libavcodec/libx264.c
6662ec29
 /*
  * H.264 encoding using the x264 library
f2250162
  * Copyright (C) 2005  Mans Rullgard <mans@mansr.com>
6662ec29
  *
b78e7197
  * This file is part of FFmpeg.
  *
  * FFmpeg is free software; you can redistribute it and/or
6662ec29
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
b78e7197
  * version 2.1 of the License, or (at your option) any later version.
6662ec29
  *
b78e7197
  * FFmpeg is distributed in the hope that it will be useful,
6662ec29
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  * Lesser General Public License for more details.
  *
  * You should have received a copy of the GNU Lesser General Public
b78e7197
  * License along with FFmpeg; if not, write to the Free Software
5509bffa
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
6662ec29
  */
 
0140d3f0
 #include "libavutil/opt.h"
6662ec29
 #include "avcodec.h"
 #include <x264.h>
337afd14
 #include <math.h>
58f7833e
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
6662ec29
 
 typedef struct X264Context {
0140d3f0
     AVClass        *class;
8a8720c1
     x264_param_t    params;
     x264_t         *enc;
     x264_picture_t  pic;
     uint8_t        *sei;
     int             sei_size;
     AVFrame         out_pic;
580fa76c
     char *preset;
     char *tune;
     char *profile;
     char *level;
0140d3f0
     int fastfirstpass;
580fa76c
     char *stats;
     char *weightp;
d1f9621d
     char *x264opts;
6662ec29
 } X264Context;
 
8a8720c1
 static void X264_log(void *p, int level, const char *fmt, va_list args)
6662ec29
 {
     static const int level_map[] = {
bb270c08
         [X264_LOG_ERROR]   = AV_LOG_ERROR,
3709f0d7
         [X264_LOG_WARNING] = AV_LOG_WARNING,
bb270c08
         [X264_LOG_INFO]    = AV_LOG_INFO,
         [X264_LOG_DEBUG]   = AV_LOG_DEBUG
6662ec29
     };
 
8a8720c1
     if (level < 0 || level > X264_LOG_DEBUG)
bb270c08
         return;
6662ec29
 
     av_vlog(p, level_map[level], fmt, args);
 }
 
 
8a8720c1
 static int encode_nals(AVCodecContext *ctx, uint8_t *buf, int size,
                        x264_nal_t *nals, int nnal, int skip_sei)
6662ec29
 {
fe022ce2
     X264Context *x4 = ctx->priv_data;
f40487c2
     uint8_t *p = buf;
2d3beede
     int i;
fe022ce2
 
     /* Write the SEI as part of the first frame. */
8a8720c1
     if (x4->sei_size > 0 && nnal > 0) {
7e362df3
         if (x4->sei_size > size) {
             av_log(ctx, AV_LOG_ERROR, "Error: nal buffer is too small\n");
             return -1;
         }
fe022ce2
         memcpy(p, x4->sei, x4->sei_size);
         p += x4->sei_size;
         x4->sei_size = 0;
7e362df3
         // why is x4->sei not freed?
fe022ce2
     }
6662ec29
 
8a8720c1
     for (i = 0; i < nnal; i++){
fe022ce2
         /* Don't put the SEI in extradata. */
8a8720c1
         if (skip_sei && nals[i].i_type == NAL_SEI) {
2d3beede
             x4->sei_size = nals[i].i_payload;
8a8720c1
             x4->sei      = av_malloc(x4->sei_size);
2d3beede
             memcpy(x4->sei, nals[i].p_payload, nals[i].i_payload);
fe022ce2
             continue;
         }
7e362df3
         if (nals[i].i_payload > (size - (p - buf))) {
             // return only complete nals which fit in buf
             av_log(ctx, AV_LOG_ERROR, "Error: nal buffer is too small\n");
             break;
         }
2d3beede
         memcpy(p, nals[i].p_payload, nals[i].i_payload);
         p += nals[i].i_payload;
6662ec29
     }
 
     return p - buf;
 }
 
8a8720c1
 static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
e49abd1d
                       int orig_bufsize, void *data)
6662ec29
 {
     X264Context *x4 = ctx->priv_data;
     AVFrame *frame = data;
     x264_nal_t *nal;
     int nnal, i;
     x264_picture_t pic_out;
e49abd1d
     int bufsize;
6662ec29
 
