git.sesse.net Git - casparcg/blob - dependencies64/ffmpeg/doc/faq.html

   1 <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
   2 <html>
   3 <!-- Created by GNU Texinfo 5.2, http://www.gnu.org/software/texinfo/ -->
   4   <head>
   5     <meta charset="utf-8">
   6     <title>
   7       FFmpeg FAQ
   8     </title>
   9     <meta name="viewport" content="width=device-width,initial-scale=1.0">
  10     <link rel="stylesheet" type="text/css" href="bootstrap.min.css">
  11     <link rel="stylesheet" type="text/css" href="style.min.css">
  12   </head>
  13   <body>
  14     <div style="width: 95%; margin: auto">
  15       <h1>
  16       FFmpeg FAQ
  17       </h1>
  18 <div align="center">
  19 </div>
  20
  21
  22 <a name="SEC_Top"></a>
  23
  24 <a name="SEC_Contents"></a>
  25 <h2 class="contents-heading">Table of Contents</h2>
  26
  27 <div class="contents">
  28
  29 <ul class="no-bullet">
  30   <li><a name="toc-General-Questions" href="#General-Questions">1 General Questions</a>
  31   <ul class="no-bullet">
  32     <li><a name="toc-Why-doesn_0027t-FFmpeg-support-feature-_005bxyz_005d_003f" href="#Why-doesn_0027t-FFmpeg-support-feature-_005bxyz_005d_003f">1.1 Why doesn&rsquo;t FFmpeg support feature [xyz]?</a></li>
  33     <li><a name="toc-FFmpeg-does-not-support-codec-XXX_002e-Can-you-include-a-Windows-DLL-loader-to-support-it_003f" href="#FFmpeg-does-not-support-codec-XXX_002e-Can-you-include-a-Windows-DLL-loader-to-support-it_003f">1.2 FFmpeg does not support codec XXX. Can you include a Windows DLL loader to support it?</a></li>
  34     <li><a name="toc-I-cannot-read-this-file-although-this-format-seems-to-be-supported-by-ffmpeg_002e" href="#I-cannot-read-this-file-although-this-format-seems-to-be-supported-by-ffmpeg_002e">1.3 I cannot read this file although this format seems to be supported by ffmpeg.</a></li>
  35     <li><a name="toc-Which-codecs-are-supported-by-Windows_003f" href="#Which-codecs-are-supported-by-Windows_003f">1.4 Which codecs are supported by Windows?</a></li>
  36   </ul></li>
  37   <li><a name="toc-Compilation" href="#Compilation">2 Compilation</a>
  38   <ul class="no-bullet">
  39     <li><a name="toc-error_003a-can_0027t-find-a-register-in-class-_0027GENERAL_005fREGS_0027-while-reloading-_0027asm_0027" href="#error_003a-can_0027t-find-a-register-in-class-_0027GENERAL_005fREGS_0027-while-reloading-_0027asm_0027">2.1 <code>error: can't find a register in class 'GENERAL_REGS' while reloading 'asm'</code></a></li>
  40     <li><a name="toc-I-have-installed-this-library-with-my-distro_0027s-package-manager_002e-Why-does-configure-not-see-it_003f" href="#I-have-installed-this-library-with-my-distro_0027s-package-manager_002e-Why-does-configure-not-see-it_003f">2.2 I have installed this library with my distro&rsquo;s package manager. Why does <code>configure</code> not see it?</a></li>
  41     <li><a name="toc-How-do-I-make-pkg_002dconfig-find-my-libraries_003f" href="#How-do-I-make-pkg_002dconfig-find-my-libraries_003f">2.3 How do I make <code>pkg-config</code> find my libraries?</a></li>
  42     <li><a name="toc-How-do-I-use-pkg_002dconfig-when-cross_002dcompiling_003f" href="#How-do-I-use-pkg_002dconfig-when-cross_002dcompiling_003f">2.4 How do I use <code>pkg-config</code> when cross-compiling?</a></li>
  43   </ul></li>
  44   <li><a name="toc-Usage" href="#Usage">3 Usage</a>
  45   <ul class="no-bullet">
  46     <li><a name="toc-ffmpeg-does-not-work_003b-what-is-wrong_003f" href="#ffmpeg-does-not-work_003b-what-is-wrong_003f">3.1 ffmpeg does not work; what is wrong?</a></li>
  47     <li><a name="toc-How-do-I-encode-single-pictures-into-movies_003f" href="#How-do-I-encode-single-pictures-into-movies_003f">3.2 How do I encode single pictures into movies?</a></li>
  48     <li><a name="toc-How-do-I-encode-movie-to-single-pictures_003f" href="#How-do-I-encode-movie-to-single-pictures_003f">3.3 How do I encode movie to single pictures?</a></li>
  49     <li><a name="toc-Why-do-I-see-a-slight-quality-degradation-with-multithreaded-MPEG_002a-encoding_003f" href="#Why-do-I-see-a-slight-quality-degradation-with-multithreaded-MPEG_002a-encoding_003f">3.4 Why do I see a slight quality degradation with multithreaded MPEG* encoding?</a></li>
  50     <li><a name="toc-How-can-I-read-from-the-standard-input-or-write-to-the-standard-output_003f" href="#How-can-I-read-from-the-standard-input-or-write-to-the-standard-output_003f">3.5 How can I read from the standard input or write to the standard output?</a></li>
  51     <li><a name="toc-_002df-jpeg-doesn_0027t-work_002e" href="#g_t_002df-jpeg-doesn_0027t-work_002e">3.6 -f jpeg doesn&rsquo;t work.</a></li>
  52     <li><a name="toc-Why-can-I-not-change-the-frame-rate_003f" href="#Why-can-I-not-change-the-frame-rate_003f">3.7 Why can I not change the frame rate?</a></li>
  53     <li><a name="toc-How-do-I-encode-Xvid-or-DivX-video-with-ffmpeg_003f" href="#How-do-I-encode-Xvid-or-DivX-video-with-ffmpeg_003f">3.8 How do I encode Xvid or DivX video with ffmpeg?</a></li>
  54     <li><a name="toc-Which-are-good-parameters-for-encoding-high-quality-MPEG_002d4_003f" href="#Which-are-good-parameters-for-encoding-high-quality-MPEG_002d4_003f">3.9 Which are good parameters for encoding high quality MPEG-4?</a></li>
  55     <li><a name="toc-Which-are-good-parameters-for-encoding-high-quality-MPEG_002d1_002fMPEG_002d2_003f" href="#Which-are-good-parameters-for-encoding-high-quality-MPEG_002d1_002fMPEG_002d2_003f">3.10 Which are good parameters for encoding high quality MPEG-1/MPEG-2?</a></li>
  56     <li><a name="toc-Interlaced-video-looks-very-bad-when-encoded-with-ffmpeg_002c-what-is-wrong_003f" href="#Interlaced-video-looks-very-bad-when-encoded-with-ffmpeg_002c-what-is-wrong_003f">3.11 Interlaced video looks very bad when encoded with ffmpeg, what is wrong?</a></li>
  57     <li><a name="toc-How-can-I-read-DirectShow-files_003f" href="#How-can-I-read-DirectShow-files_003f">3.12 How can I read DirectShow files?</a></li>
  58     <li><a name="toc-How-can-I-join-video-files_003f" href="#How-can-I-join-video-files_003f">3.13 How can I join video files?</a></li>
  59     <li><a name="toc-How-can-I-concatenate-video-files_003f" href="#How-can-I-concatenate-video-files_003f">3.14 How can I concatenate video files?</a>
  60     <ul class="no-bullet">
  61       <li><a name="toc-Concatenating-using-the-concat-filter" href="#Concatenating-using-the-concat-filter">3.14.1 Concatenating using the concat <em>filter</em></a></li>
  62       <li><a name="toc-Concatenating-using-the-concat-demuxer" href="#Concatenating-using-the-concat-demuxer">3.14.2 Concatenating using the concat <em>demuxer</em></a></li>
  63       <li><a name="toc-Concatenating-using-the-concat-protocol-_0028file-level_0029" href="#Concatenating-using-the-concat-protocol-_0028file-level_0029">3.14.3 Concatenating using the concat <em>protocol</em> (file level)</a></li>
  64       <li><a name="toc-Concatenating-using-raw-audio-and-video" href="#Concatenating-using-raw-audio-and-video">3.14.4 Concatenating using raw audio and video</a></li>
  65     </ul></li>
  66     <li><a name="toc-Using-_002df-lavfi_002c-audio-becomes-mono-for-no-apparent-reason_002e" href="#Using-_002df-lavfi_002c-audio-becomes-mono-for-no-apparent-reason_002e">3.15 Using <samp>-f lavfi</samp>, audio becomes mono for no apparent reason.</a></li>
