John Koleszar | 0ea50ce | 2010-05-18 11:58:33 -0400 | [diff] [blame] | 1 | /* |
John Koleszar | c2140b8 | 2010-09-09 08:16:39 -0400 | [diff] [blame] | 2 | * Copyright (c) 2010 The WebM project authors. All Rights Reserved. |
John Koleszar | 0ea50ce | 2010-05-18 11:58:33 -0400 | [diff] [blame] | 3 | * |
John Koleszar | 94c52e4 | 2010-06-18 12:39:21 -0400 | [diff] [blame] | 4 | * Use of this source code is governed by a BSD-style license |
John Koleszar | 09202d8 | 2010-06-04 16:19:40 -0400 | [diff] [blame] | 5 | * that can be found in the LICENSE file in the root of the source |
| 6 | * tree. An additional intellectual property rights grant can be found |
John Koleszar | 94c52e4 | 2010-06-18 12:39:21 -0400 | [diff] [blame] | 7 | * in the file PATENTS. All contributing project authors may |
John Koleszar | 09202d8 | 2010-06-04 16:19:40 -0400 | [diff] [blame] | 8 | * be found in the AUTHORS file in the root of the source tree. |
John Koleszar | 0ea50ce | 2010-05-18 11:58:33 -0400 | [diff] [blame] | 9 | */ |
| 10 | |
| 11 | |
| 12 | /*!\defgroup encoder Encoder Algorithm Interface |
| 13 | * \ingroup codec |
| 14 | * This abstraction allows applications using this encoder to easily support |
| 15 | * multiple video formats with minimal code duplication. This section describes |
| 16 | * the interface common to all encoders. |
| 17 | * @{ |
| 18 | */ |
| 19 | |
| 20 | /*!\file vpx_encoder.h |
| 21 | * \brief Describes the encoder algorithm interface to applications. |
| 22 | * |
| 23 | * This file describes the interface between an application and a |
| 24 | * video encoder algorithm. |
| 25 | * |
| 26 | */ |
| 27 | #ifdef __cplusplus |
| 28 | extern "C" { |
| 29 | #endif |
| 30 | |
| 31 | #ifndef VPX_ENCODER_H |
| 32 | #define VPX_ENCODER_H |
| 33 | #include "vpx_codec.h" |
| 34 | |
| 35 | |
| 36 | /*!\brief Current ABI version number |
| 37 | * |
| 38 | * \internal |
| 39 | * If this file is altered in any way that changes the ABI, this value |
| 40 | * must be bumped. Examples include, but are not limited to, changing |
| 41 | * types, removing or reassigning enums, adding/removing/rearranging |
| 42 | * fields to structures |
| 43 | */ |
| 44 | #define VPX_ENCODER_ABI_VERSION (2 + VPX_CODEC_ABI_VERSION) /**<\hideinitializer*/ |
| 45 | |
| 46 | |
| 47 | /*! \brief Encoder capabilities bitfield |
| 48 | * |
| 49 | * Each encoder advertises the capabilities it supports as part of its |
| 50 | * ::vpx_codec_iface_t interface structure. Capabilities are extra |
| 51 | * interfaces or functionality, and are not required to be supported |
| 52 | * by an encoder. |
| 53 | * |
| 54 | * The available flags are specifiedby VPX_CODEC_CAP_* defines. |
| 55 | */ |
| 56 | #define VPX_CODEC_CAP_PSNR 0x10000 /**< Can issue PSNR packets */ |
| 57 | |
| 58 | |
| 59 | /*! \brief Initialization-time Feature Enabling |
| 60 | * |
| 61 | * Certain codec features must be known at initialization time, to allow |
| 62 | * for proper memory allocation. |
| 63 | * |
| 64 | * The available flags are specified by VPX_CODEC_USE_* defines. |
| 65 | */ |
| 66 | #define VPX_CODEC_USE_PSNR 0x10000 /**< Calculate PSNR on each frame */ |
| 67 | |
| 68 | |
| 69 | /*!\brief Generic fixed size buffer structure |
| 70 | * |
| 71 | * This structure is able to hold a reference to any fixed size buffer. |
| 72 | */ |
| 73 | typedef struct vpx_fixed_buf |
| 74 | { |
| 75 | void *buf; /**< Pointer to the data */ |
| 76 | size_t sz; /**< Length of the buffer, in chars */ |
| 77 | } vpx_fixed_buf_t; /**< alias for struct vpx_fixed_buf */ |
| 78 | |
| 79 | |
| 80 | /*!\brief Time Stamp Type |
| 81 | * |
| 82 | * An integer, which when multiplied by the stream's time base, provides |
| 83 | * the absolute time of a sample. |
| 84 | */ |
| 85 | typedef int64_t vpx_codec_pts_t; |
| 86 | |
| 87 | |
| 88 | /*!\brief Compressed Frame Flags |
| 89 | * |
| 90 | * This type represents a bitfield containing information about a compressed |
| 91 | * frame that may be useful to an application. The most significant 16 bits |
| 92 | * can be used by an algorithm to provide additional detail, for example to |
| 93 | * support frame types that are codec specific (MPEG-1 D-frames for example) |
| 94 | */ |
| 95 | typedef uint32_t vpx_codec_frame_flags_t; |
| 96 | #define VPX_FRAME_IS_KEY 0x1 /**< frame is the start of a GOP */ |
| 97 | #define VPX_FRAME_IS_DROPPABLE 0x2 /**< frame can be dropped without affecting |