652d9d24
     x264_picture_init( &x4->pic );
8a8720c1
     x4->pic.img.i_csp   = X264_CSP_I420;
6662ec29
     x4->pic.img.i_plane = 3;
 
8a8720c1
     if (frame) {
         for (i = 0; i < 3; i++) {
             x4->pic.img.plane[i]    = frame->data[i];
67bcc870
             x4->pic.img.i_stride[i] = frame->linesize[i];
         }
6662ec29
 
8a8720c1
         x4->pic.i_pts  = frame->pts;
3ab354d7
         x4->pic.i_type =
ce5e49b0
             frame->pict_type == AV_PICTURE_TYPE_I ? X264_TYPE_KEYFRAME :
             frame->pict_type == AV_PICTURE_TYPE_P ? X264_TYPE_P :
             frame->pict_type == AV_PICTURE_TYPE_B ? X264_TYPE_B :
3ab354d7
                                             X264_TYPE_AUTO;
5f58927b
         if (x4->params.b_tff != frame->top_field_first) {
             x4->params.b_tff = frame->top_field_first;
             x264_encoder_reconfig(x4->enc, &x4->params);
         }
5655469e
         if (x4->params.vui.i_sar_height != ctx->sample_aspect_ratio.den
          || x4->params.vui.i_sar_width != ctx->sample_aspect_ratio.num) {
             x4->params.vui.i_sar_height = ctx->sample_aspect_ratio.den;
             x4->params.vui.i_sar_width = ctx->sample_aspect_ratio.num;
             x264_encoder_reconfig(x4->enc, &x4->params);
         }
84278104
     }
6662ec29
 
d545fa56
     do {
e49abd1d
         bufsize = orig_bufsize;
8a8720c1
     if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
bb270c08
         return -1;
6662ec29
 
fe022ce2
     bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
8a8720c1
     if (bufsize < 0)
bb270c08
         return -1;
d545fa56
     } while (!bufsize && !frame && x264_encoder_delayed_frames(x4->enc));
6662ec29
 
76d81909
     /* FIXME: libx264 now provides DTS, but AVFrame doesn't have a field for it. */
6662ec29
     x4->out_pic.pts = pic_out.i_pts;
 
8a8720c1
     switch (pic_out.i_type) {
6662ec29
     case X264_TYPE_IDR:
     case X264_TYPE_I:
ce5e49b0
         x4->out_pic.pict_type = AV_PICTURE_TYPE_I;
6662ec29
         break;
     case X264_TYPE_P:
ce5e49b0
         x4->out_pic.pict_type = AV_PICTURE_TYPE_P;
6662ec29
         break;
     case X264_TYPE_B:
     case X264_TYPE_BREF:
ce5e49b0
         x4->out_pic.pict_type = AV_PICTURE_TYPE_B;
6662ec29
         break;
     }
 
76d81909
     x4->out_pic.key_frame = pic_out.b_keyframe;
46f83e5b
     if (bufsize)
         x4->out_pic.quality = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
6662ec29
 
     return bufsize;
 }
 
8a8720c1
 static av_cold int X264_close(AVCodecContext *avctx)
6662ec29
 {
     X264Context *x4 = avctx->priv_data;
 
5c4e9728
     av_freep(&avctx->extradata);
fe022ce2
     av_free(x4->sei);
5c4e9728
 
