2 Copyright (C) 2017-2018 Carl Hetherington <cth@carlh.net>
4 This file is part of DCP-o-matic.
6 DCP-o-matic is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2 of the License, or
9 (at your option) any later version.
11 DCP-o-matic 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
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with DCP-o-matic. If not, see <http://www.gnu.org/licenses/>.
21 #include "ffmpeg_encoder.h"
25 #include "player_video.h"
29 #include "compose.hpp"
35 using std::runtime_error;
38 using boost::shared_ptr;
40 using boost::weak_ptr;
42 int FFmpegFileEncoder::_video_stream_index = 0;
43 int FFmpegFileEncoder::_audio_stream_index = 1;
45 FFmpegFileEncoder::FFmpegFileEncoder (
46 dcp::Size video_frame_size,
52 boost::filesystem::path output
55 , _audio_channels (channels)
57 , _video_frame_size (video_frame_size)
58 , _video_frame_rate (video_frame_rate)
59 , _audio_frame_rate (audio_frame_rate)
61 _pixel_format = pixel_format (format);
64 case EXPORT_FORMAT_PRORES:
65 _sample_format = AV_SAMPLE_FMT_S16;
66 _video_codec_name = "prores_ks";
67 _audio_codec_name = "pcm_s16le";
68 av_dict_set (&_video_options, "profile", "3", 0);
69 av_dict_set (&_video_options, "threads", "auto", 0);
71 case EXPORT_FORMAT_H264:
72 _sample_format = AV_SAMPLE_FMT_FLTP;
73 _video_codec_name = "libx264";
74 _audio_codec_name = "aac";
75 av_dict_set_int (&_video_options, "crf", x264_crf, 0);
82 avformat_alloc_output_context2 (&_format_context, 0, 0, _output.string().c_str());
83 if (!_format_context) {
84 throw runtime_error ("could not allocate FFmpeg format context");
87 _video_stream = avformat_new_stream (_format_context, _video_codec);
89 throw runtime_error ("could not create FFmpeg output video stream");
92 _audio_stream = avformat_new_stream (_format_context, _audio_codec);
94 throw runtime_error ("could not create FFmpeg output audio stream");
97 _video_stream->id = _video_stream_index;
98 _video_stream->codec = _video_codec_context;
100 _audio_stream->id = _audio_stream_index;
101 _audio_stream->codec = _audio_codec_context;
103 if (avcodec_open2 (_video_codec_context, _video_codec, &_video_options) < 0) {
104 throw runtime_error ("could not open FFmpeg video codec");
107 int r = avcodec_open2 (_audio_codec_context, _audio_codec, 0);
110 av_strerror (r, buffer, sizeof(buffer));
111 throw runtime_error (String::compose ("could not open FFmpeg audio codec (%1)", buffer));
114 if (avio_open_boost (&_format_context->pb, _output, AVIO_FLAG_WRITE) < 0) {
115 throw runtime_error ("could not open FFmpeg output file");
118 if (avformat_write_header (_format_context, 0) < 0) {
119 throw runtime_error ("could not write header to FFmpeg output file");
122 _pending_audio.reset (new AudioBuffers(channels, 0));
126 FFmpegFileEncoder::pixel_format (ExportFormat format)
129 case EXPORT_FORMAT_PRORES:
130 return AV_PIX_FMT_YUV422P10;
131 case EXPORT_FORMAT_H264:
132 return AV_PIX_FMT_YUV420P;
134 DCPOMATIC_ASSERT (false);
137 return AV_PIX_FMT_YUV422P10;
141 FFmpegFileEncoder::setup_video ()
143 _video_codec = avcodec_find_encoder_by_name (_video_codec_name.c_str());
145 throw runtime_error (String::compose ("could not find FFmpeg encoder %1", _video_codec_name));
148 _video_codec_context = avcodec_alloc_context3 (_video_codec);
149 if (!_video_codec_context) {
150 throw runtime_error ("could not allocate FFmpeg video context");
153 avcodec_get_context_defaults3 (_video_codec_context, _video_codec);
155 /* Variable quantisation */
156 _video_codec_context->global_quality = 0;
157 _video_codec_context->width = _video_frame_size.width;
158 _video_codec_context->height = _video_frame_size.height;
159 _video_codec_context->time_base = (AVRational) { 1, _video_frame_rate };
160 _video_codec_context->pix_fmt = _pixel_format;
161 _video_codec_context->flags |= AV_CODEC_FLAG_QSCALE | AV_CODEC_FLAG_GLOBAL_HEADER;
165 FFmpegFileEncoder::setup_audio ()
167 _audio_codec = avcodec_find_encoder_by_name (_audio_codec_name.c_str());
169 throw runtime_error (String::compose ("could not find FFmpeg encoder %1", _audio_codec_name));
172 _audio_codec_context = avcodec_alloc_context3 (_audio_codec);
173 if (!_audio_codec_context) {
174 throw runtime_error ("could not allocate FFmpeg audio context");
177 avcodec_get_context_defaults3 (_audio_codec_context, _audio_codec);
179 /* XXX: configurable */
180 _audio_codec_context->bit_rate = 256 * 1024;
181 _audio_codec_context->sample_fmt = _sample_format;
182 _audio_codec_context->sample_rate = _audio_frame_rate;