  67     <li><a name="toc-Why-does-FFmpeg-not-see-the-subtitles-in-my-VOB-file_003f" href="#Why-does-FFmpeg-not-see-the-subtitles-in-my-VOB-file_003f">3.16 Why does FFmpeg not see the subtitles in my VOB file?</a></li>
  68     <li><a name="toc-Why-was-the-ffmpeg-_002dsameq-option-removed_003f-What-to-use-instead_003f" href="#Why-was-the-ffmpeg-_002dsameq-option-removed_003f-What-to-use-instead_003f">3.17 Why was the <code>ffmpeg</code> <samp>-sameq</samp> option removed? What to use instead?</a></li>
  69     <li><a name="toc-I-have-a-stretched-video_002c-why-does-scaling-does-not-fix-it_003f" href="#I-have-a-stretched-video_002c-why-does-scaling-does-not-fix-it_003f">3.18 I have a stretched video, why does scaling does not fix it?</a></li>
  70   </ul></li>
  71   <li><a name="toc-Development" href="#Development">4 Development</a>
  72   <ul class="no-bullet">
  73     <li><a name="toc-Are-there-examples-illustrating-how-to-use-the-FFmpeg-libraries_002c-particularly-libavcodec-and-libavformat_003f" href="#Are-there-examples-illustrating-how-to-use-the-FFmpeg-libraries_002c-particularly-libavcodec-and-libavformat_003f">4.1 Are there examples illustrating how to use the FFmpeg libraries, particularly libavcodec and libavformat?</a></li>
  74     <li><a name="toc-Can-you-support-my-C-compiler-XXX_003f" href="#Can-you-support-my-C-compiler-XXX_003f">4.2 Can you support my C compiler XXX?</a></li>
  75     <li><a name="toc-Is-Microsoft-Visual-C_002b_002b-supported_003f" href="#Is-Microsoft-Visual-C_002b_002b-supported_003f">4.3 Is Microsoft Visual C++ supported?</a></li>
  76     <li><a name="toc-Can-you-add-automake_002c-libtool-or-autoconf-support_003f" href="#Can-you-add-automake_002c-libtool-or-autoconf-support_003f">4.4 Can you add automake, libtool or autoconf support?</a></li>
  77     <li><a name="toc-Why-not-rewrite-FFmpeg-in-object_002doriented-C_002b_002b_003f" href="#Why-not-rewrite-FFmpeg-in-object_002doriented-C_002b_002b_003f">4.5 Why not rewrite FFmpeg in object-oriented C++?</a></li>
  78     <li><a name="toc-Why-are-the-ffmpeg-programs-devoid-of-debugging-symbols_003f" href="#Why-are-the-ffmpeg-programs-devoid-of-debugging-symbols_003f">4.6 Why are the ffmpeg programs devoid of debugging symbols?</a></li>
  79     <li><a name="toc-I-do-not-like-the-LGPL_002c-can-I-contribute-code-under-the-GPL-instead_003f" href="#I-do-not-like-the-LGPL_002c-can-I-contribute-code-under-the-GPL-instead_003f">4.7 I do not like the LGPL, can I contribute code under the GPL instead?</a></li>
  80     <li><a name="toc-I_0027m-using-FFmpeg-from-within-my-C-application-but-the-linker-complains-about-missing-symbols-from-the-libraries-themselves_002e" href="#I_0027m-using-FFmpeg-from-within-my-C-application-but-the-linker-complains-about-missing-symbols-from-the-libraries-themselves_002e">4.8 I&rsquo;m using FFmpeg from within my C application but the linker complains about missing symbols from the libraries themselves.</a></li>
  81     <li><a name="toc-I_0027m-using-FFmpeg-from-within-my-C_002b_002b-application-but-the-linker-complains-about-missing-symbols-which-seem-to-be-available_002e" href="#I_0027m-using-FFmpeg-from-within-my-C_002b_002b-application-but-the-linker-complains-about-missing-symbols-which-seem-to-be-available_002e">4.9 I&rsquo;m using FFmpeg from within my C++ application but the linker complains about missing symbols which seem to be available.</a></li>
  82     <li><a name="toc-I_0027m-using-libavutil-from-within-my-C_002b_002b-application-but-the-compiler-complains-about-_0027UINT64_005fC_0027-was-not-declared-in-this-scope" href="#I_0027m-using-libavutil-from-within-my-C_002b_002b-application-but-the-compiler-complains-about-_0027UINT64_005fC_0027-was-not-declared-in-this-scope">4.10 I&rsquo;m using libavutil from within my C++ application but the compiler complains about &rsquo;UINT64_C&rsquo; was not declared in this scope</a></li>
  83     <li><a name="toc-I-have-a-file-in-memory-_002f-a-API-different-from-_002aopen_002f_002aread_002f-libc-how-do-I-use-it-with-libavformat_003f" href="#I-have-a-file-in-memory-_002f-a-API-different-from-_002aopen_002f_002aread_002f-libc-how-do-I-use-it-with-libavformat_003f">4.11 I have a file in memory / a API different from *open/*read/ libc how do I use it with libavformat?</a></li>
  84     <li><a name="toc-Where-is-the-documentation-about-ffv1_002c-msmpeg4_002c-asv1_002c-4xm_003f" href="#Where-is-the-documentation-about-ffv1_002c-msmpeg4_002c-asv1_002c-4xm_003f">4.12 Where is the documentation about ffv1, msmpeg4, asv1, 4xm?</a></li>
  85     <li><a name="toc-How-do-I-feed-H_002e263_002dRTP-_0028and-other-codecs-in-RTP_0029-to-libavcodec_003f" href="#How-do-I-feed-H_002e263_002dRTP-_0028and-other-codecs-in-RTP_0029-to-libavcodec_003f">4.13 How do I feed H.263-RTP (and other codecs in RTP) to libavcodec?</a></li>
  86     <li><a name="toc-AVStream_002er_005fframe_005frate-is-wrong_002c-it-is-much-larger-than-the-frame-rate_002e" href="#AVStream_002er_005fframe_005frate-is-wrong_002c-it-is-much-larger-than-the-frame-rate_002e">4.14 AVStream.r_frame_rate is wrong, it is much larger than the frame rate.</a></li>
  87     <li><a name="toc-Why-is-make-fate-not-running-all-tests_003f" href="#Why-is-make-fate-not-running-all-tests_003f">4.15 Why is <code>make fate</code> not running all tests?</a></li>
  88     <li><a name="toc-Why-is-make-fate-not-finding-the-samples_003f" href="#Why-is-make-fate-not-finding-the-samples_003f">4.16 Why is <code>make fate</code> not finding the samples?</a></li>
  89   </ul></li>
  90 </ul>
  91 </div>
  92
  93
  94 <a name="General-Questions"></a>
  95 <h2 class="chapter">1 General Questions<span class="pull-right"><a class="anchor hidden-xs" href="#General-Questions" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-General-Questions" aria-hidden="true">TOC</a></span></h2>
  96
  97 <a name="Why-doesn_0027t-FFmpeg-support-feature-_005bxyz_005d_003f"></a>
  98 <h3 class="section">1.1 Why doesn&rsquo;t FFmpeg support feature [xyz]?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-doesn_0027t-FFmpeg-support-feature-_005bxyz_005d_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-doesn_0027t-FFmpeg-support-feature-_005bxyz_005d_003f" aria-hidden="true">TOC</a></span></h3>
  99
 100 <p>Because no one has taken on that task yet. FFmpeg development is
 101 driven by the tasks that are important to the individual developers.
 102 If there is a feature that is important to you, the best way to get
 103 it implemented is to undertake the task yourself or sponsor a developer.
 104 </p>
 105 <a name="FFmpeg-does-not-support-codec-XXX_002e-Can-you-include-a-Windows-DLL-loader-to-support-it_003f"></a>
 106 <h3 class="section">1.2 FFmpeg does not support codec XXX. Can you include a Windows DLL loader to support it?<span class="pull-right"><a class="anchor hidden-xs" href="#FFmpeg-does-not-support-codec-XXX_002e-Can-you-include-a-Windows-DLL-loader-to-support-it_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-FFmpeg-does-not-support-codec-XXX_002e-Can-you-include-a-Windows-DLL-loader-to-support-it_003f" aria-hidden="true">TOC</a></span></h3>
 107
 108 <p>No. Windows DLLs are not portable, bloated and often slow.
