libavcodec/libx265.c

   1 /*
   2  * libx265 encoder
   3  *
   4  * Copyright (c) 2013-2014 Derek Buitenhuis
   5  *
   6  * This file is part of Libav.
   7  *
   8  * Libav is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * Libav is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with Libav; if not, write to the Free Software
  20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21  */
  22
  23 #if defined(_MSC_VER)
  24 #define X265_API_IMPORTS 1
  25 #endif
  26
  27 #include <x265.h>
  28 #include <float.h>
  29
  30 #include "libavutil/internal.h"
  31 #include "libavutil/common.h"
  32 #include "libavutil/opt.h"
  33 #include "libavutil/pixdesc.h"
  34 #include "avcodec.h"
  35 #include "internal.h"
  36
  37 typedef struct libx265Context {
  38     const AVClass *class;
  39
  40     x265_encoder *encoder;
  41     x265_param   *params;
  42
  43     float crf;
  44     char *preset;
  45     char *tune;
  46     char *x265_opts;
  47 } libx265Context;
  48
  49 static int is_keyframe(NalUnitType naltype)
  50 {
  51     switch (naltype) {
  52     case NAL_UNIT_CODED_SLICE_BLA_W_LP:
  53     case NAL_UNIT_CODED_SLICE_BLA_W_RADL:
  54     case NAL_UNIT_CODED_SLICE_BLA_N_LP:
  55     case NAL_UNIT_CODED_SLICE_IDR_W_RADL:
  56     case NAL_UNIT_CODED_SLICE_IDR_N_LP:
  57     case NAL_UNIT_CODED_SLICE_CRA:
  58         return 1;
  59     default:
  60         return 0;
  61     }
  62 }
  63
  64 static av_cold int libx265_encode_close(AVCodecContext *avctx)
  65 {
  66     libx265Context *ctx = avctx->priv_data;
  67
  68     av_frame_free(&avctx->coded_frame);
  69
  70     x265_param_free(ctx->params);
  71
  72     if (ctx->encoder)
  73         x265_encoder_close(ctx->encoder);
  74
  75     return 0;
  76 }
  77
  78 static av_cold int libx265_encode_init(AVCodecContext *avctx)
  79 {
  80     libx265Context *ctx = avctx->priv_data;
  81
  82     if (avctx->strict_std_compliance > FF_COMPLIANCE_EXPERIMENTAL &&
  83         !av_pix_fmt_desc_get(avctx->pix_fmt)->log2_chroma_w) {
  84         av_log(avctx, AV_LOG_ERROR,
  85                "4:2:2 and 4:4:4 support is not fully defined for HEVC yet. "
  86                "Set -strict experimental to encode anyway.\n");
  87         return AVERROR(ENOSYS);
  88     }
  89
  90     avctx->coded_frame = av_frame_alloc();
  91     if (!avctx->coded_frame) {
  92         av_log(avctx, AV_LOG_ERROR, "Could not allocate frame.\n");
  93         return AVERROR(ENOMEM);
  94     }
  95
  96     ctx->params = x265_param_alloc();
  97     if (!ctx->params) {
  98         av_log(avctx, AV_LOG_ERROR, "Could not allocate x265 param structure.\n");
  99         return AVERROR(ENOMEM);
 100     }
 101
 102     if (x265_param_default_preset(ctx->params, ctx->preset, ctx->tune) < 0) {
 103         int i;
 104
 105         av_log(avctx, AV_LOG_ERROR, "Error setting preset/tune %s/%s.\n", ctx->preset, ctx->tune);
 106         av_log(avctx, AV_LOG_INFO, "Possible presets:");
 107         for (i = 0; x265_preset_names[i]; i++)
 108             av_log(avctx, AV_LOG_INFO, " %s", x265_preset_names[i]);
 109
 110         av_log(avctx, AV_LOG_INFO, "\n");
 111         av_log(avctx, AV_LOG_INFO, "Possible tunes:");
 112         for (i = 0; x265_tune_names[i]; i++)
 113             av_log(avctx, AV_LOG_INFO, " %s", x265_tune_names[i]);
 114
 115         av_log(avctx, AV_LOG_INFO, "\n");
 116
 117         return AVERROR(EINVAL);
