2 * Copyright (c) 2006 Paul Richards <paul.richards@gmail.com>
4 * This file is part of FFmpeg.
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.
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.
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
22 * \file libtheoraenc.c
23 * \brief Theora encoder using libtheora.
24 * \author Paul Richards <paul.richards@gmail.com>
26 * A lot of this is copy / paste from other output codecs in
27 * libavcodec or pure guesswork (or both).
29 * I have used t_ prefixes on variables which are libtheora types
30 * and o_ prefixes on variables which are libogg types.
34 #include "libavutil/intreadwrite.h"
35 #include "libavutil/log.h"
38 /* libtheora includes */
39 #include <theora/theora.h>
41 typedef struct TheoraContext
{
46 Concatenates an ogg_packet into the extradata.
48 static int concatenate_packet(unsigned int* offset
, AVCodecContext
* avc_context
, const ogg_packet
* packet
)
50 const char* message
= NULL
;
51 uint8_t* newdata
= NULL
;
52 int newsize
= avc_context
->extradata_size
+ 2 + packet
->bytes
;
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";
61 newdata
= av_realloc(avc_context
->extradata
, newsize
);
62 if (newdata
== NULL
) {
63 message
= "av_realloc failed";
66 if (message
!= NULL
) {
67 av_log(avc_context
, AV_LOG_ERROR
, "concatenate_packet failed: %s\n", message
);
71 avc_context
->extradata
= newdata
;
72 avc_context
->extradata_size
= newsize
;
73 AV_WB16(avc_context
->extradata
+ (*offset
), packet
->bytes
);
75 memcpy( avc_context
->extradata
+ (*offset
), packet
->packet
, packet
->bytes
);
76 (*offset
) += packet
->bytes
;
80 static av_cold
int encode_init(AVCodecContext
* avc_context
)
83 theora_comment t_comment
;
86 TheoraContext
*h
= avc_context
->priv_data
;
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
;
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
;
104 t_info
.aspect_numerator
= 1;
105 t_info
.aspect_denominator
= 1;
107 t_info
.colorspace
= OC_CS_UNSPECIFIED
;
108 t_info
.pixelformat
= OC_PF_420
;
109 t_info
.keyframe_frequency
= avc_context
->gop_size
;
110 t_info
.keyframe_frequency_force
= avc_context
->gop_size
;
111 t_info
.keyframe_mindistance
= avc_context
->keyint_min
;
114 t_info
.dropframes_p
= 0;
115 t_info
.keyframe_auto_p
= 1;
116 t_info
.keyframe_data_target_bitrate
= t_info
.target_bitrate
* 1.5;
117 t_info
.keyframe_auto_threshold
= 80;
118 t_info
.noise_sensitivity
= 1;
119 t_info
.sharpness
= 0;
121 if (avc_context
->flags
& CODEC_FLAG_QSCALE
) {
122 /* to be constant with the libvorbis implementation, clip global_quality to 0 - 10
123 Theora accepts a quality parameter p, which is:
127 t_info
.quality
= av_clip(avc_context
->global_quality
/ (float)FF_QP2LAMBDA
, 0, 10) * 6.3;
128 t_info
.target_bitrate
= 0;
130 t_info
.target_bitrate
= avc_context
->bit_rate
;
134 /* Now initialise libtheora */
135 if (theora_encode_init( &(h
->t_state
), &t_info
) != 0) {
136 av_log(avc_context
, AV_LOG_ERROR
, "theora_encode_init failed\n");
140 /* Clear up theora_info struct */
141 theora_info_clear( &t_info
);
144 Output first header packet consisting of theora
145 header, comment, and tables.
147 Each one is prefixed with a 16bit size, then they
148 are concatenated together into ffmpeg's extradata.