 109 Moreover FFmpeg strives to support all codecs natively.
 110 A DLL loader is not conducive to that goal.
 111 </p>
 112 <a name="I-cannot-read-this-file-although-this-format-seems-to-be-supported-by-ffmpeg_002e"></a>
 113 <h3 class="section">1.3 I cannot read this file although this format seems to be supported by ffmpeg.<span class="pull-right"><a class="anchor hidden-xs" href="#I-cannot-read-this-file-although-this-format-seems-to-be-supported-by-ffmpeg_002e" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I-cannot-read-this-file-although-this-format-seems-to-be-supported-by-ffmpeg_002e" aria-hidden="true">TOC</a></span></h3>
 114
 115 <p>Even if ffmpeg can read the container format, it may not support all its
 116 codecs. Please consult the supported codec list in the ffmpeg
 117 documentation.
 118 </p>
 119 <a name="Which-codecs-are-supported-by-Windows_003f"></a>
 120 <h3 class="section">1.4 Which codecs are supported by Windows?<span class="pull-right"><a class="anchor hidden-xs" href="#Which-codecs-are-supported-by-Windows_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Which-codecs-are-supported-by-Windows_003f" aria-hidden="true">TOC</a></span></h3>
 121
 122 <p>Windows does not support standard formats like MPEG very well, unless you
 123 install some additional codecs.
 124 </p>
 125 <p>The following list of video codecs should work on most Windows systems:
 126 </p><dl compact="compact">
 127 <dt><samp>msmpeg4v2</samp></dt>
 128 <dd><p>.avi/.asf
 129 </p></dd>
 130 <dt><samp>msmpeg4</samp></dt>
 131 <dd><p>.asf only
 132 </p></dd>
 133 <dt><samp>wmv1</samp></dt>
 134 <dd><p>.asf only
 135 </p></dd>
 136 <dt><samp>wmv2</samp></dt>
 137 <dd><p>.asf only
 138 </p></dd>
 139 <dt><samp>mpeg4</samp></dt>
 140 <dd><p>Only if you have some MPEG-4 codec like ffdshow or Xvid installed.
 141 </p></dd>
 142 <dt><samp>mpeg1video</samp></dt>
 143 <dd><p>.mpg only
 144 </p></dd>
 145 </dl>
 146 <p>Note, ASF files often have .wmv or .wma extensions in Windows. It should also
 147 be mentioned that Microsoft claims a patent on the ASF format, and may sue
 148 or threaten users who create ASF files with non-Microsoft software. It is
 149 strongly advised to avoid ASF where possible.
 150 </p>
 151 <p>The following list of audio codecs should work on most Windows systems:
 152 </p><dl compact="compact">
 153 <dt><samp>adpcm_ima_wav</samp></dt>
 154 <dt><samp>adpcm_ms</samp></dt>
 155 <dt><samp>pcm_s16le</samp></dt>
 156 <dd><p>always
 157 </p></dd>
 158 <dt><samp>libmp3lame</samp></dt>
 159 <dd><p>If some MP3 codec like LAME is installed.
 160 </p></dd>
 161 </dl>
 162
 163
 164 <a name="Compilation"></a>
 165 <h2 class="chapter">2 Compilation<span class="pull-right"><a class="anchor hidden-xs" href="#Compilation" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Compilation" aria-hidden="true">TOC</a></span></h2>
 166
 167 <a name="error_003a-can_0027t-find-a-register-in-class-_0027GENERAL_005fREGS_0027-while-reloading-_0027asm_0027"></a>
 168 <h3 class="section">2.1 <code>error: can't find a register in class 'GENERAL_REGS' while reloading 'asm'</code><span class="pull-right"><a class="anchor hidden-xs" href="#error_003a-can_0027t-find-a-register-in-class-_0027GENERAL_005fREGS_0027-while-reloading-_0027asm_0027" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-error_003a-can_0027t-find-a-register-in-class-_0027GENERAL_005fREGS_0027-while-reloading-_0027asm_0027" aria-hidden="true">TOC</a></span></h3>
 169
 170 <p>This is a bug in gcc. Do not report it to us. Instead, please report it to
 171 the gcc developers. Note that we will not add workarounds for gcc bugs.
 172 </p>
 173 <p>Also note that (some of) the gcc developers believe this is not a bug or
 174 not a bug they should fix:
 175 <a href="http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11203">http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11203</a>.
 176 Then again, some of them do not know the difference between an undecidable
 177 problem and an NP-hard problem...
 178 </p>
 179 <a name="I-have-installed-this-library-with-my-distro_0027s-package-manager_002e-Why-does-configure-not-see-it_003f"></a>
 180 <h3 class="section">2.2 I have installed this library with my distro&rsquo;s package manager. Why does <code>configure</code> not see it?<span class="pull-right"><a class="anchor hidden-xs" href="#I-have-installed-this-library-with-my-distro_0027s-package-manager_002e-Why-does-configure-not-see-it_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I-have-installed-this-library-with-my-distro_0027s-package-manager_002e-Why-does-configure-not-see-it_003f" aria-hidden="true">TOC</a></span></h3>
 181
 182 <p>Distributions usually split libraries in several packages. The main package
 183 contains the files necessary to run programs using the library. The
 184 development package contains the files necessary to build programs using the
 185 library. Sometimes, docs and/or data are in a separate package too.
 186 </p>
 187 <p>To build FFmpeg, you need to install the development package. It is usually
 188 called <samp>libfoo-dev</samp> or <samp>libfoo-devel</samp>. You can remove it after the
 189 build is finished, but be sure to keep the main package.
 190 </p>
 191 <a name="How-do-I-make-pkg_002dconfig-find-my-libraries_003f"></a>
 192 <h3 class="section">2.3 How do I make <code>pkg-config</code> find my libraries?<span class="pull-right"><a class="anchor hidden-xs" href="#How-do-I-make-pkg_002dconfig-find-my-libraries_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-do-I-make-pkg_002dconfig-find-my-libraries_003f" aria-hidden="true">TOC</a></span></h3>
 193
 194 <p>Somewhere along with your libraries, there is a <samp>.pc</samp> file (or several)
 195 in a <samp>pkgconfig</samp> directory. You need to set environment variables to
 196 point <code>pkg-config</code> to these files.
 197 </p>
 198 <p>If you need to <em>add</em> directories to <code>pkg-config</code>&rsquo;s search list
 199 (typical use case: library installed separately), add it to
 200 <code>$PKG_CONFIG_PATH</code>:
 201 </p>
 202 <div class="example">
 203 <pre class="example">export PKG_CONFIG_PATH=/opt/x264/lib/pkgconfig:/opt/opus/lib/pkgconfig
 204 </pre></div>
 205
 206 <p>If you need to <em>replace</em> <code>pkg-config</code>&rsquo;s search list
 207 (typical use case: cross-compiling), set it in
 208 <code>$PKG_CONFIG_LIBDIR</code>:
 209 </p>
 210 <div class="example">
 211 <pre class="example">export PKG_CONFIG_LIBDIR=/home/me/cross/usr/lib/pkgconfig:/home/me/cross/usr/local/lib/pkgconfig
 212 </pre></div>
 213
 214 <p>If you need to know the library&rsquo;s internal dependencies (typical use: static
 215 linking), add the <code>--static</code> option to <code>pkg-config</code>:
 216 </p>
 217 <div class="example">
 218 <pre class="example">./configure --pkg-config-flags=--static
 219 </pre></div>
 220
 221 <a name="How-do-I-use-pkg_002dconfig-when-cross_002dcompiling_003f"></a>
 222 <h3 class="section">2.4 How do I use <code>pkg-config</code> when cross-compiling?<span class="pull-right"><a class="anchor hidden-xs" href="#How-do-I-use-pkg_002dconfig-when-cross_002dcompiling_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-do-I-use-pkg_002dconfig-when-cross_002dcompiling_003f" aria-hidden="true">TOC</a></span></h3>
 223
 224 <p>The best way is to install <code>pkg-config</code> in your cross-compilation
 225 environment. It will automatically use the cross-compilation libraries.
 226 </p>
 227 <p>You can also use <code>pkg-config</code> from the host environment by
 228 specifying explicitly <code>--pkg-config=pkg-config</code> to <code>configure</code>.
 229 In that case, you must point <code>pkg-config</code> to the correct directories
 230 using the <code>PKG_CONFIG_LIBDIR</code>, as explained in the previous entry.