 118     }
 119
 120     ctx->params->frameNumThreads = avctx->thread_count;
 121     ctx->params->fpsNum          = avctx->time_base.den;
 122     ctx->params->fpsDenom        = avctx->time_base.num * avctx->ticks_per_frame;
 123     ctx->params->sourceWidth     = avctx->width;
 124     ctx->params->sourceHeight    = avctx->height;
 125     ctx->params->bEnablePsnr     = !!(avctx->flags & CODEC_FLAG_PSNR);
 126
 127     if ((avctx->color_primaries <= AVCOL_PRI_BT2020 &&
 128          avctx->color_primaries != AVCOL_PRI_UNSPECIFIED) ||
 129         (avctx->color_trc <= AVCOL_TRC_BT2020_12 &&
 130          avctx->color_trc != AVCOL_TRC_UNSPECIFIED) ||
 131         (avctx->colorspace <= AVCOL_SPC_BT2020_CL &&
 132          avctx->colorspace != AVCOL_SPC_UNSPECIFIED)) {
 133
 134         ctx->params->vui.bEnableVideoSignalTypePresentFlag  = 1;
 135         ctx->params->vui.bEnableColorDescriptionPresentFlag = 1;
 136
 137         // x265 validates the parameters internally
 138         ctx->params->vui.colorPrimaries          = avctx->color_primaries;
 139         ctx->params->vui.transferCharacteristics = avctx->color_trc;
 140         ctx->params->vui.matrixCoeffs            = avctx->colorspace;
 141     }
 142
 143     if (avctx->sample_aspect_ratio.num > 0 && avctx->sample_aspect_ratio.den > 0) {
 144         char sar[12];
 145         int sar_num, sar_den;
 146
 147         av_reduce(&sar_num, &sar_den,
 148                   avctx->sample_aspect_ratio.num,
 149                   avctx->sample_aspect_ratio.den, 65535);
 150         snprintf(sar, sizeof(sar), "%d:%d", sar_num, sar_den);
 151         if (x265_param_parse(ctx->params, "sar", sar) == X265_PARAM_BAD_VALUE) {
 152             av_log(avctx, AV_LOG_ERROR, "Invalid SAR: %d:%d.\n", sar_num, sar_den);
 153             return AVERROR_INVALIDDATA;
 154         }
 155     }
 156
 157     switch (avctx->pix_fmt) {
 158     case AV_PIX_FMT_YUV420P:
 159     case AV_PIX_FMT_YUV420P10:
 160         ctx->params->internalCsp = X265_CSP_I420;
 161         break;
 162     case AV_PIX_FMT_YUV422P:
 163     case AV_PIX_FMT_YUV422P10:
 164         ctx->params->internalCsp = X265_CSP_I422;
 165         break;
 166     case AV_PIX_FMT_YUV444P:
 167     case AV_PIX_FMT_YUV444P10:
 168         ctx->params->internalCsp = X265_CSP_I444;
 169         break;
 170     }
 171
 172     if (ctx->crf >= 0) {
 173         char crf[6];
 174
 175         snprintf(crf, sizeof(crf), "%2.2f", ctx->crf);
 176         if (x265_param_parse(ctx->params, "crf", crf) == X265_PARAM_BAD_VALUE) {
 177             av_log(avctx, AV_LOG_ERROR, "Invalid crf: %2.2f.\n", ctx->crf);
 178             return AVERROR(EINVAL);
 179         }
 180     } else if (avctx->bit_rate > 0) {
 181         ctx->params->rc.bitrate         = avctx->bit_rate / 1000;
 182         ctx->params->rc.rateControlMode = X265_RC_ABR;
 183     }
 184
 185     if (!(avctx->flags & CODEC_FLAG_GLOBAL_HEADER))
 186         ctx->params->bRepeatHeaders = 1;
 187
 188     if (ctx->x265_opts) {
 189         AVDictionary *dict    = NULL;
 190         AVDictionaryEntry *en = NULL;
 191
 192         if (!av_dict_parse_string(&dict, ctx->x265_opts, "=", ":", 0)) {
 193             while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) {
 194                 int parse_ret = x265_param_parse(ctx->params, en->key, en->value);
 195
 196                 switch (parse_ret) {
