libavcodec/x264.c

   1 /*
   2  * H.264 encoding using the x264 library
   3  * Copyright (C) 2005  Mans Rullgard <mru@inprovide.com>
   4  *
   5  * This library is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU Lesser General Public
   7  * License as published by the Free Software Foundation; either
   8  * version 2 of the License, or (at your option) any later version.
   9  *
  10  * This library is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * Lesser General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Lesser General Public
  16  * License along with this library; if not, write to the Free Software
  17  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  18  */
  19
  20 #include "avcodec.h"
  21 #include <x264.h>
  22 #include <math.h>
  23 #include <stdio.h>
  24 #include <stdlib.h>
  25 #include <string.h>
  26
  27 typedef struct X264Context {
  28     x264_param_t params;
  29     x264_t *enc;
  30     x264_picture_t pic;
  31     AVFrame out_pic;
  32 } X264Context;
  33
  34 static void
  35 X264_log(void *p, int level, const char *fmt, va_list args)
  36 {
  37     static const int level_map[] = {
  38         [X264_LOG_ERROR]   = AV_LOG_ERROR,
  39         [X264_LOG_WARNING] = AV_LOG_ERROR,
  40         [X264_LOG_INFO]    = AV_LOG_INFO,
  41         [X264_LOG_DEBUG]   = AV_LOG_DEBUG
  42     };
  43
  44     if(level < 0 || level > X264_LOG_DEBUG)
  45         return;
  46
  47     av_vlog(p, level_map[level], fmt, args);
  48 }
  49
  50
  51 static int
  52 encode_nals(uint8_t *buf, int size, x264_nal_t *nals, int nnal)
  53 {
  54     uint8_t *p = buf;
  55     int i;
  56
  57     for(i = 0; i < nnal; i++){
  58         int s = x264_nal_encode(p, &size, 1, nals + i);
  59         if(s < 0)
  60             return -1;
  61         p += s;
  62     }
  63
  64     return p - buf;
  65 }
  66
  67 extern int
  68 X264_frame(AVCodecContext *ctx, uint8_t *buf, int bufsize, void *data)
  69 {
  70     X264Context *x4 = ctx->priv_data;
  71     AVFrame *frame = data;
  72     x264_nal_t *nal;
  73     int nnal, i;
  74     x264_picture_t pic_out;
  75
  76     x4->pic.img.i_csp = X264_CSP_I420;
  77     x4->pic.img.i_plane = 3;
  78
  79     for(i = 0; i < 3; i++){
  80         x4->pic.img.plane[i] = frame->data[i];
  81         x4->pic.img.i_stride[i] = frame->linesize[i];
  82     }
  83
  84     x4->pic.i_pts = frame->pts;
  85     x4->pic.i_type = X264_TYPE_AUTO;
  86
  87     if(x264_encoder_encode(x4->enc, &nal, &nnal, &x4->pic, &pic_out))
  88         return -1;
  89
  90     bufsize = encode_nals(buf, bufsize, nal, nnal);
  91     if(bufsize < 0)
  92         return -1;
  93
  94     /* FIXME: dts */
  95     x4->out_pic.pts = pic_out.i_pts;
  96
  97     switch(pic_out.i_type){
  98     case X264_TYPE_IDR:
  99     case X264_TYPE_I:
 100         x4->out_pic.pict_type = FF_I_TYPE;
 101         break;
 102     case X264_TYPE_P:
 103         x4->out_pic.pict_type = FF_P_TYPE;
 104         break;
 105     case X264_TYPE_B:
 106     case X264_TYPE_BREF:
 107         x4->out_pic.pict_type = FF_B_TYPE;
 108         break;
 109     }
 110
 111     x4->out_pic.key_frame = pic_out.i_type == X264_TYPE_IDR;
 112     x4->out_pic.quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
 113
 114     return bufsize;
 115 }
 116
 117 static int
 118 X264_close(AVCodecContext *avctx)
 119 {
 120     X264Context *x4 = avctx->priv_data;
 121
 122     if(x4->enc)
 123         x264_encoder_close(x4->enc);
 124
