Theora encoding via libtheora.

courtesy of Paul Richards, paul.richards gmail com


git-svn-id: file:///var/local/repositories/ffmpeg/trunk@7698 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Changelog

@@ -68,6 +68,7 @@ version <next>
 - Intel Music decoder
 - Musepack decoder
 - Flash Screen Video encoder
+- Theora encoding via libtheora
 
 version 0.4.9-pre1:
 

configure

@@ -89,6 +89,7 @@ show_help(){
   echo "  --enable-mp3lame         enable MP3 encoding via libmp3lame [default=no]"
   echo "  --enable-libnut          enable NUT support via libnut [default=no]"
   echo "  --enable-libogg          enable Ogg support via libogg [default=no]"
+  echo "  --enable-libtheora       enable Theora support via libtheora [default=no]"
   echo "  --enable-vorbis          enable Vorbis support via libvorbis [default=no]"
   echo "  --enable-x264            enable H.264 encoding via x264 [default=no]"
   echo "  --enable-xvid            enable Xvid encoding via xvidcore [default=no]"
@@ -444,6 +445,7 @@ CONFIG_LIST='
     libmp3lame
     libnut
     libogg
+    libtheora
     libvorbis
     memalign_hack
     mpegaudio_hp
@@ -514,6 +516,7 @@ dts_decoder_deps="libdts"
 faac_encoder_deps="libfaac"
 libgsm_decoder_deps="libgsm"
 libgsm_encoder_deps="libgsm"
+libtheora_encoder_deps="libtheora"
 mp3lame_encoder_deps="libmp3lame"
 oggvorbis_decoder_deps="libvorbis"
 oggvorbis_encoder_deps="libvorbis"
@@ -651,6 +654,7 @@ libgsm="no"
 libmp3lame="no"
 libnut="no"
 libogg="no"
+libtheora="no"
 libvorbis="no"
 xvid="no"
 x264="no"
@@ -1003,6 +1007,9 @@ for opt do
   --enable-libogg) libogg="yes"
     pkg_requires="$pkg_requires ogg >= 1.1"
   ;;
+  --enable-libtheora) libtheora="yes"
+    pkg_requires="$pkg_requires theora"
+  ;;
   --enable-vorbis) libvorbis="yes"
     pkg_requires="$pkg_requires vorbis vorbisenc"
   ;;
@@ -1179,6 +1186,10 @@ EOF
     exit 1;
 fi
 
+if test "$libtheora" = "yes" && test "$libogg" = "no"; then
+    die "libogg must be enabled to enable libtheora."
+fi
+
 if test "$libvorbis" = "yes" && test "$libogg" = "no"; then
     die "libogg must be enabled to enable libvorbis."
 fi
@@ -1532,6 +1543,7 @@ done
 enabled libdts  && require libdts dts.h dts_init -ldts -lm
 enabled libgsm  && require libgsm gsm.h gsm_create -lgsm
 enabled libmp3lame && require LAME lame/lame.h lame_init -lmp3lame -lm
+enabled libtheora && require libtheora theora/theora.h theora_info_init -ltheora -logg
 enabled libvorbis && require libvorbis vorbis/vorbisenc.h vorbis_info_init -lvorbis -lvorbisenc -logg
 enabled libogg  && require libogg ogg/ogg.h ogg_sync_init -logg
 enabled libnut  && require libnut libnut.h nut_demuxer_init -lnut
@@ -1868,6 +1880,7 @@ echo "libgsm enabled            $libgsm"
 echo "libmp3lame enabled        $libmp3lame"
 echo "libnut enabled            $libnut"
 echo "libogg enabled            $libogg"
+echo "libtheora enabled         $libtheora"
 echo "libvorbis enabled         $libvorbis"
 echo "x264 enabled              $x264"
 echo "XviD enabled              $xvid"

doc/ffmpeg-doc.texi

@@ -969,7 +969,7 @@ following image formats are supported:
 @item On2 VP3                @tab     @tab  X @tab still experimental
 @item On2 VP5                @tab     @tab  X @tab fourcc: VP50
 @item On2 VP6                @tab     @tab  X @tab fourcc: VP60,VP61,VP62
-@item Theora                 @tab     @tab  X @tab still experimental
+@item Theora                 @tab  X  @tab  X @tab still experimental
 @item Intel Indeo 3          @tab     @tab  X
 @item FLV                    @tab  X  @tab  X @tab Sorenson H.263 used in Flash
 @item Flash Screen Video     @tab     @tab  X @tab fourcc: FSV1

libavcodec/Makefile

@@ -258,6 +258,7 @@ OBJS-$(CONFIG_LIBFAAC)                 += faac.o
 OBJS-$(CONFIG_LIBFAAD)                 += faad.o
 OBJS-$(CONFIG_LIBGSM)                  += libgsm.o
 OBJS-$(CONFIG_LIBMP3LAME)              += mp3lameaudio.o
+OBJS-$(CONFIG_LIBTHEORA)               += libtheoraenc.o
 OBJS-$(CONFIG_LIBVORBIS)               += oggvorbis.o
 OBJS-$(CONFIG_X264)                    += x264.o
 OBJS-$(CONFIG_XVID)                    += xvidff.o xvid_rc.o

