Commit | Line | Data |
---|---|---|
150d2772 DB |
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 St, Fifth Floor, Boston, MA 02110-1301 USA | |
19 | */ | |
20 | ||
21 | /*! | |
22 | * \file theoraenc.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 "avcodec.h" | |
35 | #include "log.h" | |
36 | ||
37 | /* libtheora includes */ | |
38 | #include <theora/theora.h> | |
39 | ||
40 | typedef struct TheoraContext{ | |
41 | theora_state t_state; | |
42 | } TheoraContext; | |
43 | ||
44 | /*! | |
45 | Concatenates an ogg_packet into the extradata. | |
46 | */ | |
47 | static int concatenate_packet(unsigned int* offset, AVCodecContext* avc_context, const ogg_packet* packet) | |
48 | { | |
49 | char* message = NULL; | |
50 | uint8_t* newdata = NULL; | |
51 | int newsize = avc_context->extradata_size + 2 + packet->bytes; | |
52 | ||
53 | if (packet->bytes < 0) { | |
54 | message = "ogg_packet has negative size"; | |
55 | } else if (packet->bytes > 0xffff) { | |
56 | message = "ogg_packet is larger than 65535 bytes"; | |
57 | } else if (newsize < avc_context->extradata_size) { | |
58 | message = "extradata_size would overflow"; | |
59 | } else { | |
60 | newdata = av_realloc(avc_context->extradata, newsize); | |
61 | if (newdata == NULL) { | |
62 | message = "av_realloc failed"; | |
63 | } | |
64 | } | |
65 | if (message != NULL) { | |
66 | av_log(avc_context, AV_LOG_ERROR, "concatenate_packet failed: %s\n", message); | |
67 | return -1; | |
68 | } | |
69 | ||
70 | avc_context->extradata = newdata; | |
71 | avc_context->extradata_size = newsize; | |
72 | avc_context->extradata[ (*offset)++ ] = packet->bytes >> 8; | |
73 | avc_context->extradata[ (*offset)++ ] = packet->bytes & 0xff; | |
74 | memcpy( avc_context->extradata + (*offset), packet->packet, packet->bytes ); | |
75 | (*offset) += packet->bytes; | |
76 | return 0; | |
77 | } | |
78 | ||
79 | static int encode_init(AVCodecContext* avc_context) | |
80 | { | |
81 | theora_info t_info; | |
82 | theora_comment t_comment; | |
83 | ogg_packet o_packet; | |
84 | unsigned int offset; | |
85 | TheoraContext *h = avc_context->priv_data; | |
86 | ||
87 | /* Set up the theora_info struct */ | |
88 | theora_info_init( &t_info ); | |
89 | t_info.width = avc_context->width; | |
90 | t_info.height = avc_context->height; | |
91 | t_info.frame_width = avc_context->width; | |
92 | t_info.frame_height = avc_context->height; | |
93 | t_info.offset_x = 0; | |
94 | t_info.offset_y = 0; | |
95 | t_info.fps_numerator = avc_context->time_base.den; | |
96 | t_info.fps_denominator = avc_context->time_base.num; | |
97 | if (avc_context->sample_aspect_ratio.num != 0) { | |
98 | t_info.aspect_numerator = avc_context->sample_aspect_ratio.num; | |
99 | t_info.aspect_denominator = avc_context->sample_aspect_ratio.den; | |
100 | } else { | |
101 | t_info.aspect_numerator = 1; | |
102 | t_info.aspect_denominator = 1; | |
103 | } | |
104 | t_info.colorspace = OC_CS_UNSPECIFIED; | |
105 | t_info.pixelformat = OC_PF_420; | |
106 | t_info.target_bitrate = avc_context->bit_rate; | |
107 | t_info.keyframe_frequency = avc_context->gop_size; | |
108 | t_info.keyframe_frequency_force = avc_context->gop_size; | |
109 | t_info.keyframe_mindistance = avc_context->keyint_min; | |
110 | t_info.quality = 0; | |
111 | ||
112 | t_info.quick_p = 1; | |
113 | t_info.dropframes_p = 0; | |
114 | t_info.keyframe_auto_p = 1; | |
115 | t_info.keyframe_data_target_bitrate = t_info.target_bitrate * 1.5; | |
116 | t_info.keyframe_auto_threshold = 80; | |
117 | t_info.noise_sensitivity = 1; | |
118 | t_info.sharpness = 0; | |
119 | ||
120 | /* Now initialise libtheora */ | |
121 | if (theora_encode_init( &(h->t_state), &t_info ) != 0) { | |
122 | av_log(avc_context, AV_LOG_ERROR, "theora_encode_init failed\n"); | |
123 | return -1; | |
124 | } | |
125 | ||
126 | /* Clear up theora_info struct */ | |
127 | theora_info_clear( &t_info ); | |
128 | ||
129 | /* | |
130 | Output first header packet consisting of theora | |
131 | header, comment, and tables. | |
132 | ||
133 | Each one is prefixed with a 16bit size, then they | |
134 | are concatenated together into ffmpeg's extradata. | |
135 | */ | |
136 | offset = 0; | |
137 | ||
138 | /* Header */ | |
139 | theora_encode_header( &(h->t_state), &o_packet ); | |
140 | if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) { | |
141 | return -1; | |
142 | } | |
143 | ||
144 | /* Comment */ | |
145 | theora_comment_init( &t_comment ); | |