| 98 | the stream (no future frame depends on |
| 99 | this one) */ |
| 100 | #define VPX_FRAME_IS_INVISIBLE 0x4 /**< frame should be decoded but will not |
| 101 | be shown */ |
| 102 | |
| 103 | |
| 104 | /*!\brief Encoder output packet variants |
| 105 | * |
| 106 | * This enumeration lists the different kinds of data packets that can be |
| 107 | * returned by calls to vpx_codec_get_cx_data(). Algorithms \ref MAY |
| 108 | * extend this list to provide additional functionality. |
| 109 | */ |
| 110 | enum vpx_codec_cx_pkt_kind |
| 111 | { |
| 112 | VPX_CODEC_CX_FRAME_PKT, /**< Compressed video frame */ |
| 113 | VPX_CODEC_STATS_PKT, /**< Two-pass statistics for this frame */ |
| 114 | VPX_CODEC_PSNR_PKT, /**< PSNR statistics for this frame */ |
| 115 | VPX_CODEC_CUSTOM_PKT = 256 /**< Algorithm extensions */ |
| 116 | }; |
| 117 | |
| 118 | |
| 119 | /*!\brief Encoder output packet |
| 120 | * |
| 121 | * This structure contains the different kinds of output data the encoder |
| 122 | * may produce while compressing a frame. |
| 123 | */ |
| 124 | typedef struct vpx_codec_cx_pkt |
| 125 | { |
| 126 | enum vpx_codec_cx_pkt_kind kind; /**< packet variant */ |
| 127 | union |
| 128 | { |
| 129 | struct |
| 130 | { |
| 131 | void *buf; /**< compressed data buffer */ |
| 132 | size_t sz; /**< length of compressed data */ |
| 133 | vpx_codec_pts_t pts; /**< time stamp to show frame |
| 134 | (in timebase units) */ |
| 135 | unsigned long duration; /**< duration to show frame |
| 136 | (in timebase units) */ |
| 137 | vpx_codec_frame_flags_t flags; /**< flags for this frame */ |
| 138 | } frame; /**< data for compressed frame packet */ |
| 139 | struct vpx_fixed_buf twopass_stats; /**< data for two-pass packet */ |
| 140 | struct vpx_psnr_pkt |
| 141 | { |
| 142 | unsigned int samples[4]; /**< Number of samples, total/y/u/v */ |
| 143 | uint64_t sse[4]; /**< sum squared error, total/y/u/v */ |
| 144 | double psnr[4]; /**< PSNR, total/y/u/v */ |
| 145 | } psnr; /**< data for PSNR packet */ |
| 146 | struct vpx_fixed_buf raw; /**< data for arbitrary packets */ |
| 147 | |
| 148 | /* This packet size is fixed to allow codecs to extend this |
| 149 | * interface without having to manage storage for raw packets, |
| 150 | * ie if it's smaller than 128 bytes, you can store in the |
| 151 | * packet list directly. |
| 152 | */ |
| 153 | char pad[128 - sizeof(enum vpx_codec_cx_pkt_kind)]; /**< fixed sz */ |
| 154 | } data; /**< packet data */ |
| 155 | } vpx_codec_cx_pkt_t; /**< alias for struct vpx_codec_cx_pkt */ |
| 156 | |
| 157 | |
| 158 | /*!\brief Rational Number |
| 159 | * |
| 160 | * This structure holds a fractional value. |
| 161 | */ |
| 162 | typedef struct vpx_rational |
| 163 | { |
| 164 | int num; /**< fraction numerator */ |
| 165 | int den; /**< fraction denominator */ |
| 166 | } vpx_rational_t; /**< alias for struct vpx_rational */ |
| 167 | |
| 168 | |
| 169 | /*!\brief Multi-pass Encoding Pass */ |
| 170 | enum vpx_enc_pass |
| 171 | { |
| 172 | VPX_RC_ONE_PASS, /**< Single pass mode */ |
| 173 | VPX_RC_FIRST_PASS, /**< First pass of multi-pass mode */ |
Timothy B. Terriberry | c4d7e5e | 2010-10-27 16:04:02 -0700 | [diff] [blame] | 174 | VPX_RC_LAST_PASS /**< Final pass of multi-pass mode */ |
John Koleszar | 0ea50ce | 2010-05-18 11:58:33 -0400 | [diff] [blame] | 175 | }; |
| 176 | |
| 177 | |
| 178 | /*!\brief Rate control mode */ |
| 179 | enum vpx_rc_mode |
| 180 | { |
| 181 | VPX_VBR, /**< Variable Bit Rate (VBR) mode */ |
Paul Wilkins | e0846c9 | 2011-01-07 18:29:37 +0000 | [diff] [blame] | 182 | VPX_CBR, /**< Constant Bit Rate (CBR) mode */ |
| 183 | VPX_CQ /**< Constant Quality (CQ) mode */ |
John Koleszar | 0ea50ce | 2010-05-18 11:58:33 -0400 | [diff] [blame] | 184 | }; |
| 185 | |
| 186 | |
| 187 | /*!\brief Keyframe placement mode. |
| 188 | * |
| 189 | * This enumeration determines whether keyframes are placed automatically by |
| 190 | * the encoder or whether this behavior is disabled. Older releases of this |
| 191 | * SDK were implemented such that VPX_KF_FIXED meant keyframes were disabled. |
| 192 | * This name is confusing for this behavior, so the new symbols to be used |
| 193 | * are VPX_KF_AUTO and VPX_KF_DISABLED. |
| 194 | */ |
| 195 | enum vpx_kf_mode |
| 196 | { |
| 197 | VPX_KF_FIXED, /**< deprecated, implies VPX_KF_DISABLED */ |