 231 </p>
 232 <p>As an intermediate solution, you can place in your cross-compilation
 233 environment a script that calls the host <code>pkg-config</code> with
 234 <code>PKG_CONFIG_LIBDIR</code> set. That script can look like that:
 235 </p>
 236 <div class="example">
 237 <pre class="example">#!/bin/sh
 238 PKG_CONFIG_LIBDIR=/path/to/cross/lib/pkgconfig
 239 export PKG_CONFIG_LIBDIR
 240 exec /usr/bin/pkg-config &quot;$@&quot;
 241 </pre></div>
 242
 243 <a name="Usage"></a>
 244 <h2 class="chapter">3 Usage<span class="pull-right"><a class="anchor hidden-xs" href="#Usage" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Usage" aria-hidden="true">TOC</a></span></h2>
 245
 246 <a name="ffmpeg-does-not-work_003b-what-is-wrong_003f"></a>
 247 <h3 class="section">3.1 ffmpeg does not work; what is wrong?<span class="pull-right"><a class="anchor hidden-xs" href="#ffmpeg-does-not-work_003b-what-is-wrong_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-ffmpeg-does-not-work_003b-what-is-wrong_003f" aria-hidden="true">TOC</a></span></h3>
 248
 249 <p>Try a <code>make distclean</code> in the ffmpeg source directory before the build.
 250 If this does not help see
 251 (<a href="http://ffmpeg.org/bugreports.html">http://ffmpeg.org/bugreports.html</a>).
 252 </p>
 253 <a name="How-do-I-encode-single-pictures-into-movies_003f"></a>
 254 <h3 class="section">3.2 How do I encode single pictures into movies?<span class="pull-right"><a class="anchor hidden-xs" href="#How-do-I-encode-single-pictures-into-movies_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-do-I-encode-single-pictures-into-movies_003f" aria-hidden="true">TOC</a></span></h3>
 255
 256 <p>First, rename your pictures to follow a numerical sequence.
 257 For example, img1.jpg, img2.jpg, img3.jpg,...
 258 Then you may run:
 259 </p>
 260 <div class="example">
 261 <pre class="example">ffmpeg -f image2 -i img%d.jpg /tmp/a.mpg
 262 </pre></div>
 263
 264 <p>Notice that &lsquo;<samp>%d</samp>&rsquo; is replaced by the image number.
 265 </p>
 266 <p><samp>img%03d.jpg</samp> means the sequence <samp>img001.jpg</samp>, <samp>img002.jpg</samp>, etc.
 267 </p>
 268 <p>Use the <samp>-start_number</samp> option to declare a starting number for
 269 the sequence. This is useful if your sequence does not start with
 270 <samp>img001.jpg</samp> but is still in a numerical order. The following
 271 example will start with <samp>img100.jpg</samp>:
 272 </p>
 273 <div class="example">
 274 <pre class="example">ffmpeg -f image2 -start_number 100 -i img%d.jpg /tmp/a.mpg
 275 </pre></div>
 276
 277 <p>If you have large number of pictures to rename, you can use the
 278 following command to ease the burden. The command, using the bourne
 279 shell syntax, symbolically links all files in the current directory
 280 that match <code>*jpg</code> to the <samp>/tmp</samp> directory in the sequence of
 281 <samp>img001.jpg</samp>, <samp>img002.jpg</samp> and so on.
 282 </p>
 283 <div class="example">
 284 <pre class="example">x=1; for i in *jpg; do counter=$(printf %03d $x); ln -s &quot;$i&quot; /tmp/img&quot;$counter&quot;.jpg; x=$(($x+1)); done
 285 </pre></div>
 286
 287 <p>If you want to sequence them by oldest modified first, substitute
 288 <code>$(ls -r -t *jpg)</code> in place of <code>*jpg</code>.
 289 </p>
 290 <p>Then run:
 291 </p>
 292 <div class="example">
 293 <pre class="example">ffmpeg -f image2 -i /tmp/img%03d.jpg /tmp/a.mpg
 294 </pre></div>
 295
 296 <p>The same logic is used for any image format that ffmpeg reads.
 297 </p>
 298 <p>You can also use <code>cat</code> to pipe images to ffmpeg:
 299 </p>
 300 <div class="example">
 301 <pre class="example">cat *.jpg | ffmpeg -f image2pipe -c:v mjpeg -i - output.mpg
 302 </pre></div>
 303
 304 <a name="How-do-I-encode-movie-to-single-pictures_003f"></a>
 305 <h3 class="section">3.3 How do I encode movie to single pictures?<span class="pull-right"><a class="anchor hidden-xs" href="#How-do-I-encode-movie-to-single-pictures_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-do-I-encode-movie-to-single-pictures_003f" aria-hidden="true">TOC</a></span></h3>
 306
 307 <p>Use:
 308 </p>
 309 <div class="example">
 310 <pre class="example">ffmpeg -i movie.mpg movie%d.jpg
 311 </pre></div>
 312
 313 <p>The <samp>movie.mpg</samp> used as input will be converted to
 314 <samp>movie1.jpg</samp>, <samp>movie2.jpg</samp>, etc...
 315 </p>
 316 <p>Instead of relying on file format self-recognition, you may also use
 317 </p><dl compact="compact">
 318 <dt><samp>-c:v ppm</samp></dt>
 319 <dt><samp>-c:v png</samp></dt>
 320 <dt><samp>-c:v mjpeg</samp></dt>
 321 </dl>
 322 <p>to force the encoding.
 323 </p>
 324 <p>Applying that to the previous example:
 325 </p><div class="example">
 326 <pre class="example">ffmpeg -i movie.mpg -f image2 -c:v mjpeg menu%d.jpg
 327 </pre></div>
 328
 329 <p>Beware that there is no &quot;jpeg&quot; codec. Use &quot;mjpeg&quot; instead.
 330 </p>
 331 <a name="Why-do-I-see-a-slight-quality-degradation-with-multithreaded-MPEG_002a-encoding_003f"></a>
 332 <h3 class="section">3.4 Why do I see a slight quality degradation with multithreaded MPEG* encoding?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-do-I-see-a-slight-quality-degradation-with-multithreaded-MPEG_002a-encoding_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-do-I-see-a-slight-quality-degradation-with-multithreaded-MPEG_002a-encoding_003f" aria-hidden="true">TOC</a></span></h3>
 333
 334 <p>For multithreaded MPEG* encoding, the encoded slices must be independent,
 335 otherwise thread n would practically have to wait for n-1 to finish, so it&rsquo;s
 336 quite logical that there is a small reduction of quality. This is not a bug.
 337 </p>
 338 <a name="How-can-I-read-from-the-standard-input-or-write-to-the-standard-output_003f"></a>
 339 <h3 class="section">3.5 How can I read from the standard input or write to the standard output?<span class="pull-right"><a class="anchor hidden-xs" href="#How-can-I-read-from-the-standard-input-or-write-to-the-standard-output_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-can-I-read-from-the-standard-input-or-write-to-the-standard-output_003f" aria-hidden="true">TOC</a></span></h3>
 340
 341 <p>Use <samp>-</samp> as file name.
 342 </p>
 343 <a name="g_t_002df-jpeg-doesn_0027t-work_002e"></a>
 344 <h3 class="section">3.6 -f jpeg doesn&rsquo;t work.<span class="pull-right"><a class="anchor hidden-xs" href="#_002df-jpeg-doesn_0027t-work_002e" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-_002df-jpeg-doesn_0027t-work_002e" aria-hidden="true">TOC</a></span></h3>
 345
 346 <p>Try &rsquo;-f image2 test%d.jpg&rsquo;.
 347 </p>
 348 <a name="Why-can-I-not-change-the-frame-rate_003f"></a>
 349 <h3 class="section">3.7 Why can I not change the frame rate?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-can-I-not-change-the-frame-rate_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-can-I-not-change-the-frame-rate_003f" aria-hidden="true">TOC</a></span></h3>
 350
 351 <p>Some codecs, like MPEG-1/2, only allow a small number of fixed frame rates.