libavcodec/allcodecs.c

@@ -164,6 +164,7 @@ void avcodec_register_all(void)
     REGISTER_ENCDEC (FLAC, flac);
     REGISTER_DECODER(IMC, imc);
     REGISTER_ENCDEC (LIBGSM, libgsm);
+    REGISTER_ENCODER(LIBTHEORA, libtheora);
     REGISTER_DECODER(MACE3, mace3);
     REGISTER_DECODER(MACE6, mace6);
     REGISTER_ENCDEC (MP2, mp2);

libavcodec/avcodec.h

@@ -2162,6 +2162,7 @@ extern AVCodec h264_encoder;
 extern AVCodec huffyuv_encoder;
 extern AVCodec jpegls_encoder;
 extern AVCodec libgsm_encoder;
+extern AVCodec libtheora_encoder;
 extern AVCodec ljpeg_encoder;
 extern AVCodec mdec_encoder;
 extern AVCodec mjpeg_encoder;

libavcodec/libtheoraenc.c

+/*
+ * Copyright (c) 2006 Paul Richards <paul.richards@gmail.com>
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+ */
+
+/*!
+ * \file theoraenc.c
+ * \brief Theora encoder using libtheora.
+ * \author Paul Richards <paul.richards@gmail.com>
+ *
+ * A lot of this is copy / paste from other output codecs in
+ * libavcodec or pure guesswork (or both).
+ *
+ * I have used t_ prefixes on variables which are libtheora types
+ * and o_ prefixes on variables which are libogg types.
+ */
+
+/* FFmpeg includes */
+#include "avcodec.h"
+#include "log.h"
+
+/* libtheora includes */
+#include <theora/theora.h>
+
+typedef struct TheoraContext{
+    theora_state t_state;
+} TheoraContext;
+
+/*!
+    Concatenates an ogg_packet into the extradata.
+*/
+static int concatenate_packet(unsigned int* offset, AVCodecContext* avc_context, const ogg_packet* packet)
+{
+    char* message = NULL;
+    uint8_t* newdata = NULL;
+    int newsize = avc_context->extradata_size + 2 + packet->bytes;
+
+    if (packet->bytes < 0) {
+        message = "ogg_packet has negative size";
+    } else if (packet->bytes > 0xffff) {
+        message = "ogg_packet is larger than 65535 bytes";
+    } else if (newsize < avc_context->extradata_size) {
+        message = "extradata_size would overflow";
+    } else {
+        newdata = av_realloc(avc_context->extradata, newsize);
+        if (newdata == NULL) {
+            message = "av_realloc failed";
+        }
+    }
+    if (message != NULL) {
+        av_log(avc_context, AV_LOG_ERROR, "concatenate_packet failed: %s\n", message);
+        return -1;
+    }
+
+    avc_context->extradata = newdata;
+    avc_context->extradata_size = newsize;
+    avc_context->extradata[ (*offset)++ ] = packet->bytes >> 8;
+    avc_context->extradata[ (*offset)++ ] = packet->bytes & 0xff;
+    memcpy( avc_context->extradata + (*offset), packet->packet, packet->bytes );
+    (*offset) += packet->bytes;
+    return 0;
+}
+
+static int encode_init(AVCodecContext* avc_context)
+{
+    theora_info t_info;
+    theora_comment t_comment;
+    ogg_packet o_packet;
+    unsigned int offset;
+    TheoraContext *h = avc_context->priv_data;
+
+    /* Set up the theora_info struct */
+    theora_info_init( &t_info );
+    t_info.width = avc_context->width;
+    t_info.height = avc_context->height;
+    t_info.frame_width = avc_context->width;
+    t_info.frame_height = avc_context->height;
+    t_info.offset_x = 0;
+    t_info.offset_y = 0;
+    t_info.fps_numerator = avc_context->time_base.den;
+    t_info.fps_denominator = avc_context->time_base.num;
+    if (avc_context->sample_aspect_ratio.num != 0) {
+        t_info.aspect_numerator = avc_context->sample_aspect_ratio.num;
+        t_info.aspect_denominator = avc_context->sample_aspect_ratio.den;
+    } else {
+        t_info.aspect_numerator = 1;
+        t_info.aspect_denominator = 1;
+    }
+    t_info.colorspace = OC_CS_UNSPECIFIED;
+    t_info.pixelformat = OC_PF_420;
+    t_info.target_bitrate = avc_context->bit_rate;
+    t_info.keyframe_frequency = avc_context->gop_size;
+    t_info.keyframe_frequency_force = avc_context->gop_size;
+    t_info.keyframe_mindistance = avc_context->keyint_min;
+    t_info.quality = 0;
+
+    t_info.quick_p = 1;
+    t_info.dropframes_p = 0;
+    t_info.keyframe_auto_p = 1;
+    t_info.keyframe_data_target_bitrate = t_info.target_bitrate * 1.5;
+    t_info.keyframe_auto_threshold = 80;
+    t_info.noise_sensitivity = 1;
+    t_info.sharpness = 0;
+
+    /* Now initialise libtheora */
+    if (theora_encode_init( &(h->t_state), &t_info ) != 0) {
+        av_log(avc_context, AV_LOG_ERROR, "theora_encode_init failed\n");
+        return -1;
+    }
+
+    /* Clear up theora_info struct */
+    theora_info_clear( &t_info );
+
+    /*