| 198 | VPX_KF_AUTO, /**< Encoder determines optimal placement automatically */ |
| 199 | VPX_KF_DISABLED = 0 /**< Encoder does not place keyframes. */ |
| 200 | }; |
| 201 | |
| 202 | |
| 203 | /*!\brief Encoded Frame Flags |
| 204 | * |
| 205 | * This type indicates a bitfield to be passed to vpx_codec_encode(), defining |
| 206 | * per-frame boolean values. By convention, bits common to all codecs will be |
| 207 | * named VPX_EFLAG_*, and bits specific to an algorithm will be named |
| 208 | * /algo/_eflag_*. The lower order 16 bits are reserved for common use. |
| 209 | */ |
| 210 | typedef long vpx_enc_frame_flags_t; |
| 211 | #define VPX_EFLAG_FORCE_KF (1<<0) /**< Force this frame to be a keyframe */ |
| 212 | |
| 213 | |
| 214 | /*!\brief Encoder configuration structure |
| 215 | * |
| 216 | * This structure contains the encoder settings that have common representations |
| 217 | * across all codecs. This doesn't imply that all codecs support all features, |
| 218 | * however. |
| 219 | */ |
| 220 | typedef struct vpx_codec_enc_cfg |
| 221 | { |
| 222 | /* |
| 223 | * generic settings (g) |
| 224 | */ |
| 225 | |
| 226 | /*!\brief Algorithm specific "usage" value |
| 227 | * |
| 228 | * Algorithms may define multiple values for usage, which may convey the |
| 229 | * intent of how the application intends to use the stream. If this value |
| 230 | * is non-zero, consult the documentation for the codec to determine its |
| 231 | * meaning. |
| 232 | */ |
| 233 | unsigned int g_usage; |
| 234 | |
| 235 | |
| 236 | /*!\brief Maximum number of threads to use |
| 237 | * |
| 238 | * For multi-threaded implementations, use no more than this number of |
| 239 | * threads. The codec may use fewer threads than allowed. The value |
| 240 | * 0 is equivalent to the value 1. |
| 241 | */ |
| 242 | unsigned int g_threads; |
| 243 | |
| 244 | |
| 245 | /*!\brief Bitstream profile to use |
| 246 | * |
| 247 | * Some codecs support a notion of multiple bitstream profiles. Typically |
| 248 | * this maps to a set of features that are turned on or off. Often the |
| 249 | * profile to use is determined by the features of the intended decoder. |
| 250 | * Consult the documentation for the codec to determine the valid values |
| 251 | * for this parameter, or set to zero for a sane default. |
| 252 | */ |
| 253 | unsigned int g_profile; /**< profile of bitstream to use */ |
| 254 | |
| 255 | |
| 256 | |
| 257 | /*!\brief Width of the frame |
| 258 | * |
| 259 | * This value identifies the presentation resolution of the frame, |
| 260 | * in pixels. Note that the frames passed as input to the encoder must |
| 261 | * have this resolution. Frames will be presented by the decoder in this |
| 262 | * resolution, independent of any spatial resampling the encoder may do. |
| 263 | */ |
| 264 | unsigned int g_w; |
| 265 | |
| 266 | |
| 267 | /*!\brief Height of the frame |
| 268 | * |
| 269 | * This value identifies the presentation resolution of the frame, |
| 270 | * in pixels. Note that the frames passed as input to the encoder must |
| 271 | * have this resolution. Frames will be presented by the decoder in this |
| 272 | * resolution, independent of any spatial resampling the encoder may do. |
| 273 | */ |
| 274 | unsigned int g_h; |
| 275 | |
| 276 | |
| 277 | /*!\brief Stream timebase units |
| 278 | * |
| 279 | * Indicates the smallest interval of time, in seconds, used by the stream. |
| 280 | * For fixed frame rate material, or variable frame rate material where |
| 281 | * frames are timed at a multiple of a given clock (ex: video capture), |
| 282 | * the \ref RECOMMENDED method is to set the timebase to the reciprocal |
| 283 | * of the frame rate (ex: 1001/30000 for 29.970 Hz NTSC). This allows the |
| 284 | * pts to correspond to the frame number, which can be handy. For |
| 285 | * re-encoding video from containers with absolute time timestamps, the |
| 286 | * \ref RECOMMENDED method is to set the timebase to that of the parent |
| 287 | * container or multimedia framework (ex: 1/1000 for ms, as in FLV). |
| 288 | */ |
| 289 | struct vpx_rational g_timebase; |
| 290 | |
| 291 | |
| 292 | /*!\brief Enable error resilient mode. |
| 293 | * |
| 294 | * Error resilient mode indicates to the encoder that it should take |
| 295 | * measures appropriate for streaming over lossy or noisy links, if |