 352 Choose a different codec with the -c:v command line option.
 353 </p>
 354 <a name="How-do-I-encode-Xvid-or-DivX-video-with-ffmpeg_003f"></a>
 355 <h3 class="section">3.8 How do I encode Xvid or DivX video with ffmpeg?<span class="pull-right"><a class="anchor hidden-xs" href="#How-do-I-encode-Xvid-or-DivX-video-with-ffmpeg_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-do-I-encode-Xvid-or-DivX-video-with-ffmpeg_003f" aria-hidden="true">TOC</a></span></h3>
 356
 357 <p>Both Xvid and DivX (version 4+) are implementations of the ISO MPEG-4
 358 standard (note that there are many other coding formats that use this
 359 same standard). Thus, use &rsquo;-c:v mpeg4&rsquo; to encode in these formats. The
 360 default fourcc stored in an MPEG-4-coded file will be &rsquo;FMP4&rsquo;. If you want
 361 a different fourcc, use the &rsquo;-vtag&rsquo; option. E.g., &rsquo;-vtag xvid&rsquo; will
 362 force the fourcc &rsquo;xvid&rsquo; to be stored as the video fourcc rather than the
 363 default.
 364 </p>
 365 <a name="Which-are-good-parameters-for-encoding-high-quality-MPEG_002d4_003f"></a>
 366 <h3 class="section">3.9 Which are good parameters for encoding high quality MPEG-4?<span class="pull-right"><a class="anchor hidden-xs" href="#Which-are-good-parameters-for-encoding-high-quality-MPEG_002d4_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Which-are-good-parameters-for-encoding-high-quality-MPEG_002d4_003f" aria-hidden="true">TOC</a></span></h3>
 367
 368 <p>&rsquo;-mbd rd -flags +mv4+aic -trellis 2 -cmp 2 -subcmp 2 -g 300 -pass 1/2&rsquo;,
 369 things to try: &rsquo;-bf 2&rsquo;, &rsquo;-flags qprd&rsquo;, &rsquo;-flags mv0&rsquo;, &rsquo;-flags skiprd&rsquo;.
 370 </p>
 371 <a name="Which-are-good-parameters-for-encoding-high-quality-MPEG_002d1_002fMPEG_002d2_003f"></a>
 372 <h3 class="section">3.10 Which are good parameters for encoding high quality MPEG-1/MPEG-2?<span class="pull-right"><a class="anchor hidden-xs" href="#Which-are-good-parameters-for-encoding-high-quality-MPEG_002d1_002fMPEG_002d2_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Which-are-good-parameters-for-encoding-high-quality-MPEG_002d1_002fMPEG_002d2_003f" aria-hidden="true">TOC</a></span></h3>
 373
 374 <p>&rsquo;-mbd rd -trellis 2 -cmp 2 -subcmp 2 -g 100 -pass 1/2&rsquo;
 375 but beware the &rsquo;-g 100&rsquo; might cause problems with some decoders.
 376 Things to try: &rsquo;-bf 2&rsquo;, &rsquo;-flags qprd&rsquo;, &rsquo;-flags mv0&rsquo;, &rsquo;-flags skiprd.
 377 </p>
 378 <a name="Interlaced-video-looks-very-bad-when-encoded-with-ffmpeg_002c-what-is-wrong_003f"></a>
 379 <h3 class="section">3.11 Interlaced video looks very bad when encoded with ffmpeg, what is wrong?<span class="pull-right"><a class="anchor hidden-xs" href="#Interlaced-video-looks-very-bad-when-encoded-with-ffmpeg_002c-what-is-wrong_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Interlaced-video-looks-very-bad-when-encoded-with-ffmpeg_002c-what-is-wrong_003f" aria-hidden="true">TOC</a></span></h3>
 380
 381 <p>You should use &rsquo;-flags +ilme+ildct&rsquo; and maybe &rsquo;-flags +alt&rsquo; for interlaced
 382 material, and try &rsquo;-top 0/1&rsquo; if the result looks really messed-up.
 383 </p>
 384 <a name="How-can-I-read-DirectShow-files_003f"></a>
 385 <h3 class="section">3.12 How can I read DirectShow files?<span class="pull-right"><a class="anchor hidden-xs" href="#How-can-I-read-DirectShow-files_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-can-I-read-DirectShow-files_003f" aria-hidden="true">TOC</a></span></h3>
 386
 387 <p>If you have built FFmpeg with <code>./configure --enable-avisynth</code>
 388 (only possible on MinGW/Cygwin platforms),
 389 then you may use any file that DirectShow can read as input.
 390 </p>
 391 <p>Just create an &quot;input.avs&quot; text file with this single line ...
 392 </p><div class="example">
 393 <pre class="example">DirectShowSource(&quot;C:\path to your file\yourfile.asf&quot;)
 394 </pre></div>
 395 <p>... and then feed that text file to ffmpeg:
 396 </p><div class="example">
 397 <pre class="example">ffmpeg -i input.avs
 398 </pre></div>
 399
 400 <p>For ANY other help on AviSynth, please visit the
 401 <a href="http://www.avisynth.org/">AviSynth homepage</a>.
 402 </p>
 403 <a name="How-can-I-join-video-files_003f"></a>
 404 <h3 class="section">3.13 How can I join video files?<span class="pull-right"><a class="anchor hidden-xs" href="#How-can-I-join-video-files_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-can-I-join-video-files_003f" aria-hidden="true">TOC</a></span></h3>
 405
 406 <p>To &quot;join&quot; video files is quite ambiguous. The following list explains the
 407 different kinds of &quot;joining&quot; and points out how those are addressed in
 408 FFmpeg. To join video files may mean:
 409 </p>
 410 <ul>
 411 <li> To put them one after the other: this is called to <em>concatenate</em> them
 412 (in short: concat) and is addressed
 413 <a href="#How-can-I-concatenate-video-files">in this very faq</a>.
 414
 415 </li><li> To put them together in the same file, to let the user choose between the
 416 different versions (example: different audio languages): this is called to
 417 <em>multiplex</em> them together (in short: mux), and is done by simply
 418 invoking ffmpeg with several <samp>-i</samp> options.
 419
 420 </li><li> For audio, to put all channels together in a single stream (example: two
 421 mono streams into one stereo stream): this is sometimes called to
 422 <em>merge</em> them, and can be done using the
 423 <a href="http://ffmpeg.org/ffmpeg-filters.html#amerge"><code>amerge</code></a> filter.
 424
 425 </li><li> For audio, to play one on top of the other: this is called to <em>mix</em>
 426 them, and can be done by first merging them into a single stream and then
 427 using the <a href="http://ffmpeg.org/ffmpeg-filters.html#pan"><code>pan</code></a> filter to mix
 428 the channels at will.
 429
 430 </li><li> For video, to display both together, side by side or one on top of a part of
 431 the other; it can be done using the
 432 <a href="http://ffmpeg.org/ffmpeg-filters.html#overlay"><code>overlay</code></a> video filter.
 433
 434 </li></ul>
 435
 436 <a name="How-can-I-concatenate-video-files"></a><a name="How-can-I-concatenate-video-files_003f"></a>
 437 <h3 class="section">3.14 How can I concatenate video files?<span class="pull-right"><a class="anchor hidden-xs" href="#How-can-I-concatenate-video-files_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-can-I-concatenate-video-files_003f" aria-hidden="true">TOC</a></span></h3>
 438
 439 <p>There are several solutions, depending on the exact circumstances.
 440 </p>
 441 <a name="Concatenating-using-the-concat-filter"></a>
 442 <h4 class="subsection">3.14.1 Concatenating using the concat <em>filter</em><span class="pull-right"><a class="anchor hidden-xs" href="#Concatenating-using-the-concat-filter" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Concatenating-using-the-concat-filter" aria-hidden="true">TOC</a></span></h4>
 443
 444 <p>FFmpeg has a <a href="http://ffmpeg.org/ffmpeg-filters.html#concat"><code>concat</code></a> filter designed specifically for that, with examples in the
 445 documentation. This operation is recommended if you need to re-encode.
 446 </p>
 447 <a name="Concatenating-using-the-concat-demuxer"></a>
 448 <h4 class="subsection">3.14.2 Concatenating using the concat <em>demuxer</em><span class="pull-right"><a class="anchor hidden-xs" href="#Concatenating-using-the-concat-demuxer" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Concatenating-using-the-concat-demuxer" aria-hidden="true">TOC</a></span></h4>
 449
 450 <p>FFmpeg has a <a href="http://www.ffmpeg.org/ffmpeg-formats.html#concat"><code>concat</code></a> demuxer which you can use when you want to avoid a re-encode and
 451 your format doesn&rsquo;t support file level concatenation.
 452 </p>
 453 <a name="Concatenating-using-the-concat-protocol-_0028file-level_0029"></a>
 454 <h4 class="subsection">3.14.3 Concatenating using the concat <em>protocol</em> (file level)<span class="pull-right"><a class="anchor hidden-xs" href="#Concatenating-using-the-concat-protocol-_0028file-level_0029" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Concatenating-using-the-concat-protocol-_0028file-level_0029" aria-hidden="true">TOC</a></span></h4>
 455
 456 <p>FFmpeg has a <a href="http://ffmpeg.org/ffmpeg-protocols.html#concat"><code>concat</code></a> protocol designed specifically for that, with examples in the
 457 documentation.
