.dranger.com/ffmpeg/tutorial05.html > > In the tutorial it says not to trust most of the DTS and PTS variables > and uses an extremely convoluted alternative. Is that tutorial still > correct that most of the DTS and PTS values can not be relied upon in > the latest versions of libavformat and libavcode? And is the method > outlined in the tutorial the only viable way to get the correct time? > > Charlie What I was trying to get across was not that the PTS and DTS are not to be trusted. Instead, sometimes a frame may not have a proper PTS, and other times you may have to repeat a frame, so we have to adjust our information accordingly (this is what the synchronize_video() function is for). Then, when we display it, we have to figure out how long to delay before showing the next picture. So we see what the last gap in between PTSs was, and use that as our first guess. Then we compare the PTS of a given frame to where the audio is at. As you said, you can easily figure out where the audio is at, and we do this with our global audio_clock variable which is set in the audio_decode_frame() function. Then we make our delay larger or smaller based on that discrepancy. Then we compare the delay with the Actual Clock to make sure we don't drift. I don't think this is that convoluted, it's basically doing what you're saying: get the PTS and changing it based on what time it is and how the audio is being played. Maybe there might be a better way to do this, but I am basing this off of the ffplay source code. Check out ffplay.c for how they do it, too. --Stephen

2 1

converting a MPEG-TS file to .mp4
by catchmark3＠gmail.com 24 Aug '07

24 Aug '07

Hi all I tried to convert a MPEG-TS file to mp4. Used the following command ffmpeg -i input.trp -vcodec copy -acodec copy -f mp4 out.mp4 Am gettign the following warnings and the output video I get is full of gray frames and does not have video [h264 @ 009CC230]non existing PPS referenced [h264 @ 009CC230]decode_slice_header error [h264 @ 009CC230]no frame! Any idea why I am getting the error messages. -Mark

1 0