libavcodec/libtheoraenc.c

   1 /*
   2  * Copyright (c) 2006 Paul Richards <paul.richards@gmail.com>
   3  *
   4  * This file is part of FFmpeg.
   5  *
   6  * FFmpeg is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * FFmpeg is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with FFmpeg; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19  */
  20
  21 /*!
  22  * \file libtheoraenc.c
  23  * \brief Theora encoder using libtheora.
  24  * \author Paul Richards <paul.richards@gmail.com>
  25  *
  26  * A lot of this is copy / paste from other output codecs in
  27  * libavcodec or pure guesswork (or both).
  28  *
  29  * I have used t_ prefixes on variables which are libtheora types
  30  * and o_ prefixes on variables which are libogg types.
  31  */
  32
  33 /* FFmpeg includes */
  34 #include "libavutil/intreadwrite.h"
  35 #include "libavutil/log.h"
  36 #include "avcodec.h"
  37
  38 /* libtheora includes */
  39 #include <theora/theora.h>
  40
  41 typedef struct TheoraContext{
  42     theora_state t_state;
  43 } TheoraContext;
  44
  45 /*!
  46     Concatenates an ogg_packet into the extradata.
  47 */
  48 static int concatenate_packet(unsigned int* offset, AVCodecContext* avc_context, const ogg_packet* packet)
  49 {
  50     const char* message = NULL;
  51     uint8_t* newdata = NULL;
  52     int newsize = avc_context->extradata_size + 2 + packet->bytes;
  53
  54     if (packet->bytes < 0) {
  55         message = "ogg_packet has negative size";
  56     } else if (packet->bytes > 0xffff) {
  57         message = "ogg_packet is larger than 65535 bytes";
  58     } else if (newsize < avc_context->extradata_size) {
  59         message = "extradata_size would overflow";
  60     } else {
  61         newdata = av_realloc(avc_context->extradata, newsize);
  62         if (newdata == NULL) {
  63             message = "av_realloc failed";
  64         }
  65     }
  66     if (message != NULL) {
  67         av_log(avc_context, AV_LOG_ERROR, "concatenate_packet failed: %s\n", message);
  68         return -1;
  69     }
  70
  71     avc_context->extradata = newdata;
  72     avc_context->extradata_size = newsize;
  73     AV_WB16(avc_context->extradata + (*offset), packet->bytes);
  74     *offset += 2;
  75     memcpy( avc_context->extradata + (*offset), packet->packet, packet->bytes );
  76     (*offset) += packet->bytes;
  77     return 0;
  78 }
  79
  80 static av_cold int encode_init(AVCodecContext* avc_context)
  81 {
  82     theora_info t_info;
  83     theora_comment t_comment;
  84     ogg_packet o_packet;
  85     unsigned int offset;
  86     TheoraContext *h = avc_context->priv_data;
  87
  88     /* Set up the theora_info struct */
  89     theora_info_init( &t_info );
  90     t_info.width = FFALIGN(avc_context->width, 16);
  91     t_info.height = FFALIGN(avc_context->height, 16);
  92     t_info.frame_width = avc_context->width;
  93     t_info.frame_height = avc_context->height;
  94     t_info.offset_x = 0;
  95     t_info.offset_y = avc_context->height & 0xf;
  96     /* Swap numerator and denominator as time_base in AVCodecContext gives the
  97      * time period between frames, but theora_info needs the framerate.  */
  98     t_info.fps_numerator = avc_context->time_base.den;
  99     t_info.fps_denominator = avc_context->time_base.num;
 100     if (avc_context->sample_aspect_ratio.num != 0) {
 101         t_info.aspect_numerator = avc_context->sample_aspect_ratio.num;
 102         t_info.aspect_denominator = avc_context->sample_aspect_ratio.den;
 103     } else {
 104         t_info.aspect_numerator = 1;
 105         t_info.aspect_denominator = 1;
 106     }
 107     t_info.colorspace = OC_CS_UNSPECIFIED;
 108     t_info.pixelformat = OC_PF_420;
 109     t_info.target_bitrate = avc_context->bit_rate;
 110     t_info.keyframe_frequency = avc_context->gop_size;
 111     t_info.keyframe_frequency_force = avc_context->gop_size;
 112     t_info.keyframe_mindistance = avc_context->keyint_min;
 113     t_info.quality = 0;
 114
 115     t_info.quick_p = 1;
 116     t_info.dropframes_p = 0;
 117     t_info.keyframe_auto_p = 1;
 118     t_info.keyframe_data_target_bitrate = t_info.target_bitrate * 1.5;
 119     t_info.keyframe_auto_threshold = 80;
 120     t_info.noise_sensitivity = 1;
 121     t_info.sharpness = 0;
 122
 123     /* Now initialise libtheora */
 124     if (theora_encode_init( &(h->t_state), &t_info ) != 0) {
 125         av_log(avc_context, AV_LOG_ERROR, "theora_encode_init failed\n");
 126         return -1;
 127     }
 128
 129     /* Clear up theora_info struct */
 130     theora_info_clear( &t_info );
 131
 132     /*
 133         Output first header packet consisting of theora
 134         header, comment, and tables.
 135
 136         Each one is prefixed with a 16bit size, then they
 137         are concatenated together into ffmpeg's extradata.