 458 </p>
 459 <p>A few multimedia containers (MPEG-1, MPEG-2 PS, DV) allow one to concatenate
 460 video by merely concatenating the files containing them.
 461 </p>
 462 <p>Hence you may concatenate your multimedia files by first transcoding them to
 463 these privileged formats, then using the humble <code>cat</code> command (or the
 464 equally humble <code>copy</code> under Windows), and finally transcoding back to your
 465 format of choice.
 466 </p>
 467 <div class="example">
 468 <pre class="example">ffmpeg -i input1.avi -qscale:v 1 intermediate1.mpg
 469 ffmpeg -i input2.avi -qscale:v 1 intermediate2.mpg
 470 cat intermediate1.mpg intermediate2.mpg &gt; intermediate_all.mpg
 471 ffmpeg -i intermediate_all.mpg -qscale:v 2 output.avi
 472 </pre></div>
 473
 474 <p>Additionally, you can use the <code>concat</code> protocol instead of <code>cat</code> or
 475 <code>copy</code> which will avoid creation of a potentially huge intermediate file.
 476 </p>
 477 <div class="example">
 478 <pre class="example">ffmpeg -i input1.avi -qscale:v 1 intermediate1.mpg
 479 ffmpeg -i input2.avi -qscale:v 1 intermediate2.mpg
 480 ffmpeg -i concat:&quot;intermediate1.mpg|intermediate2.mpg&quot; -c copy intermediate_all.mpg
 481 ffmpeg -i intermediate_all.mpg -qscale:v 2 output.avi
 482 </pre></div>
 483
 484 <p>Note that you may need to escape the character &quot;|&quot; which is special for many
 485 shells.
 486 </p>
 487 <p>Another option is usage of named pipes, should your platform support it:
 488 </p>
 489 <div class="example">
 490 <pre class="example">mkfifo intermediate1.mpg
 491 mkfifo intermediate2.mpg
 492 ffmpeg -i input1.avi -qscale:v 1 -y intermediate1.mpg &lt; /dev/null &amp;
 493 ffmpeg -i input2.avi -qscale:v 1 -y intermediate2.mpg &lt; /dev/null &amp;
 494 cat intermediate1.mpg intermediate2.mpg |\
 495 ffmpeg -f mpeg -i - -c:v mpeg4 -acodec libmp3lame output.avi
 496 </pre></div>
 497
 498 <a name="Concatenating-using-raw-audio-and-video"></a>
 499 <h4 class="subsection">3.14.4 Concatenating using raw audio and video<span class="pull-right"><a class="anchor hidden-xs" href="#Concatenating-using-raw-audio-and-video" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Concatenating-using-raw-audio-and-video" aria-hidden="true">TOC</a></span></h4>
 500
 501 <p>Similarly, the yuv4mpegpipe format, and the raw video, raw audio codecs also
 502 allow concatenation, and the transcoding step is almost lossless.
 503 When using multiple yuv4mpegpipe(s), the first line needs to be discarded
 504 from all but the first stream. This can be accomplished by piping through
 505 <code>tail</code> as seen below. Note that when piping through <code>tail</code> you
 506 must use command grouping, <code>{  ;}</code>, to background properly.
 507 </p>
 508 <p>For example, let&rsquo;s say we want to concatenate two FLV files into an
 509 output.flv file:
 510 </p>
 511 <div class="example">
 512 <pre class="example">mkfifo temp1.a
 513 mkfifo temp1.v
 514 mkfifo temp2.a
 515 mkfifo temp2.v
 516 mkfifo all.a
 517 mkfifo all.v
 518 ffmpeg -i input1.flv -vn -f u16le -acodec pcm_s16le -ac 2 -ar 44100 - &gt; temp1.a &lt; /dev/null &amp;
 519 ffmpeg -i input2.flv -vn -f u16le -acodec pcm_s16le -ac 2 -ar 44100 - &gt; temp2.a &lt; /dev/null &amp;
 520 ffmpeg -i input1.flv -an -f yuv4mpegpipe - &gt; temp1.v &lt; /dev/null &amp;
 521 { ffmpeg -i input2.flv -an -f yuv4mpegpipe - &lt; /dev/null | tail -n +2 &gt; temp2.v ; } &amp;
 522 cat temp1.a temp2.a &gt; all.a &amp;
 523 cat temp1.v temp2.v &gt; all.v &amp;
 524 ffmpeg -f u16le -acodec pcm_s16le -ac 2 -ar 44100 -i all.a \
 525        -f yuv4mpegpipe -i all.v \
 526        -y output.flv
 527 rm temp[12].[av] all.[av]
 528 </pre></div>
 529
 530 <a name="Using-_002df-lavfi_002c-audio-becomes-mono-for-no-apparent-reason_002e"></a>
 531 <h3 class="section">3.15 Using <samp>-f lavfi</samp>, audio becomes mono for no apparent reason.<span class="pull-right"><a class="anchor hidden-xs" href="#Using-_002df-lavfi_002c-audio-becomes-mono-for-no-apparent-reason_002e" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Using-_002df-lavfi_002c-audio-becomes-mono-for-no-apparent-reason_002e" aria-hidden="true">TOC</a></span></h3>
 532
 533 <p>Use <samp>-dumpgraph -</samp> to find out exactly where the channel layout is
 534 lost.
 535 </p>
 536 <p>Most likely, it is through <code>auto-inserted aresample</code>. Try to understand
 537 why the converting filter was needed at that place.
 538 </p>
 539 <p>Just before the output is a likely place, as <samp>-f lavfi</samp> currently
 540 only support packed S16.
 541 </p>
 542 <p>Then insert the correct <code>aformat</code> explicitly in the filtergraph,
 543 specifying the exact format.
 544 </p>
 545 <div class="example">
 546 <pre class="example">aformat=sample_fmts=s16:channel_layouts=stereo
 547 </pre></div>
 548
 549 <a name="Why-does-FFmpeg-not-see-the-subtitles-in-my-VOB-file_003f"></a>
 550 <h3 class="section">3.16 Why does FFmpeg not see the subtitles in my VOB file?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-does-FFmpeg-not-see-the-subtitles-in-my-VOB-file_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-does-FFmpeg-not-see-the-subtitles-in-my-VOB-file_003f" aria-hidden="true">TOC</a></span></h3>
 551
 552 <p>VOB and a few other formats do not have a global header that describes
 553 everything present in the file. Instead, applications are supposed to scan
 554 the file to see what it contains. Since VOB files are frequently large, only
 555 the beginning is scanned. If the subtitles happen only later in the file,
 556 they will not be initially detected.
 557 </p>
 558 <p>Some applications, including the <code>ffmpeg</code> command-line tool, can only
 559 work with streams that were detected during the initial scan; streams that
 560 are detected later are ignored.
 561 </p>
 562 <p>The size of the initial scan is controlled by two options: <code>probesize</code>
 563 (default ~5 Mo) and <code>analyzeduration</code> (default 5,000,000 µs = 5 s). For
 564 the subtitle stream to be detected, both values must be large enough.
 565 </p>
 566 <a name="Why-was-the-ffmpeg-_002dsameq-option-removed_003f-What-to-use-instead_003f"></a>
 567 <h3 class="section">3.17 Why was the <code>ffmpeg</code> <samp>-sameq</samp> option removed? What to use instead?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-was-the-ffmpeg-_002dsameq-option-removed_003f-What-to-use-instead_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-was-the-ffmpeg-_002dsameq-option-removed_003f-What-to-use-instead_003f" aria-hidden="true">TOC</a></span></h3>
 568
 569 <p>The <samp>-sameq</samp> option meant &quot;same quantizer&quot;, and made sense only in a
 570 very limited set of cases. Unfortunately, a lot of people mistook it for
 571 &quot;same quality&quot; and used it in places where it did not make sense: it had
 572 roughly the expected visible effect, but achieved it in a very inefficient
 573 way.
 574 </p>
 575 <p>Each encoder has its own set of options to set the quality-vs-size balance,
 576 use the options for the encoder you are using to set the quality level to a
 577 point acceptable for your tastes. The most common options to do that are
 578 <samp>-qscale</samp> and <samp>-qmax</samp>, but you should peruse the documentation
 579 of the encoder you chose.