 125     return 0;
 126 }
 127
 128 extern int
 129 X264_init(AVCodecContext *avctx)
 130 {
 131     X264Context *x4 = avctx->priv_data;
 132
 133     x264_param_default(&x4->params);
 134
 135     x4->params.pf_log = X264_log;
 136     x4->params.p_log_private = avctx;
 137
 138     x4->params.i_keyint_max = avctx->gop_size;
 139     x4->params.rc.i_bitrate = avctx->bit_rate / 1000;
 140     x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
 141     x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate / 1000;
 142     x4->params.rc.b_stat_write = (avctx->flags & CODEC_FLAG_PASS1);
 143     if(avctx->flags & CODEC_FLAG_PASS2) x4->params.rc.b_stat_read = 1;
 144     else{
 145         if(avctx->crf){
 146             x4->params.rc.i_rc_method = X264_RC_CRF;
 147             x4->params.rc.i_rf_constant = avctx->crf;
 148         }else if(avctx->cqp > -1){
 149             x4->params.rc.i_rc_method = X264_RC_CQP;
 150             x4->params.rc.i_qp_constant = avctx->cqp;
 151         }
 152     }
 153
 154     // if neither crf nor cqp modes are selected we have to enable the RC
 155     // we do it this way because we cannot check if the bitrate has been set
 156     if(!(avctx->crf || (avctx->cqp > -1))) x4->params.rc.i_rc_method = X264_RC_ABR;
 157
 158     x4->params.i_bframe = avctx->max_b_frames;
 159     x4->params.b_cabac = avctx->coder_type == FF_CODER_TYPE_AC;
 160     x4->params.b_bframe_adaptive = avctx->b_frame_strategy;
 161     x4->params.i_bframe_bias = avctx->bframebias;
 162     x4->params.b_bframe_pyramid = (avctx->flags2 & CODEC_FLAG2_BPYRAMID);
 163
 164     x4->params.i_keyint_min = avctx->keyint_min;
 165     if(x4->params.i_keyint_min > x4->params.i_keyint_max)
 166         x4->params.i_keyint_min = x4->params.i_keyint_max;
 167
 168     x4->params.i_scenecut_threshold = avctx->scenechange_threshold;
 169
 170     x4->params.b_deblocking_filter = (avctx->flags & CODEC_FLAG_LOOP_FILTER);
 171     x4->params.i_deblocking_filter_alphac0 = avctx->deblockalpha;
 172     x4->params.i_deblocking_filter_beta = avctx->deblockbeta;
 173
 174     x4->params.rc.i_qp_min = avctx->qmin;
 175     x4->params.rc.i_qp_max = avctx->qmax;
 176     x4->params.rc.i_qp_step = avctx->max_qdiff;
 177
 178     x4->params.rc.f_qcompress = avctx->qcompress;  /* 0.0 => cbr, 1.0 => constant qp */
 179     x4->params.rc.f_qblur = avctx->qblur;        /* temporally blur quants */
 180     x4->params.rc.f_complexity_blur = avctx->complexityblur;
 181
 182     x4->params.i_frame_reference = avctx->refs;
 183
 184     x4->params.i_width = avctx->width;
 185     x4->params.i_height = avctx->height;
 186     x4->params.vui.i_sar_width = avctx->sample_aspect_ratio.num;
 187     x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
 188     x4->params.i_fps_num = avctx->time_base.den;
 189     x4->params.i_fps_den = avctx->time_base.num;
 190
 191     x4->params.analyse.inter = 0;
 192     if(avctx->partitions){
 193         if(avctx->partitions & X264_PART_I4X4)
 194             x4->params.analyse.inter |= X264_ANALYSE_I4x4;
 195         if(avctx->partitions & X264_PART_I8X8)
 196             x4->params.analyse.inter |= X264_ANALYSE_I8x8;
 197         if(avctx->partitions & X264_PART_P8X8)
 198             x4->params.analyse.inter |= X264_ANALYSE_PSUB16x16;
 199         if(avctx->partitions & X264_PART_P4X4)
 200             x4->params.analyse.inter |= X264_ANALYSE_PSUB8x8;
 201         if(avctx->partitions & X264_PART_B8X8)