 138     */
 139     offset = 0;
 140
 141     /* Header */
 142     theora_encode_header( &(h->t_state), &o_packet );
 143     if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
 144         return -1;
 145     }
 146
 147     /* Comment */
 148     theora_comment_init( &t_comment );
 149     theora_encode_comment( &t_comment, &o_packet );
 150     if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
 151         return -1;
 152     }
 153     /* Clear up theora_comment struct before we reset the packet */
 154     theora_comment_clear( &t_comment );
 155     /* And despite documentation to the contrary, theora_comment_clear
 156      * does not release the packet */
 157     ogg_packet_clear(&o_packet);
 158
 159     /* Tables */
 160     theora_encode_tables( &(h->t_state), &o_packet );
 161     if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) {
 162         return -1;
 163     }
 164
 165     /* Set up the output AVFrame */
 166     avc_context->coded_frame= avcodec_alloc_frame();
 167
 168     return 0;
 169 }
 170
 171 static int encode_frame(
 172     AVCodecContext* avc_context,
 173     uint8_t *outbuf,
 174     int buf_size,
 175     void *data)
 176 {
 177     yuv_buffer t_yuv_buffer;
 178     TheoraContext *h = avc_context->priv_data;
 179     AVFrame *frame = data;
 180     ogg_packet o_packet;
 181     int result;
 182
 183     assert(avc_context->pix_fmt == PIX_FMT_YUV420P);
 184
 185     /* Copy planes to the theora yuv_buffer */
 186     if (frame->linesize[1] != frame->linesize[2]) {
 187         av_log(avc_context, AV_LOG_ERROR, "U and V stride differ\n");
 188         return -1;
 189     }
 190
 191     t_yuv_buffer.y_width = FFALIGN(avc_context->width, 16);
 192     t_yuv_buffer.y_height = FFALIGN(avc_context->height, 16);
 193     t_yuv_buffer.y_stride = frame->linesize[0];
 194     t_yuv_buffer.uv_width = t_yuv_buffer.y_width / 2;
 195     t_yuv_buffer.uv_height = t_yuv_buffer.y_height / 2;
 196     t_yuv_buffer.uv_stride = frame->linesize[1];
 197
 198     t_yuv_buffer.y = frame->data[0];
 199     t_yuv_buffer.u = frame->data[1];
 200     t_yuv_buffer.v = frame->data[2];
 201
 202     /* Now call into theora_encode_YUVin */
 203     result = theora_encode_YUVin( &(h->t_state), &t_yuv_buffer );
 204     if (result != 0) {
 205         const char* message;
 206         switch (result) {
 207             case -1:
 208                 message = "differing frame sizes";
 209                 break;
 210             case OC_EINVAL:
 211                 message = "encoder is not ready or is finished";
 212                 break;
 213             default:
 214                 message = "unknown reason";
 215                 break;
 216         }
 217         av_log(avc_context, AV_LOG_ERROR, "theora_encode_YUVin failed (%s) [%d]\n", message, result);
 218         return -1;
 219     }
 220
 221     /* Pick up returned ogg_packet */
 222     result = theora_encode_packetout( &(h->t_state), 0, &o_packet );
 223     switch (result) {
 224         case 0:
 225             /* No packet is ready */
 226             return 0;
 227         case 1:
 228             /* Success, we have a packet */
 229             break;
 230         default:
 231             av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed [%d]\n", result);
 232             return -1;
 233     }
 234
 235     /* Copy ogg_packet content out to buffer */
 236     if (buf_size < o_packet.bytes) {
 237         av_log(avc_context, AV_LOG_ERROR, "encoded frame too large\n");
 238         return -1;
 239     }
 240     memcpy(outbuf, o_packet.packet, o_packet.bytes);
 241
 242     return o_packet.bytes;
 243 }
 244
 245 static av_cold int encode_close(AVCodecContext* avc_context)
 246 {
 247     ogg_packet o_packet;
 248     TheoraContext *h = avc_context->priv_data;
 249     int result;
 250     const char* message;
 251
 252     result = theora_encode_packetout( &(h->t_state), 1, &o_packet );
 253     theora_clear( &(h->t_state) );
 254     av_freep(&avc_context->coded_frame);
 255     av_freep(&avc_context->extradata);
 256     avc_context->extradata_size = 0;
 257
 258     switch (result) {
 259         case 0:/* No packet is ready */
 260         case -1:/* Encoding finished */
 261             return 0;
 262         case 1:
 263             /* We have a packet */
 264             message = "gave us a packet";
 265             break;
 266         default:
 267             message = "unknown reason";
 268             break;
 269     }
 270     av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed (%s) [%d]\n", message, result);
 271     return -1;
 272 }
 273
 274 static const enum PixelFormat supported_pixel_formats[] = { PIX_FMT_YUV420P, PIX_FMT_NONE };
 275
 276 /*! AVCodec struct exposed to libavcodec */
 277 AVCodec libtheora_encoder =
 278 {
 279     .name = "libtheora",
 280     .type = CODEC_TYPE_VIDEO,
 281     .id = CODEC_ID_THEORA,
 282     .priv_data_size = sizeof(TheoraContext),
 283     .init = encode_init,
 284     .close = encode_close,
 285     .encode = encode_frame,
 286     .pix_fmts = supported_pixel_formats,
 287     .long_name = NULL_IF_CONFIG_SMALL("libtheora Theora"),
 288 };