 580 </p>
 581 <a name="I-have-a-stretched-video_002c-why-does-scaling-does-not-fix-it_003f"></a>
 582 <h3 class="section">3.18 I have a stretched video, why does scaling does not fix it?<span class="pull-right"><a class="anchor hidden-xs" href="#I-have-a-stretched-video_002c-why-does-scaling-does-not-fix-it_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I-have-a-stretched-video_002c-why-does-scaling-does-not-fix-it_003f" aria-hidden="true">TOC</a></span></h3>
 583
 584 <p>A lot of video codecs and formats can store the <em>aspect ratio</em> of the
 585 video: this is the ratio between the width and the height of either the full
 586 image (DAR, display aspect ratio) or individual pixels (SAR, sample aspect
 587 ratio). For example, EGA screens at resolution 640×350 had 4:3 DAR and 35:48
 588 SAR.
 589 </p>
 590 <p>Most still image processing work with square pixels, i.e. 1:1 SAR, but a lot
 591 of video standards, especially from the analogic-numeric transition era, use
 592 non-square pixels.
 593 </p>
 594 <p>Most processing filters in FFmpeg handle the aspect ratio to avoid
 595 stretching the image: cropping adjusts the DAR to keep the SAR constant,
 596 scaling adjusts the SAR to keep the DAR constant.
 597 </p>
 598 <p>If you want to stretch, or “unstretch”, the image, you need to override the
 599 information with the
 600 <a href="http://ffmpeg.org/ffmpeg-filters.html#setdar_002c-setsar"><code>setdar or setsar filters</code></a>.
 601 </p>
 602 <p>Do not forget to examine carefully the original video to check whether the
 603 stretching comes from the image or from the aspect ratio information.
 604 </p>
 605 <p>For example, to fix a badly encoded EGA capture, use the following commands,
 606 either the first one to upscale to square pixels or the second one to set
 607 the correct aspect ratio or the third one to avoid transcoding (may not work
 608 depending on the format / codec / player / phase of the moon):
 609 </p>
 610 <div class="example">
 611 <pre class="example">ffmpeg -i ega_screen.nut -vf scale=640:480,setsar=1 ega_screen_scaled.nut
 612 ffmpeg -i ega_screen.nut -vf setdar=4/3 ega_screen_anamorphic.nut
 613 ffmpeg -i ega_screen.nut -aspect 4/3 -c copy ega_screen_overridden.nut
 614 </pre></div>
 615
 616 <a name="Development"></a>
 617 <h2 class="chapter">4 Development<span class="pull-right"><a class="anchor hidden-xs" href="#Development" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Development" aria-hidden="true">TOC</a></span></h2>
 618
 619 <a name="Are-there-examples-illustrating-how-to-use-the-FFmpeg-libraries_002c-particularly-libavcodec-and-libavformat_003f"></a>
 620 <h3 class="section">4.1 Are there examples illustrating how to use the FFmpeg libraries, particularly libavcodec and libavformat?<span class="pull-right"><a class="anchor hidden-xs" href="#Are-there-examples-illustrating-how-to-use-the-FFmpeg-libraries_002c-particularly-libavcodec-and-libavformat_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Are-there-examples-illustrating-how-to-use-the-FFmpeg-libraries_002c-particularly-libavcodec-and-libavformat_003f" aria-hidden="true">TOC</a></span></h3>
 621
 622 <p>Yes. Check the <samp>doc/examples</samp> directory in the source
 623 repository, also available online at:
 624 <a href="https://github.com/FFmpeg/FFmpeg/tree/master/doc/examples">https://github.com/FFmpeg/FFmpeg/tree/master/doc/examples</a>.
 625 </p>
 626 <p>Examples are also installed by default, usually in
 627 <code>$PREFIX/share/ffmpeg/examples</code>.
 628 </p>
 629 <p>Also you may read the Developers Guide of the FFmpeg documentation. Alternatively,
 630 examine the source code for one of the many open source projects that
 631 already incorporate FFmpeg at (<a href="projects.html">projects.html</a>).
 632 </p>
 633 <a name="Can-you-support-my-C-compiler-XXX_003f"></a>
 634 <h3 class="section">4.2 Can you support my C compiler XXX?<span class="pull-right"><a class="anchor hidden-xs" href="#Can-you-support-my-C-compiler-XXX_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Can-you-support-my-C-compiler-XXX_003f" aria-hidden="true">TOC</a></span></h3>
 635
 636 <p>It depends. If your compiler is C99-compliant, then patches to support
 637 it are likely to be welcome if they do not pollute the source code
 638 with <code>#ifdef</code>s related to the compiler.
 639 </p>
 640 <a name="Is-Microsoft-Visual-C_002b_002b-supported_003f"></a>
 641 <h3 class="section">4.3 Is Microsoft Visual C++ supported?<span class="pull-right"><a class="anchor hidden-xs" href="#Is-Microsoft-Visual-C_002b_002b-supported_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Is-Microsoft-Visual-C_002b_002b-supported_003f" aria-hidden="true">TOC</a></span></h3>
 642
 643 <p>Yes. Please see the <a href="platform.html">Microsoft Visual C++</a>
 644 section in the FFmpeg documentation.
 645 </p>
 646 <a name="Can-you-add-automake_002c-libtool-or-autoconf-support_003f"></a>
 647 <h3 class="section">4.4 Can you add automake, libtool or autoconf support?<span class="pull-right"><a class="anchor hidden-xs" href="#Can-you-add-automake_002c-libtool-or-autoconf-support_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Can-you-add-automake_002c-libtool-or-autoconf-support_003f" aria-hidden="true">TOC</a></span></h3>
 648
 649 <p>No. These tools are too bloated and they complicate the build.
 650 </p>
 651 <a name="Why-not-rewrite-FFmpeg-in-object_002doriented-C_002b_002b_003f"></a>
 652 <h3 class="section">4.5 Why not rewrite FFmpeg in object-oriented C++?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-not-rewrite-FFmpeg-in-object_002doriented-C_002b_002b_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-not-rewrite-FFmpeg-in-object_002doriented-C_002b_002b_003f" aria-hidden="true">TOC</a></span></h3>
 653
 654 <p>FFmpeg is already organized in a highly modular manner and does not need to
 655 be rewritten in a formal object language. Further, many of the developers
 656 favor straight C; it works for them. For more arguments on this matter,
 657 read <a href="http://www.tux.org/lkml/#s15">&quot;Programming Religion&quot;</a>.
 658 </p>
 659 <a name="Why-are-the-ffmpeg-programs-devoid-of-debugging-symbols_003f"></a>
 660 <h3 class="section">4.6 Why are the ffmpeg programs devoid of debugging symbols?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-are-the-ffmpeg-programs-devoid-of-debugging-symbols_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-are-the-ffmpeg-programs-devoid-of-debugging-symbols_003f" aria-hidden="true">TOC</a></span></h3>
 661
 662 <p>The build process creates <code>ffmpeg_g</code>, <code>ffplay_g</code>, etc. which
 663 contain full debug information. Those binaries are stripped to create
 664 <code>ffmpeg</code>, <code>ffplay</code>, etc. If you need the debug information, use
 665 the *_g versions.
 666 </p>
 667 <a name="I-do-not-like-the-LGPL_002c-can-I-contribute-code-under-the-GPL-instead_003f"></a>
 668 <h3 class="section">4.7 I do not like the LGPL, can I contribute code under the GPL instead?<span class="pull-right"><a class="anchor hidden-xs" href="#I-do-not-like-the-LGPL_002c-can-I-contribute-code-under-the-GPL-instead_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I-do-not-like-the-LGPL_002c-can-I-contribute-code-under-the-GPL-instead_003f" aria-hidden="true">TOC</a></span></h3>
 669
 670 <p>Yes, as long as the code is optional and can easily and cleanly be placed
 671 under #if CONFIG_GPL without breaking anything. So, for example, a new codec
 672 or filter would be OK under GPL while a bug fix to LGPL code would not.
 673 </p>
 674 <a name="I_0027m-using-FFmpeg-from-within-my-C-application-but-the-linker-complains-about-missing-symbols-from-the-libraries-themselves_002e"></a>
 675 <h3 class="section">4.8 I&rsquo;m using FFmpeg from within my C application but the linker complains about missing symbols from the libraries themselves.<span class="pull-right"><a class="anchor hidden-xs" href="#I_0027m-using-FFmpeg-from-within-my-C-application-but-the-linker-complains-about-missing-symbols-from-the-libraries-themselves_002e" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I_0027m-using-FFmpeg-from-within-my-C-application-but-the-linker-complains-about-missing-symbols-from-the-libraries-themselves_002e" aria-hidden="true">TOC</a></span></h3>
 676
 677 <p>FFmpeg builds static libraries by default. In static libraries, dependencies
 678 are not handled. That has two consequences. First, you must specify the
 679 libraries in dependency order: <code>-lavdevice</code> must come before
 680 <code>-lavformat</code>, <code>-lavutil</code> must come after everything else, etc.