| 296 | * possible. Set to 1 to enable this feature, 0 to disable it. |
| 297 | */ |
| 298 | unsigned int g_error_resilient; |
| 299 | |
| 300 | |
| 301 | /*!\brief Multi-pass Encoding Mode |
| 302 | * |
| 303 | * This value should be set to the current phase for multi-pass encoding. |
| 304 | * For single pass, set to #VPX_RC_ONE_PASS. |
| 305 | */ |
| 306 | enum vpx_enc_pass g_pass; |
| 307 | |
| 308 | |
| 309 | /*!\brief Allow lagged encoding |
| 310 | * |
| 311 | * If set, this value allows the encoder to consume a number of input |
| 312 | * frames before producing output frames. This allows the encoder to |
| 313 | * base decisions for the current frame on future frames. This does |
| 314 | * increase the latency of the encoding pipeline, so it is not appropriate |
| 315 | * in all situations (ex: realtime encoding). |
| 316 | * |
| 317 | * Note that this is a maximum value -- the encoder may produce frames |
| 318 | * sooner than the given limit. Set this value to 0 to disable this |
| 319 | * feature. |
| 320 | */ |
| 321 | unsigned int g_lag_in_frames; |
| 322 | |
| 323 | |
| 324 | /* |
| 325 | * rate control settings (rc) |
| 326 | */ |
| 327 | |
| 328 | /*!\brief Temporal resampling configuration, if supported by the codec. |
| 329 | * |
| 330 | * Temporal resampling allows the codec to "drop" frames as a strategy to |
| 331 | * meet its target data rate. This can cause temporal discontinuities in |
| 332 | * the encoded video, which may appear as stuttering during playback. This |
| 333 | * trade-off is often acceptable, but for many applications is not. It can |
| 334 | * be disabled in these cases. |
| 335 | * |
| 336 | * Note that not all codecs support this feature. All vpx VPx codecs do. |
| 337 | * For other codecs, consult the documentation for that algorithm. |
| 338 | * |
| 339 | * This threshold is described as a percentage of the target data buffer. |
| 340 | * When the data buffer falls below this percentage of fullness, a |
| 341 | * dropped frame is indicated. Set the threshold to zero (0) to disable |
| 342 | * this feature. |
| 343 | */ |
| 344 | unsigned int rc_dropframe_thresh; |
| 345 | |
| 346 | |
| 347 | /*!\brief Enable/disable spatial resampling, if supported by the codec. |
| 348 | * |
| 349 | * Spatial resampling allows the codec to compress a lower resolution |
| 350 | * version of the frame, which is then upscaled by the encoder to the |
| 351 | * correct presentation resolution. This increases visual quality at |
| 352 | * low data rates, at the expense of CPU time on the encoder/decoder. |
| 353 | */ |
| 354 | unsigned int rc_resize_allowed; |
| 355 | |
| 356 | |
| 357 | /*!\brief Spatial resampling up watermark. |
| 358 | * |
| 359 | * This threshold is described as a percentage of the target data buffer. |
| 360 | * When the data buffer rises above this percentage of fullness, the |
| 361 | * encoder will step up to a higher resolution version of the frame. |
| 362 | */ |
| 363 | unsigned int rc_resize_up_thresh; |
| 364 | |
| 365 | |
| 366 | /*!\brief Spatial resampling down watermark. |
| 367 | * |
| 368 | * This threshold is described as a percentage of the target data buffer. |
| 369 | * When the data buffer falls below this percentage of fullness, the |
| 370 | * encoder will step down to a lower resolution version of the frame. |
| 371 | */ |
| 372 | unsigned int rc_resize_down_thresh; |
| 373 | |
| 374 | |
| 375 | /*!\brief Rate control algorithm to use. |
| 376 | * |
| 377 | * Indicates whether the end usage of this stream is to be streamed over |
| 378 | * a bandwidth constrained link, indicating that Constant Bit Rate (CBR) |
| 379 | * mode should be used, or whether it will be played back on a high |
| 380 | * bandwidth link, as from a local disk, where higher variations in |
| 381 | * bitrate are acceptable. |
| 382 | */ |
| 383 | enum vpx_rc_mode rc_end_usage; |
| 384 | |
| 385 | |
| 386 | /*!\brief Two-pass stats buffer. |
| 387 | * |
| 388 | * A buffer containing all of the stats packets produced in the first |
| 389 | * pass, concatenated. |
| 390 | */ |
| 391 | struct vpx_fixed_buf rc_twopass_stats_in; |
| 392 | |
| 393 | |
| 394 | /*!\brief Target data rate |
| 395 | * |
| 396 | * Target bandwidth to use for this stream, in kilobits per second. |
| 397 | */ |
| 398 | unsigned int rc_target_bitrate; |
| 399 | |
| 400 | |
| 401 | /* |
| 402 | * quantizer settings |