183 _audio_codec_context->channel_layout = av_get_default_channel_layout (_audio_channels);
184 _audio_codec_context->channels = _audio_channels;
188 FFmpegFileEncoder::flush ()
190 if (_pending_audio->frames() > 0) {
191 audio_frame (_pending_audio->frames ());
194 bool flushed_video = false;
195 bool flushed_audio = false;
197 while (!flushed_video || !flushed_audio) {
199 av_init_packet (&packet);
204 avcodec_encode_video2 (_video_codec_context, &packet, 0, &got_packet);
206 packet.stream_index = 0;
207 av_interleaved_write_frame (_format_context, &packet);
209 flushed_video = true;
211 av_packet_unref (&packet);
213 av_init_packet (&packet);
217 avcodec_encode_audio2 (_audio_codec_context, &packet, 0, &got_packet);
219 packet.stream_index = 0;
220 av_interleaved_write_frame (_format_context, &packet);
222 flushed_audio = true;
224 av_packet_unref (&packet);
227 av_write_trailer (_format_context);
229 avcodec_close (_video_codec_context);
230 avcodec_close (_audio_codec_context);
231 avio_close (_format_context->pb);
232 avformat_free_context (_format_context);
236 FFmpegFileEncoder::video (shared_ptr<PlayerVideo> video, DCPTime time)
238 shared_ptr<Image> image = video->image (
239 bind (&PlayerVideo::force, _1, _pixel_format),
244 AVFrame* frame = av_frame_alloc ();
245 DCPOMATIC_ASSERT (frame);
247 _pending_images[image->data()[0]] = image;
248 for (int i = 0; i < 3; ++i) {
249 AVBufferRef* buffer = av_buffer_create(image->data()[i], image->stride()[i] * image->size().height, &buffer_free, this, 0);
250 frame->buf[i] = av_buffer_ref (buffer);
251 frame->data[i] = buffer->data;
252 frame->linesize[i] = image->stride()[i];
253 av_buffer_unref (&buffer);
256 frame->width = image->size().width;
257 frame->height = image->size().height;
258 frame->format = _pixel_format;
259 frame->pts = time.seconds() / av_q2d (_video_stream->time_base);
262 av_init_packet (&packet);
267 if (avcodec_encode_video2 (_video_codec_context, &packet, frame, &got_packet) < 0) {
268 throw EncodeError ("FFmpeg video encode failed");
271 if (got_packet && packet.size) {
272 packet.stream_index = _video_stream_index;
273 av_interleaved_write_frame (_format_context, &packet);
274 av_packet_unref (&packet);
277 av_frame_free (&frame);
281 /** Called when the player gives us some audio */
283 FFmpegFileEncoder::audio (shared_ptr<AudioBuffers> audio)
285 _pending_audio->append (audio);
287 int frame_size = _audio_codec_context->frame_size;
288 if (frame_size == 0) {
289 /* codec has AV_CODEC_CAP_VARIABLE_FRAME_SIZE */
290 frame_size = _audio_frame_rate / _video_frame_rate;
293 while (_pending_audio->frames() >= frame_size) {
294 audio_frame (frame_size);
299 FFmpegFileEncoder::audio_frame (int size)
301 DCPOMATIC_ASSERT (size);
303 AVFrame* frame = av_frame_alloc ();
304 DCPOMATIC_ASSERT (frame);
306 int const channels = _pending_audio->channels();
307 DCPOMATIC_ASSERT (channels);
309 int const buffer_size = av_samples_get_buffer_size (0, channels, size, _audio_codec_context->sample_fmt, 0);
310 DCPOMATIC_ASSERT (buffer_size >= 0);
312 void* samples = av_malloc (buffer_size);
313 DCPOMATIC_ASSERT (samples);
315 frame->nb_samples = size;
316 int r = avcodec_fill_audio_frame (frame, channels, _audio_codec_context->sample_fmt, (const uint8_t *) samples, buffer_size, 0);
317 DCPOMATIC_ASSERT (r >= 0);
319 float** p = _pending_audio->data ();
320 switch (_audio_codec_context->sample_fmt) {
321 case AV_SAMPLE_FMT_S16:
323 int16_t* q = reinterpret_cast<int16_t*> (samples);
324 for (int i = 0; i < size; ++i) {
325 for (int j = 0; j < channels; ++j) {
326 *q++ = p[j][i] * 32767;
331 case AV_SAMPLE_FMT_FLTP:
333 float* q = reinterpret_cast<float*> (samples);
334 for (int i = 0; i < channels; ++i) {
335 memcpy (q, p[i], sizeof(float) * size);
341 DCPOMATIC_ASSERT (false);
345 av_init_packet (&packet);
350 if (avcodec_encode_audio2 (_audio_codec_context, &packet, frame, &got_packet) < 0) {
351 throw EncodeError ("FFmpeg audio encode failed");
354 if (got_packet && packet.size) {
355 packet.stream_index = _audio_stream_index;
356 av_interleaved_write_frame (_format_context, &packet);
357 av_packet_unref (&packet);
361 av_frame_free (&frame);
363 _pending_audio->trim_start (size);
367 FFmpegFileEncoder::subtitle (PlayerText, DCPTimePeriod)
373 FFmpegFileEncoder::buffer_free (void* opaque, uint8_t* data)
375 reinterpret_cast<FFmpegFileEncoder*>(opaque)->buffer_free2(data);
379 FFmpegFileEncoder::buffer_free2 (uint8_t* data)
381 _pending_images.erase (data);