 681 Second, external libraries that are used in FFmpeg have to be specified too.
 682 </p>
 683 <p>An easy way to get the full list of required libraries in dependency order
 684 is to use <code>pkg-config</code>.
 685 </p>
 686 <div class="example">
 687 <pre class="example">c99 -o program program.c $(pkg-config --cflags --libs libavformat libavcodec)
 688 </pre></div>
 689
 690 <p>See <samp>doc/example/Makefile</samp> and <samp>doc/example/pc-uninstalled</samp> for
 691 more details.
 692 </p>
 693 <a name="I_0027m-using-FFmpeg-from-within-my-C_002b_002b-application-but-the-linker-complains-about-missing-symbols-which-seem-to-be-available_002e"></a>
 694 <h3 class="section">4.9 I&rsquo;m using FFmpeg from within my C++ application but the linker complains about missing symbols which seem to be available.<span class="pull-right"><a class="anchor hidden-xs" href="#I_0027m-using-FFmpeg-from-within-my-C_002b_002b-application-but-the-linker-complains-about-missing-symbols-which-seem-to-be-available_002e" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I_0027m-using-FFmpeg-from-within-my-C_002b_002b-application-but-the-linker-complains-about-missing-symbols-which-seem-to-be-available_002e" aria-hidden="true">TOC</a></span></h3>
 695
 696 <p>FFmpeg is a pure C project, so to use the libraries within your C++ application
 697 you need to explicitly state that you are using a C library. You can do this by
 698 encompassing your FFmpeg includes using <code>extern &quot;C&quot;</code>.
 699 </p>
 700 <p>See <a href="http://www.parashift.com/c++-faq-lite/mixing-c-and-cpp.html#faq-32.3">http://www.parashift.com/c++-faq-lite/mixing-c-and-cpp.html#faq-32.3</a>
 701 </p>
 702 <a name="I_0027m-using-libavutil-from-within-my-C_002b_002b-application-but-the-compiler-complains-about-_0027UINT64_005fC_0027-was-not-declared-in-this-scope"></a>
 703 <h3 class="section">4.10 I&rsquo;m using libavutil from within my C++ application but the compiler complains about &rsquo;UINT64_C&rsquo; was not declared in this scope<span class="pull-right"><a class="anchor hidden-xs" href="#I_0027m-using-libavutil-from-within-my-C_002b_002b-application-but-the-compiler-complains-about-_0027UINT64_005fC_0027-was-not-declared-in-this-scope" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I_0027m-using-libavutil-from-within-my-C_002b_002b-application-but-the-compiler-complains-about-_0027UINT64_005fC_0027-was-not-declared-in-this-scope" aria-hidden="true">TOC</a></span></h3>
 704
 705 <p>FFmpeg is a pure C project using C99 math features, in order to enable C++
 706 to use them you have to append -D__STDC_CONSTANT_MACROS to your CXXFLAGS
 707 </p>
 708 <a name="I-have-a-file-in-memory-_002f-a-API-different-from-_002aopen_002f_002aread_002f-libc-how-do-I-use-it-with-libavformat_003f"></a>
 709 <h3 class="section">4.11 I have a file in memory / a API different from *open/*read/ libc how do I use it with libavformat?<span class="pull-right"><a class="anchor hidden-xs" href="#I-have-a-file-in-memory-_002f-a-API-different-from-_002aopen_002f_002aread_002f-libc-how-do-I-use-it-with-libavformat_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-I-have-a-file-in-memory-_002f-a-API-different-from-_002aopen_002f_002aread_002f-libc-how-do-I-use-it-with-libavformat_003f" aria-hidden="true">TOC</a></span></h3>
 710
 711 <p>You have to create a custom AVIOContext using <code>avio_alloc_context</code>,
 712 see <samp>libavformat/aviobuf.c</samp> in FFmpeg and <samp>libmpdemux/demux_lavf.c</samp> in MPlayer or MPlayer2 sources.
 713 </p>
 714 <a name="Where-is-the-documentation-about-ffv1_002c-msmpeg4_002c-asv1_002c-4xm_003f"></a>
 715 <h3 class="section">4.12 Where is the documentation about ffv1, msmpeg4, asv1, 4xm?<span class="pull-right"><a class="anchor hidden-xs" href="#Where-is-the-documentation-about-ffv1_002c-msmpeg4_002c-asv1_002c-4xm_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Where-is-the-documentation-about-ffv1_002c-msmpeg4_002c-asv1_002c-4xm_003f" aria-hidden="true">TOC</a></span></h3>
 716
 717 <p>see <a href="http://www.ffmpeg.org/~michael/">http://www.ffmpeg.org/~michael/</a>
 718 </p>
 719 <a name="How-do-I-feed-H_002e263_002dRTP-_0028and-other-codecs-in-RTP_0029-to-libavcodec_003f"></a>
 720 <h3 class="section">4.13 How do I feed H.263-RTP (and other codecs in RTP) to libavcodec?<span class="pull-right"><a class="anchor hidden-xs" href="#How-do-I-feed-H_002e263_002dRTP-_0028and-other-codecs-in-RTP_0029-to-libavcodec_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-How-do-I-feed-H_002e263_002dRTP-_0028and-other-codecs-in-RTP_0029-to-libavcodec_003f" aria-hidden="true">TOC</a></span></h3>
 721
 722 <p>Even if peculiar since it is network oriented, RTP is a container like any
 723 other. You have to <em>demux</em> RTP before feeding the payload to libavcodec.
 724 In this specific case please look at RFC 4629 to see how it should be done.
 725 </p>
 726 <a name="AVStream_002er_005fframe_005frate-is-wrong_002c-it-is-much-larger-than-the-frame-rate_002e"></a>
 727 <h3 class="section">4.14 AVStream.r_frame_rate is wrong, it is much larger than the frame rate.<span class="pull-right"><a class="anchor hidden-xs" href="#AVStream_002er_005fframe_005frate-is-wrong_002c-it-is-much-larger-than-the-frame-rate_002e" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-AVStream_002er_005fframe_005frate-is-wrong_002c-it-is-much-larger-than-the-frame-rate_002e" aria-hidden="true">TOC</a></span></h3>
 728
 729 <p><code>r_frame_rate</code> is NOT the average frame rate, it is the smallest frame rate
 730 that can accurately represent all timestamps. So no, it is not
 731 wrong if it is larger than the average!
 732 For example, if you have mixed 25 and 30 fps content, then <code>r_frame_rate</code>
 733 will be 150 (it is the least common multiple).
 734 If you are looking for the average frame rate, see <code>AVStream.avg_frame_rate</code>.
 735 </p>
 736 <a name="Why-is-make-fate-not-running-all-tests_003f"></a>
 737 <h3 class="section">4.15 Why is <code>make fate</code> not running all tests?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-is-make-fate-not-running-all-tests_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-is-make-fate-not-running-all-tests_003f" aria-hidden="true">TOC</a></span></h3>
 738
 739 <p>Make sure you have the fate-suite samples and the <code>SAMPLES</code> Make variable
 740 or <code>FATE_SAMPLES</code> environment variable or the <code>--samples</code>
 741 <code>configure</code> option is set to the right path.
 742 </p>
 743 <a name="Why-is-make-fate-not-finding-the-samples_003f"></a>
 744 <h3 class="section">4.16 Why is <code>make fate</code> not finding the samples?<span class="pull-right"><a class="anchor hidden-xs" href="#Why-is-make-fate-not-finding-the-samples_003f" aria-hidden="true">#</a> <a class="anchor hidden-xs"href="#toc-Why-is-make-fate-not-finding-the-samples_003f" aria-hidden="true">TOC</a></span></h3>
 745
 746 <p>Do you happen to have a <code>~</code> character in the samples path to indicate a
 747 home directory? The value is used in ways where the shell cannot expand it,
 748 causing FATE to not find files. Just replace <code>~</code> by the full path.
 749 </p>
 750
 751       <p style="font-size: small;">
 752         This document was generated using <a href="http://www.gnu.org/software/texinfo/"><em>makeinfo</em></a>.
 753       </p>
 754     </div>
 755   </body>
 756 </html>