| 403 | */ |
| 404 | |
| 405 | |
| 406 | /*!\brief Minimum (Best Quality) Quantizer |
| 407 | * |
| 408 | * The quantizer is the most direct control over the quality of the |
| 409 | * encoded image. The range of valid values for the quantizer is codec |
| 410 | * specific. Consult the documentation for the codec to determine the |
| 411 | * values to use. To determine the range programmatically, call |
| 412 | * vpx_codec_enc_config_default() with a usage value of 0. |
| 413 | */ |
| 414 | unsigned int rc_min_quantizer; |
| 415 | |
| 416 | |
| 417 | /*!\brief Maximum (Worst Quality) Quantizer |
| 418 | * |
| 419 | * The quantizer is the most direct control over the quality of the |
| 420 | * encoded image. The range of valid values for the quantizer is codec |
| 421 | * specific. Consult the documentation for the codec to determine the |
| 422 | * values to use. To determine the range programmatically, call |
| 423 | * vpx_codec_enc_config_default() with a usage value of 0. |
| 424 | */ |
| 425 | unsigned int rc_max_quantizer; |
| 426 | |
| 427 | |
| 428 | /* |
| 429 | * bitrate tolerance |
| 430 | */ |
| 431 | |
| 432 | |
| 433 | /*!\brief Rate control undershoot tolerance |
| 434 | * |
| 435 | * This value, expressed as a percentage of the target bitrate, describes |
| 436 | * the target bitrate for easier frames, allowing bits to be saved for |
| 437 | * harder frames. Set to zero to use the codec default. |
| 438 | */ |
| 439 | unsigned int rc_undershoot_pct; |
| 440 | |
| 441 | |
| 442 | /*!\brief Rate control overshoot tolerance |
| 443 | * |
| 444 | * This value, expressed as a percentage of the target bitrate, describes |
| 445 | * the maximum allowed bitrate for a given frame. Set to zero to use the |
| 446 | * codec default. |
| 447 | */ |
| 448 | unsigned int rc_overshoot_pct; |
| 449 | |
| 450 | |
| 451 | /* |
| 452 | * decoder buffer model parameters |
| 453 | */ |
| 454 | |
| 455 | |
| 456 | /*!\brief Decoder Buffer Size |
| 457 | * |
| 458 | * This value indicates the amount of data that may be buffered by the |
| 459 | * decoding application. Note that this value is expressed in units of |
| 460 | * time (milliseconds). For example, a value of 5000 indicates that the |
| 461 | * client will buffer (at least) 5000ms worth of encoded data. Use the |
| 462 | * target bitrate (#rc_target_bitrate) to convert to bits/bytes, if |
| 463 | * necessary. |
| 464 | */ |
| 465 | unsigned int rc_buf_sz; |
| 466 | |
| 467 | |
| 468 | /*!\brief Decoder Buffer Initial Size |
| 469 | * |
| 470 | * This value indicates the amount of data that will be buffered by the |
| 471 | * decoding application prior to beginning playback. This value is |
| 472 | * expressed in units of time (milliseconds). Use the target bitrate |
| 473 | * (#rc_target_bitrate) to convert to bits/bytes, if necessary. |
| 474 | */ |
| 475 | unsigned int rc_buf_initial_sz; |
| 476 | |
| 477 | |
| 478 | /*!\brief Decoder Buffer Optimal Size |
| 479 | * |
| 480 | * This value indicates the amount of data that the encoder should try |
| 481 | * to maintain in the decoder's buffer. This value is expressed in units |
| 482 | * of time (milliseconds). Use the target bitrate (#rc_target_bitrate) |
| 483 | * to convert to bits/bytes, if necessary. |
| 484 | */ |
| 485 | unsigned int rc_buf_optimal_sz; |
| 486 | |
| 487 | |
| 488 | /* |
| 489 | * 2 pass rate control parameters |
| 490 | */ |
| 491 | |
| 492 | |
| 493 | /*!\brief Two-pass mode CBR/VBR bias |
| 494 | * |
| 495 | * Bias, expressed on a scale of 0 to 100, for determining target size |
| 496 | * for the current frame. The value 0 indicates the optimal CBR mode |
| 497 | * value should be used. The value 100 indicates the optimal VBR mode |
| 498 | * value should be used. Values in between indicate which way the |
| 499 | * encoder should "lean." |
| 500 | */ |
| 501 | unsigned int rc_2pass_vbr_bias_pct; /**< RC mode bias between CBR and VBR(0-100: 0->CBR, 100->VBR) */ |
| 502 | |
| 503 | |
| 504 | /*!\brief Two-pass mode per-GOP minimum bitrate |
| 505 | * |
| 506 | * This value, expressed as a percentage of the target bitrate, indicates |
| 507 | * the minimum bitrate to be used for a single GOP (aka "section") |
| 508 | */ |
| 509 | unsigned int rc_2pass_vbr_minsection_pct; |
| 510 | |
| 511 | |
| 512 | /*!\brief Two-pass mode per-GOP maximum bitrate |
| 513 | * |