+        Output first header packet consisting of theora
+        header, comment, and tables.
+
+        Each one is prefixed with a 16bit size, then they
+        are concatenated together into ffmpeg's extradata.
+    */
+    offset = 0;
+
+    /* Header */
+    theora_encode_header( &(h->t_state), &o_packet );
+    if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
+        return -1;
+    }
+
+    /* Comment */
+    theora_comment_init( &t_comment );
+    theora_encode_comment( &t_comment, &o_packet );
+    if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
+        return -1;
+    }
+
+    /* Tables */
+    theora_encode_tables( &(h->t_state), &o_packet );
+    if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
+        return -1;
+    }
+
+    /* Clear up theora_comment struct */
+    theora_comment_clear( &t_comment );
+
+    /* Set up the output AVFrame */
+    avc_context->coded_frame= avcodec_alloc_frame();
+
+    return 0;
+}
+
+static int encode_frame(
+    AVCodecContext* avc_context,
+    uint8_t *outbuf,
+    int buf_size,
+    void *data)
+{
+    yuv_buffer t_yuv_buffer;
+    TheoraContext *h = avc_context->priv_data;
+    AVFrame *frame = data;
+    ogg_packet o_packet;
+    int result;
+
+    assert(avc_context->pix_fmt == PIX_FMT_YUV420P);
+
+    /* Copy planes to the theora yuv_buffer */
+    if (frame->linesize[1] != frame->linesize[2]) {
+        av_log(avc_context, AV_LOG_ERROR, "U and V stride differ\n");
+        return -1;
+    }
+
+    t_yuv_buffer.y_width = avc_context->width;
+    t_yuv_buffer.y_height = avc_context->height;
+    t_yuv_buffer.y_stride = frame->linesize[0];
+    t_yuv_buffer.uv_width = t_yuv_buffer.y_width / 2;
+    t_yuv_buffer.uv_height = t_yuv_buffer.y_height / 2;
+    t_yuv_buffer.uv_stride = frame->linesize[1];
+
+    t_yuv_buffer.y = frame->data[0];
+    t_yuv_buffer.u = frame->data[1];
+    t_yuv_buffer.v = frame->data[2];
+
+    /* Now call into theora_encode_YUVin */
+    result = theora_encode_YUVin( &(h->t_state), &t_yuv_buffer );
+    if (result != 0) {
+        const char* message;
+        switch (result) {
+            case -1:
+                message = "differing frame sizes";
+                break;
+            case OC_EINVAL:
+                message = "encoder is not ready or is finished";
+                break;
+            default:
+                message = "unknown reason";
+                break;
+        }
+        av_log(avc_context, AV_LOG_ERROR, "theora_encode_YUVin failed (%s) [%d]\n", message, result);
+        return -1;
+    }
+
+    /* Pick up returned ogg_packet */
+    result = theora_encode_packetout( &(h->t_state), 0, &o_packet );
+    switch (result) {
+        case 0:
+            /* No packet is ready */
+            return 0;
+        case 1:
+            /* Success, we have a packet */
+            break;
+        default:
+            av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed [%d]\n", result);
+            return -1;
+    }
+
+    /* Copy ogg_packet content out to buffer */
+    if (buf_size < o_packet.bytes) {
+        av_log(avc_context, AV_LOG_ERROR, "encoded frame too large\n");
+        return -1;
+    }
+    memcpy(outbuf, o_packet.packet, o_packet.bytes);
+
+    return o_packet.bytes;
+}
+
+static int encode_close(AVCodecContext* avc_context)
+{
+    ogg_packet o_packet;
+    TheoraContext *h = avc_context->priv_data;
+    int result;
+    const char* message;
+
+    result = theora_encode_packetout( &(h->t_state), 1, &o_packet );
+    theora_clear( &(h->t_state) );
+    switch (result) {
+        case 0:/* No packet is ready */
+        case -1:/* Encoding finished */
+            return 0;
+        case 1:
+            /* We have a packet */
+            message = "gave us a packet";
+            break;
+        default:
+            message = "unknown reason";
+            break;
+    }
+    av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed (%s) [%d]\n", message, result);
+    return -1;
+}
+
+static const enum PixelFormat supported_pixel_formats[] = { PIX_FMT_YUV420P, -1 };
+
+/*! AVCodec struct exposed to libavcodec */
+AVCodec libtheora_encoder =
+{
+    .name = "libtheora",
+    .type = CODEC_TYPE_VIDEO,
+    .id = CODEC_ID_THEORA,
+    .priv_data_size = sizeof(TheoraContext),
+    .init = encode_init,
+    .close = encode_close,
+    .encode = encode_frame,
+    .pix_fmts = supported_pixel_formats,
+};