 197                 case X265_PARAM_BAD_NAME:
 198                     av_log(avctx, AV_LOG_WARNING,
 199                           "Unknown option: %s.\n", en->key);
 200                     break;
 201                 case X265_PARAM_BAD_VALUE:
 202                     av_log(avctx, AV_LOG_WARNING,
 203                           "Invalid value for %s: %s.\n", en->key, en->value);
 204                     break;
 205                 default:
 206                     break;
 207                 }
 208             }
 209             av_dict_free(&dict);
 210         }
 211     }
 212
 213     ctx->encoder = x265_encoder_open(ctx->params);
 214     if (!ctx->encoder) {
 215         av_log(avctx, AV_LOG_ERROR, "Cannot open libx265 encoder.\n");
 216         libx265_encode_close(avctx);
 217         return AVERROR_INVALIDDATA;
 218     }
 219
 220     if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
 221         x265_nal *nal;
 222         int nnal;
 223
 224         avctx->extradata_size = x265_encoder_headers(ctx->encoder, &nal, &nnal);
 225         if (avctx->extradata_size <= 0) {
 226             av_log(avctx, AV_LOG_ERROR, "Cannot encode headers.\n");
 227             libx265_encode_close(avctx);
 228             return AVERROR_INVALIDDATA;
 229         }
 230
 231         avctx->extradata = av_malloc(avctx->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
 232         if (!avctx->extradata) {
 233             av_log(avctx, AV_LOG_ERROR,
 234                    "Cannot allocate HEVC header of size %d.\n", avctx->extradata_size);
 235             libx265_encode_close(avctx);
 236             return AVERROR(ENOMEM);
 237         }
 238
 239         memcpy(avctx->extradata, nal[0].payload, avctx->extradata_size);
 240     }
 241
 242     return 0;
 243 }
 244
 245 static int libx265_encode_frame(AVCodecContext *avctx, AVPacket *pkt,
 246                                 const AVFrame *pic, int *got_packet)
 247 {
 248     libx265Context *ctx = avctx->priv_data;
 249     x265_picture x265pic;
 250     x265_picture x265pic_out = { { 0 } };
 251     x265_nal *nal;
 252     uint8_t *dst;
 253     int payload = 0;
 254     int nnal;
 255     int ret;
 256     int i;
 257
 258     x265_picture_init(ctx->params, &x265pic);
 259
 260     if (pic) {
 261         for (i = 0; i < 3; i++) {
 262            x265pic.planes[i] = pic->data[i];
 263            x265pic.stride[i] = pic->linesize[i];
 264         }
 265
 266         x265pic.pts      = pic->pts;
 267         x265pic.bitDepth = av_pix_fmt_desc_get(avctx->pix_fmt)->comp[0].depth_minus1 + 1;
 268
 269         x265pic.sliceType = pic->pict_type == AV_PICTURE_TYPE_I ? X265_TYPE_I :
 270                             pic->pict_type == AV_PICTURE_TYPE_P ? X265_TYPE_P :
 271                             pic->pict_type == AV_PICTURE_TYPE_B ? X265_TYPE_B :
 272                             X265_TYPE_AUTO;
 273     }
 274
 275     ret = x265_encoder_encode(ctx->encoder, &nal, &nnal,
 276                               pic ? &x265pic : NULL, &x265pic_out);
 277     if (ret < 0)
 278         return AVERROR_UNKNOWN;
 279
 280     if (!nnal)
 281         return 0;
 282
 283     for (i = 0; i < nnal; i++)
 284         payload += nal[i].sizeBytes;
 285
 286     ret = ff_alloc_packet(pkt, payload);
 287     if (ret < 0) {
 288         av_log(avctx, AV_LOG_ERROR, "Error getting output packet.\n");
 289         return ret;
 290     }
 291     dst = pkt->data;