| 514 | * This value, expressed as a percentage of the target bitrate, indicates |
| 515 | * the maximum bitrate to be used for a single GOP (aka "section") |
| 516 | */ |
| 517 | unsigned int rc_2pass_vbr_maxsection_pct; |
| 518 | |
| 519 | |
| 520 | /* |
| 521 | * keyframing settings (kf) |
| 522 | */ |
| 523 | |
| 524 | /*!\brief Keyframe placement mode |
| 525 | * |
| 526 | * This value indicates whether the encoder should place keyframes at a |
| 527 | * fixed interval, or determine the optimal placement automatically |
| 528 | * (as governed by the #kf_min_dist and #kf_max_dist parameters) |
| 529 | */ |
| 530 | enum vpx_kf_mode kf_mode; |
| 531 | |
| 532 | |
| 533 | /*!\brief Keyframe minimum interval |
| 534 | * |
| 535 | * This value, expressed as a number of frames, prevents the encoder from |
| 536 | * placing a keyframe nearer than kf_min_dist to the previous keyframe. At |
| 537 | * least kf_min_dist frames non-keyframes will be coded before the next |
| 538 | * keyframe. Set kf_min_dist equal to kf_max_dist for a fixed interval. |
| 539 | */ |
| 540 | unsigned int kf_min_dist; |
| 541 | |
| 542 | |
| 543 | /*!\brief Keyframe maximum interval |
| 544 | * |
| 545 | * This value, expressed as a number of frames, forces the encoder to code |
| 546 | * a keyframe if one has not been coded in the last kf_max_dist frames. |
| 547 | * A value of 0 implies all frames will be keyframes. Set kf_min_dist |
| 548 | * equal to kf_max_dist for a fixed interval. |
| 549 | */ |
| 550 | unsigned int kf_max_dist; |
| 551 | |
| 552 | } vpx_codec_enc_cfg_t; /**< alias for struct vpx_codec_enc_cfg */ |
| 553 | |
| 554 | |
| 555 | /*!\brief Initialize an encoder instance |
| 556 | * |
| 557 | * Initializes a encoder context using the given interface. Applications |
| 558 | * should call the vpx_codec_enc_init convenience macro instead of this |
| 559 | * function directly, to ensure that the ABI version number parameter |
| 560 | * is properly initialized. |
| 561 | * |
| 562 | * In XMA mode (activated by setting VPX_CODEC_USE_XMA in the flags |
| 563 | * parameter), the storage pointed to by the cfg parameter must be |
| 564 | * kept readable and stable until all memory maps have been set. |
| 565 | * |
| 566 | * \param[in] ctx Pointer to this instance's context. |
| 567 | * \param[in] iface Pointer to the algorithm interface to use. |
| 568 | * \param[in] cfg Configuration to use, if known. May be NULL. |
| 569 | * \param[in] flags Bitfield of VPX_CODEC_USE_* flags |
| 570 | * \param[in] ver ABI version number. Must be set to |
| 571 | * VPX_ENCODER_ABI_VERSION |
| 572 | * \retval #VPX_CODEC_OK |
| 573 | * The decoder algorithm initialized. |
| 574 | * \retval #VPX_CODEC_MEM_ERROR |
| 575 | * Memory allocation failed. |
| 576 | */ |
| 577 | vpx_codec_err_t vpx_codec_enc_init_ver(vpx_codec_ctx_t *ctx, |
| 578 | vpx_codec_iface_t *iface, |
| 579 | vpx_codec_enc_cfg_t *cfg, |
| 580 | vpx_codec_flags_t flags, |
| 581 | int ver); |
| 582 | |
| 583 | |
| 584 | /*!\brief Convenience macro for vpx_codec_enc_init_ver() |
| 585 | * |
| 586 | * Ensures the ABI version parameter is properly set. |
| 587 | */ |
| 588 | #define vpx_codec_enc_init(ctx, iface, cfg, flags) \ |
| 589 | vpx_codec_enc_init_ver(ctx, iface, cfg, flags, VPX_ENCODER_ABI_VERSION) |
| 590 | |
| 591 | |
| 592 | /*!\brief Get a default configuration |
| 593 | * |
| 594 | * Initializes a encoder configuration structure with default values. Supports |
| 595 | * the notion of "usages" so that an algorithm may offer different default |
| 596 | * settings depending on the user's intended goal. This function \ref SHOULD |
| 597 | * be called by all applications to initialize the configuration structure |
| 598 | * before specializing the configuration with application specific values. |
| 599 | * |
| 600 | * \param[in] iface Pointer to the algorithm interface to use. |
| 601 | * \param[out] cfg Configuration buffer to populate |
| 602 | * \param[in] usage End usage. Set to 0 or use codec specific values. |
| 603 | * |
| 604 | * \retval #VPX_CODEC_OK |
| 605 | * The configuration was populated. |
| 606 | * \retval #VPX_CODEC_INCAPABLE |
| 607 | * Interface is not an encoder interface. |
| 608 | * \retval #VPX_CODEC_INVALID_PARAM |
| 609 | * A parameter was NULL, or the usage value was not recognized. |
| 610 | */ |
| 611 | vpx_codec_err_t vpx_codec_enc_config_default(vpx_codec_iface_t *iface, |