 202             x4->params.analyse.inter |= X264_ANALYSE_BSUB16x16;
 203     }
 204
 205     x4->params.analyse.i_direct_mv_pred = avctx->directpred;
 206
 207     x4->params.analyse.b_weighted_bipred = (avctx->flags2 & CODEC_FLAG2_WPRED);
 208
 209     if(avctx->me_method == ME_EPZS)
 210         x4->params.analyse.i_me_method = X264_ME_DIA;
 211     else if(avctx->me_method == ME_HEX)
 212         x4->params.analyse.i_me_method = X264_ME_HEX;
 213     else if(avctx->me_method == ME_UMH)
 214         x4->params.analyse.i_me_method = X264_ME_UMH;
 215     else if(avctx->me_method == ME_FULL)
 216         x4->params.analyse.i_me_method = X264_ME_ESA;
 217     else x4->params.analyse.i_me_method = X264_ME_HEX;
 218
 219     x4->params.analyse.i_me_range = avctx->me_range;
 220     x4->params.analyse.i_subpel_refine = avctx->me_subpel_quality;
 221
 222     x4->params.analyse.b_bframe_rdo = (avctx->flags2 & CODEC_FLAG2_BRDO);
 223     x4->params.analyse.b_mixed_references =
 224         (avctx->flags2 & CODEC_FLAG2_MIXED_REFS);
 225     x4->params.analyse.b_chroma_me = (avctx->me_cmp & FF_CMP_CHROMA);
 226     x4->params.analyse.b_transform_8x8 = (avctx->flags2 & CODEC_FLAG2_8X8DCT);
 227     x4->params.analyse.b_fast_pskip = (avctx->flags2 & CODEC_FLAG2_FASTPSKIP);
 228
 229     x4->params.analyse.i_trellis = avctx->trellis;
 230     x4->params.analyse.i_noise_reduction = avctx->noise_reduction;
 231
 232     if(avctx->level > 0) x4->params.i_level_idc = avctx->level;
 233
 234     x4->params.rc.f_rate_tolerance =
 235         (float)avctx->bit_rate_tolerance/avctx->bit_rate;
 236
 237     if((avctx->rc_buffer_size != 0) &&
 238             (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)){
 239         x4->params.rc.f_vbv_buffer_init =
 240             (float)avctx->rc_initial_buffer_occupancy/avctx->rc_buffer_size;
 241     }
 242     else x4->params.rc.f_vbv_buffer_init = 0.9;
 243
 244     x4->params.rc.f_ip_factor = 1/fabs(avctx->i_quant_factor);
 245     x4->params.rc.f_pb_factor = avctx->b_quant_factor;
 246     x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
 247     x4->params.rc.psz_rc_eq = avctx->rc_eq;
 248
 249     x4->params.analyse.b_psnr = (avctx->flags & CODEC_FLAG_PSNR);
 250     x4->params.i_log_level = X264_LOG_DEBUG;
 251
 252     x4->params.b_aud = (avctx->flags2 & CODEC_FLAG2_AUD);
 253
 254     x4->params.i_threads = avctx->thread_count;
 255
 256     if(avctx->flags & CODEC_FLAG_GLOBAL_HEADER){
 257         x4->params.b_repeat_headers = 0;
 258     }
 259
 260     x4->enc = x264_encoder_open(&x4->params);
 261     if(!x4->enc)
 262         return -1;
 263
 264     avctx->coded_frame = &x4->out_pic;
 265
 266     if(avctx->flags & CODEC_FLAG_GLOBAL_HEADER){
 267         x264_nal_t *nal;
 268         int nnal, i, s = 0;
 269
 270         x264_encoder_headers(x4->enc, &nal, &nnal);
 271
 272         /* 5 bytes NAL header + worst case escaping */
 273         for(i = 0; i < nnal; i++)
 274             s += 5 + nal[i].i_payload * 4 / 3;
 275
 276         avctx->extradata = av_malloc(s);
 277         avctx->extradata_size = encode_nals(avctx->extradata, s, nal, nnal);
 278     }
 279
 280     return 0;
 281 }
 282
 283 AVCodec x264_encoder = {
 284     .name = "h264",
 285     .type = CODEC_TYPE_VIDEO,
 286     .id = CODEC_ID_H264,
 287     .priv_data_size = sizeof(X264Context),
 288     .init = X264_init,
 289     .encode = X264_frame,
 290     .close = X264_close,
 291     .pix_fmts = (enum PixelFormat[]) { PIX_FMT_YUV420P, -1 }
 292 };