 292
 293     for (i = 0; i < nnal; i++) {
 294         memcpy(dst, nal[i].payload, nal[i].sizeBytes);
 295         dst += nal[i].sizeBytes;
 296
 297         if (is_keyframe(nal[i].type))
 298             pkt->flags |= AV_PKT_FLAG_KEY;
 299     }
 300
 301     pkt->pts = x265pic_out.pts;
 302     pkt->dts = x265pic_out.dts;
 303
 304     switch (x265pic_out.sliceType) {
 305     case X265_TYPE_IDR:
 306     case X265_TYPE_I:
 307         avctx->coded_frame->pict_type = AV_PICTURE_TYPE_I;
 308         break;
 309     case X265_TYPE_P:
 310         avctx->coded_frame->pict_type = AV_PICTURE_TYPE_P;
 311         break;
 312     case X265_TYPE_B:
 313         avctx->coded_frame->pict_type = AV_PICTURE_TYPE_B;
 314         break;
 315     }
 316
 317     *got_packet = 1;
 318     return 0;
 319 }
 320
 321 static const enum AVPixelFormat x265_csp_eight[] = {
 322     AV_PIX_FMT_YUV420P,
 323     AV_PIX_FMT_YUV422P,
 324     AV_PIX_FMT_YUV444P,
 325     AV_PIX_FMT_NONE
 326 };
 327
 328 static const enum AVPixelFormat x265_csp_twelve[] = {
 329     AV_PIX_FMT_YUV420P,
 330     AV_PIX_FMT_YUV422P,
 331     AV_PIX_FMT_YUV444P,
 332     AV_PIX_FMT_YUV420P10,
 333     AV_PIX_FMT_YUV422P10,
 334     AV_PIX_FMT_YUV444P10,
 335     AV_PIX_FMT_NONE
 336 };
 337
 338 static av_cold void libx265_encode_init_csp(AVCodec *codec)
 339 {
 340     if (x265_max_bit_depth == 8)
 341         codec->pix_fmts = x265_csp_eight;
 342     else if (x265_max_bit_depth == 12)
 343         codec->pix_fmts = x265_csp_twelve;
 344 }
 345
 346 #define OFFSET(x) offsetof(libx265Context, x)
 347 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
 348 static const AVOption options[] = {
 349     { "crf",         "set the x265 crf",                                                            OFFSET(crf),       AV_OPT_TYPE_FLOAT,  { .dbl = -1 }, -1, FLT_MAX, VE },
 350     { "preset",      "set the x265 preset",                                                         OFFSET(preset),    AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
 351     { "tune",        "set the x265 tune parameter",                                                 OFFSET(tune),      AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
 352     { "x265-params", "set the x265 configuration using a :-separated list of key=value parameters", OFFSET(x265_opts), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
 353     { NULL }
 354 };
 355
 356 static const AVClass class = {
 357     .class_name = "libx265",
 358     .item_name  = av_default_item_name,
 359     .option     = options,
 360     .version    = LIBAVUTIL_VERSION_INT,
 361 };
 362
 363 static const AVCodecDefault x265_defaults[] = {
 364     { "b", "0" },
 365     { NULL },
 366 };
 367
 368 AVCodec ff_libx265_encoder = {
 369     .name             = "libx265",
 370     .long_name        = NULL_IF_CONFIG_SMALL("libx265 H.265 / HEVC"),
 371     .type             = AVMEDIA_TYPE_VIDEO,
 372     .id               = AV_CODEC_ID_HEVC,
 373     .init             = libx265_encode_init,
 374     .init_static_data = libx265_encode_init_csp,
 375     .encode2          = libx265_encode_frame,
 376     .close            = libx265_encode_close,
 377     .priv_data_size   = sizeof(libx265Context),
 378     .priv_class       = &class,
 379     .defaults         = x265_defaults,
 380     .capabilities     = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS,
 381 };