| 612 | vpx_codec_enc_cfg_t *cfg, |
| 613 | unsigned int usage); |
| 614 | |
| 615 | |
| 616 | /*!\brief Set or change configuration |
| 617 | * |
| 618 | * Reconfigures an encoder instance according to the given configuration. |
| 619 | * |
| 620 | * \param[in] ctx Pointer to this instance's context |
| 621 | * \param[in] cfg Configuration buffer to use |
| 622 | * |
| 623 | * \retval #VPX_CODEC_OK |
| 624 | * The configuration was populated. |
| 625 | * \retval #VPX_CODEC_INCAPABLE |
| 626 | * Interface is not an encoder interface. |
| 627 | * \retval #VPX_CODEC_INVALID_PARAM |
| 628 | * A parameter was NULL, or the usage value was not recognized. |
| 629 | */ |
| 630 | vpx_codec_err_t vpx_codec_enc_config_set(vpx_codec_ctx_t *ctx, |
| 631 | const vpx_codec_enc_cfg_t *cfg); |
| 632 | |
| 633 | |
| 634 | /*!\brief Get global stream headers |
| 635 | * |
| 636 | * Retrieves a stream level global header packet, if supported by the codec. |
| 637 | * |
| 638 | * \param[in] ctx Pointer to this instance's context |
| 639 | * |
| 640 | * \retval NULL |
| 641 | * Encoder does not support global header |
| 642 | * \retval Non-NULL |
| 643 | * Pointer to buffer containing global header packet |
| 644 | */ |
| 645 | vpx_fixed_buf_t *vpx_codec_get_global_headers(vpx_codec_ctx_t *ctx); |
| 646 | |
| 647 | |
| 648 | #define VPX_DL_REALTIME (1) /**< deadline parameter analogous to |
| 649 | * VPx REALTIME mode. */ |
| 650 | #define VPX_DL_GOOD_QUALITY (1000000) /**< deadline parameter analogous to |
| 651 | * VPx GOOD QUALITY mode. */ |
| 652 | #define VPX_DL_BEST_QUALITY (0) /**< deadline parameter analogous to |
| 653 | * VPx BEST QUALITY mode. */ |
| 654 | /*!\brief Encode a frame |
| 655 | * |
| 656 | * Encodes a video frame at the given "presentation time." The presentation |
| 657 | * time stamp (PTS) \ref MUST be strictly increasing. |
| 658 | * |
| 659 | * The encoder supports the notion of a soft real-time deadline. Given a |
| 660 | * non-zero value to the deadline parameter, the encoder will make a "best |
| 661 | * effort" guarantee to return before the given time slice expires. It is |
| 662 | * implicit that limiting the available time to encode will degrade the |
| 663 | * output quality. The encoder can be given an unlimited time to produce the |
| 664 | * best possible frame by specifying a deadline of '0'. This deadline |
| 665 | * supercedes the VPx notion of "best quality, good quality, realtime". |
| 666 | * Applications that wish to map these former settings to the new deadline |
| 667 | * based system can use the symbols #VPX_DL_REALTIME, #VPX_DL_GOOD_QUALITY, |
| 668 | * and #VPX_DL_BEST_QUALITY. |
| 669 | * |
| 670 | * When the last frame has been passed to the encoder, this function should |
| 671 | * continue to be called, with the img parameter set to NULL. This will |
| 672 | * signal the end-of-stream condition to the encoder and allow it to encode |
| 673 | * any held buffers. Encoding is complete when vpx_codec_encode() is called |
| 674 | * and vpx_codec_get_cx_data() returns no data. |
| 675 | * |
| 676 | * \param[in] ctx Pointer to this instance's context |
| 677 | * \param[in] img Image data to encode, NULL to flush. |
| 678 | * \param[in] pts Presentation time stamp, in timebase units. |
| 679 | * \param[in] duration Duration to show frame, in timebase units. |
| 680 | * \param[in] flags Flags to use for encoding this frame. |
| 681 | * \param[in] deadline Time to spend encoding, in microseconds. (0=infinite) |
| 682 | * |
| 683 | * \retval #VPX_CODEC_OK |
| 684 | * The configuration was populated. |
| 685 | * \retval #VPX_CODEC_INCAPABLE |
| 686 | * Interface is not an encoder interface. |
| 687 | * \retval #VPX_CODEC_INVALID_PARAM |
| 688 | * A parameter was NULL, the image format is unsupported, etc. |
| 689 | */ |
| 690 | vpx_codec_err_t vpx_codec_encode(vpx_codec_ctx_t *ctx, |
| 691 | const vpx_image_t *img, |
| 692 | vpx_codec_pts_t pts, |
| 693 | unsigned long duration, |
| 694 | vpx_enc_frame_flags_t flags, |
| 695 | unsigned long deadline); |
| 696 | |
| 697 | |
| 698 | /*!\brief Set compressed data output buffer |
| 699 | * |
| 700 | * Sets the buffer that the codec should output the compressed data |
| 701 | * into. This call effectively sets the buffer pointer returned in the |
| 702 | * next VPX_CODEC_CX_FRAME_PKT packet. Subsequent packets will be |