146 | theora_encode_comment( &t_comment, &o_packet ); | |
147 | if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) { | |
148 | return -1; | |
149 | } | |
150 | ||
151 | /* Tables */ | |
152 | theora_encode_tables( &(h->t_state), &o_packet ); | |
153 | if (concatenate_packet( &offset, avc_context, &o_packet ) != 0) { | |
154 | return -1; | |
155 | } | |
156 | ||
157 | /* Clear up theora_comment struct */ | |
158 | theora_comment_clear( &t_comment ); | |
159 | ||
160 | /* Set up the output AVFrame */ | |
161 | avc_context->coded_frame= avcodec_alloc_frame(); | |
162 | ||
163 | return 0; | |
164 | } | |
165 | ||
166 | static int encode_frame( | |
167 | AVCodecContext* avc_context, | |
168 | uint8_t *outbuf, | |
169 | int buf_size, | |
170 | void *data) | |
171 | { | |
172 | yuv_buffer t_yuv_buffer; | |
173 | TheoraContext *h = avc_context->priv_data; | |
174 | AVFrame *frame = data; | |
175 | ogg_packet o_packet; | |
176 | int result; | |
177 | ||
178 | assert(avc_context->pix_fmt == PIX_FMT_YUV420P); | |
179 | ||
180 | /* Copy planes to the theora yuv_buffer */ | |
181 | if (frame->linesize[1] != frame->linesize[2]) { | |
182 | av_log(avc_context, AV_LOG_ERROR, "U and V stride differ\n"); | |
183 | return -1; | |
184 | } | |
185 | ||
186 | t_yuv_buffer.y_width = avc_context->width; | |
187 | t_yuv_buffer.y_height = avc_context->height; | |
188 | t_yuv_buffer.y_stride = frame->linesize[0]; | |
189 | t_yuv_buffer.uv_width = t_yuv_buffer.y_width / 2; | |
190 | t_yuv_buffer.uv_height = t_yuv_buffer.y_height / 2; | |
191 | t_yuv_buffer.uv_stride = frame->linesize[1]; | |
192 | ||
193 | t_yuv_buffer.y = frame->data[0]; | |
194 | t_yuv_buffer.u = frame->data[1]; | |
195 | t_yuv_buffer.v = frame->data[2]; | |
196 | ||
197 | /* Now call into theora_encode_YUVin */ | |
198 | result = theora_encode_YUVin( &(h->t_state), &t_yuv_buffer ); | |
199 | if (result != 0) { | |
200 | const char* message; | |
201 | switch (result) { | |
202 | case -1: | |
203 | message = "differing frame sizes"; | |
204 | break; | |
205 | case OC_EINVAL: | |
206 | message = "encoder is not ready or is finished"; | |
207 | break; | |
208 | default: | |
209 | message = "unknown reason"; | |
210 | break; | |
211 | } | |
212 | av_log(avc_context, AV_LOG_ERROR, "theora_encode_YUVin failed (%s) [%d]\n", message, result); | |
213 | return -1; | |
214 | } | |
215 | ||
216 | /* Pick up returned ogg_packet */ | |
217 | result = theora_encode_packetout( &(h->t_state), 0, &o_packet ); | |
218 | switch (result) { | |
219 | case 0: | |
220 | /* No packet is ready */ | |
221 | return 0; | |
222 | case 1: | |
223 | /* Success, we have a packet */ | |
224 | break; | |
225 | default: | |
226 | av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed [%d]\n", result); | |
227 | return -1; | |
228 | } | |
229 | ||
230 | /* Copy ogg_packet content out to buffer */ | |
231 | if (buf_size < o_packet.bytes) { | |
232 | av_log(avc_context, AV_LOG_ERROR, "encoded frame too large\n"); | |
233 | return -1; | |
234 | } | |
235 | memcpy(outbuf, o_packet.packet, o_packet.bytes); | |
236 | ||
237 | return o_packet.bytes; | |
238 | } | |
239 | ||
240 | static int encode_close(AVCodecContext* avc_context) | |
241 | { | |
242 | ogg_packet o_packet; | |
243 | TheoraContext *h = avc_context->priv_data; | |
244 | int result; | |
245 | const char* message; | |
246 | ||
247 | result = theora_encode_packetout( &(h->t_state), 1, &o_packet ); | |
248 | theora_clear( &(h->t_state) ); | |
249 | switch (result) { | |
250 | case 0:/* No packet is ready */ | |
251 | case -1:/* Encoding finished */ | |
252 | return 0; | |
253 | case 1: | |
254 | /* We have a packet */ | |
255 | message = "gave us a packet"; | |
256 | break; | |
257 | default: | |
258 | message = "unknown reason"; | |
259 | break; | |
260 | } | |
261 | av_log(avc_context, AV_LOG_ERROR, "theora_encode_packetout failed (%s) [%d]\n", message, result); | |
262 | return -1; | |
263 | } | |
264 | ||
265 | static const enum PixelFormat supported_pixel_formats[] = { PIX_FMT_YUV420P, -1 }; | |
266 | ||
267 | /*! AVCodec struct exposed to libavcodec */ | |
268 | AVCodec libtheora_encoder = | |
269 | { | |
270 | .name = "libtheora", | |
271 | .type = CODEC_TYPE_VIDEO, | |
272 | .id = CODEC_ID_THEORA, | |
273 | .priv_data_size = sizeof(TheoraContext), | |
274 | .init = encode_init, | |
275 | .close = encode_close, | |
276 | .encode = encode_frame, | |
277 | .pix_fmts = supported_pixel_formats, | |
278 | }; |