153 theora_encode_header( &(h
->t_state
), &o_packet
);
154 if (concatenate_packet( &offset
, avc_context
, &o_packet
) != 0) {
159 theora_comment_init( &t_comment
);
160 theora_encode_comment( &t_comment
, &o_packet
);
161 if (concatenate_packet( &offset
, avc_context
, &o_packet
) != 0) {
164 /* Clear up theora_comment struct before we reset the packet */
165 theora_comment_clear( &t_comment
);
166 /* And despite documentation to the contrary, theora_comment_clear
167 * does not release the packet */
168 ogg_packet_clear(&o_packet
);
171 theora_encode_tables( &(h
->t_state
), &o_packet
);
172 if (concatenate_packet( &offset
, avc_context
, &o_packet
) != 0) {
176 /* Set up the output AVFrame */
177 avc_context
->coded_frame
= avcodec_alloc_frame();
182 static int encode_frame(
183 AVCodecContext
* avc_context
,
188 yuv_buffer t_yuv_buffer
;
189 TheoraContext
*h
= avc_context
->priv_data
;
190 AVFrame
*frame
= data
;
194 assert(avc_context
->pix_fmt
== PIX_FMT_YUV420P
);
196 /* Copy planes to the theora yuv_buffer */
197 if (frame
->linesize
[1] != frame
->linesize
[2]) {
198 av_log(avc_context
, AV_LOG_ERROR
, "U and V stride differ\n");
202 t_yuv_buffer
.y_width
= FFALIGN(avc_context
->width
, 16);
203 t_yuv_buffer
.y_height
= FFALIGN(avc_context
->height
, 16);
204 t_yuv_buffer
.y_stride
= frame
->linesize
[0];
205 t_yuv_buffer
.uv_width
= t_yuv_buffer
.y_width
/ 2;
206 t_yuv_buffer
.uv_height
= t_yuv_buffer
.y_height
/ 2;
207 t_yuv_buffer
.uv_stride
= frame
->linesize
[1];
209 t_yuv_buffer
.y
= frame
->data
[0];
210 t_yuv_buffer
.u
= frame
->data
[1];
211 t_yuv_buffer
.v
= frame
->data
[2];
213 /* Now call into theora_encode_YUVin */
214 result
= theora_encode_YUVin( &(h
->t_state
), &t_yuv_buffer
);
219 message
= "differing frame sizes";
222 message
= "encoder is not ready or is finished";
225 message
= "unknown reason";
228 av_log(avc_context
, AV_LOG_ERROR
, "theora_encode_YUVin failed (%s) [%d]\n", message
, result
);
232 /* Pick up returned ogg_packet */
233 result
= theora_encode_packetout( &(h
->t_state
), 0, &o_packet
);
236 /* No packet is ready */
239 /* Success, we have a packet */
242 av_log(avc_context
, AV_LOG_ERROR
, "theora_encode_packetout failed [%d]\n", result
);
246 /* Copy ogg_packet content out to buffer */
247 if (buf_size
< o_packet
.bytes
) {
248 av_log(avc_context
, AV_LOG_ERROR
, "encoded frame too large\n");
251 memcpy(outbuf
, o_packet
.packet
, o_packet
.bytes
);
253 // HACK: does not take codec delay into account (neither does the decoder though)
254 avc_context
->coded_frame
->pts
= frame
->pts
;
256 return o_packet
.bytes
;
259 static av_cold
int encode_close(AVCodecContext
* avc_context
)
262 TheoraContext
*h
= avc_context
->priv_data
;
266 result
= theora_encode_packetout( &(h
->t_state
), 1, &o_packet
);
267 theora_clear( &(h
->t_state
) );
268 av_freep(&avc_context
->coded_frame
);
269 av_freep(&avc_context
->extradata
);
270 avc_context
->extradata_size
= 0;
273 case 0:/* No packet is ready */
274 case -1:/* Encoding finished */
277 /* We have a packet */
278 message
= "gave us a packet";
281 message
= "unknown reason";
284 av_log(avc_context
, AV_LOG_ERROR
, "theora_encode_packetout failed (%s) [%d]\n", message
, result
);
288 static const enum PixelFormat supported_pixel_formats
[] = { PIX_FMT_YUV420P
, PIX_FMT_NONE
};
290 /*! AVCodec struct exposed to libavcodec */
291 AVCodec libtheora_encoder
=
294 .type
= CODEC_TYPE_VIDEO
,
295 .id
= CODEC_ID_THEORA
,
296 .priv_data_size
= sizeof(TheoraContext
),
298 .close
= encode_close
,
299 .encode
= encode_frame
,
300 .pix_fmts
= supported_pixel_formats
,
301 .long_name
= NULL_IF_CONFIG_SMALL("libtheora Theora"),