| 703 | * appended into this buffer. The buffer is preserved across frames, |
| 704 | * so applications must periodically call this function after flushing |
| 705 | * the accumulated compressed data to disk or to the network to reset |
| 706 | * the pointer to the buffer's head. |
| 707 | * |
| 708 | * `pad_before` bytes will be skipped before writing the compressed |
| 709 | * data, and `pad_after` bytes will be appended to the packet. The size |
| 710 | * of the packet will be the sum of the size of the actual compressed |
| 711 | * data, pad_before, and pad_after. The padding bytes will be preserved |
| 712 | * (not overwritten). |
| 713 | * |
| 714 | * Note that calling this function does not guarantee that the returned |
| 715 | * compressed data will be placed into the specified buffer. In the |
| 716 | * event that the encoded data will not fit into the buffer provided, |
| 717 | * the returned packet \ref MAY point to an internal buffer, as it would |
| 718 | * if this call were never used. In this event, the output packet will |
| 719 | * NOT have any padding, and the application must free space and copy it |
| 720 | * to the proper place. This is of particular note in configurations |
| 721 | * that may output multiple packets for a single encoded frame (e.g., lagged |
| 722 | * encoding) or if the application does not reset the buffer periodically. |
| 723 | * |
| 724 | * Applications may restore the default behavior of the codec providing |
| 725 | * the compressed data buffer by calling this function with a NULL |
| 726 | * buffer. |
| 727 | * |
| 728 | * Applications \ref MUSTNOT call this function during iteration of |
| 729 | * vpx_codec_get_cx_data(). |
| 730 | * |
| 731 | * \param[in] ctx Pointer to this instance's context |
| 732 | * \param[in] buf Buffer to store compressed data into |
| 733 | * \param[in] pad_before Bytes to skip before writing compressed data |
| 734 | * \param[in] pad_after Bytes to skip after writing compressed data |
| 735 | * |
| 736 | * \retval #VPX_CODEC_OK |
| 737 | * The buffer was set successfully. |
| 738 | * \retval #VPX_CODEC_INVALID_PARAM |
| 739 | * A parameter was NULL, the image format is unsupported, etc. |
| 740 | */ |
| 741 | vpx_codec_err_t vpx_codec_set_cx_data_buf(vpx_codec_ctx_t *ctx, |
| 742 | const vpx_fixed_buf_t *buf, |
| 743 | unsigned int pad_before, |
| 744 | unsigned int pad_after); |
| 745 | |
| 746 | |
| 747 | /*!\brief Encoded data iterator |
| 748 | * |
| 749 | * Iterates over a list of data packets to be passed from the encoder to the |
| 750 | * application. The different kinds of packets available are enumerated in |
| 751 | * #vpx_codec_cx_pkt_kind. |
| 752 | * |
| 753 | * #VPX_CODEC_CX_FRAME_PKT packets should be passed to the application's |
| 754 | * muxer. Multiple compressed frames may be in the list. |
| 755 | * #VPX_CODEC_STATS_PKT packets should be appended to a global buffer. |
| 756 | * |
| 757 | * The application \ref MUST silently ignore any packet kinds that it does |
| 758 | * not recognize or support. |
| 759 | * |
| 760 | * The data buffers returned from this function are only guaranteed to be |
| 761 | * valid until the application makes another call to any vpx_codec_* function. |
| 762 | * |
| 763 | * \param[in] ctx Pointer to this instance's context |
| 764 | * \param[in,out] iter Iterator storage, initialized to NULL |
| 765 | * |
| 766 | * \return Returns a pointer to an output data packet (compressed frame data, |
| 767 | * two-pass statistics, etc.) or NULL to signal end-of-list. |
| 768 | * |
| 769 | */ |
| 770 | const vpx_codec_cx_pkt_t *vpx_codec_get_cx_data(vpx_codec_ctx_t *ctx, |
| 771 | vpx_codec_iter_t *iter); |
| 772 | |
| 773 | |
| 774 | /*!\brief Get Preview Frame |
| 775 | * |
| 776 | * Returns an image that can be used as a preview. Shows the image as it would |
| 777 | * exist at the decompressor. The application \ref MUST NOT write into this |
| 778 | * image buffer. |
| 779 | * |
| 780 | * \param[in] ctx Pointer to this instance's context |
| 781 | * |
| 782 | * \return Returns a pointer to a preview image, or NULL if no image is |
| 783 | * available. |
| 784 | * |
| 785 | */ |
| 786 | const vpx_image_t *vpx_codec_get_preview_frame(vpx_codec_ctx_t *ctx); |
| 787 | |
| 788 | |
| 789 | /*!@} - end defgroup encoder*/ |
| 790 | |
| 791 | #endif |
| 792 | #ifdef __cplusplus |
| 793 | } |
| 794 | #endif |