8a8720c1
     if (x4->enc)
bb270c08
         x264_encoder_close(x4->enc);
6662ec29
 
     return 0;
 }
 
20a16048
 /**
  * Detect default settings and use default profile to avoid libx264 failure.
  */
 static void check_default_settings(AVCodecContext *avctx)
 {
     X264Context *x4 = avctx->priv_data;
 
     int score = 0;
     score += x4->params.analyse.i_me_range == 0;
     score += x4->params.rc.i_qp_step == 3;
     score += x4->params.i_keyint_max == 12;
     score += x4->params.rc.i_qp_min == 2;
     score += x4->params.rc.i_qp_max == 31;
     score += x4->params.rc.f_qcompress == 0.5;
     score += fabs(x4->params.rc.f_ip_factor - 1.25) < 0.01;
     score += fabs(x4->params.rc.f_pb_factor - 1.25) < 0.01;
     score += x4->params.analyse.inter == 0 && x4->params.analyse.i_subpel_refine == 8;
     if (score >= 5) {
         av_log(avctx, AV_LOG_ERROR, "Default settings detected, using medium profile\n");
580fa76c
         x4->preset = av_strdup("medium");
28768579
         if (avctx->bit_rate == 200*1000)
20a16048
             avctx->crf = 23;
     }
 }
 
f92b0084
 #define OPT_STR(opt, param)                                             \
     do {                                                                \
         if (param && x264_param_parse(&x4->params, opt, param) < 0) {   \
             av_log(avctx, AV_LOG_ERROR,                                 \
                    "bad value for '%s': '%s'\n", opt, param);           \
             return -1;                                                  \
         }                                                               \
     } while (0);                                                        \
 
8a8720c1
 static av_cold int X264_init(AVCodecContext *avctx)
6662ec29
 {
     X264Context *x4 = avctx->priv_data;
 
fe022ce2
     x4->sei_size = 0;
6662ec29
     x264_param_default(&x4->params);
 
8a8720c1
     x4->params.i_keyint_max         = avctx->gop_size;
58f7833e
 
8a8720c1
     x4->params.i_bframe          = avctx->max_b_frames;
     x4->params.b_cabac           = avctx->coder_type == FF_CODER_TYPE_AC;
0bc4c436
     x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
8a8720c1
     x4->params.i_bframe_bias     = avctx->bframebias;
19538981
     x4->params.i_bframe_pyramid  = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? X264_B_PYRAMID_NORMAL : X264_B_PYRAMID_NONE;
58f7833e
 
     x4->params.i_keyint_min = avctx->keyint_min;
8a8720c1
     if (x4->params.i_keyint_min > x4->params.i_keyint_max)
58f7833e
         x4->params.i_keyint_min = x4->params.i_keyint_max;
 
8a8720c1
     x4->params.i_scenecut_threshold        = avctx->scenechange_threshold;
58f7833e
 
f543f636
     x4->params.b_deblocking_filter         = avctx->flags & CODEC_FLAG_LOOP_FILTER;
58f7833e
     x4->params.i_deblocking_filter_alphac0 = avctx->deblockalpha;
8a8720c1
     x4->params.i_deblocking_filter_beta    = avctx->deblockbeta;
f9791751
 
8a8720c1
     x4->params.rc.i_qp_min                 = avctx->qmin;
     x4->params.rc.i_qp_max                 = avctx->qmax;
     x4->params.rc.i_qp_step                = avctx->max_qdiff;
6662ec29
 
8a8720c1
     x4->params.rc.f_qcompress       = avctx->qcompress; /* 0.0 => cbr, 1.0 => constant qp */
     x4->params.rc.f_qblur           = avctx->qblur;     /* temporally blur quants */
58f7833e
     x4->params.rc.f_complexity_blur = avctx->complexityblur;
18ae520b
 
8a8720c1
     x4->params.i_frame_reference    = avctx->refs;
68c6d60f
 
8a8720c1
     x4->params.analyse.inter    = 0;
     if (avctx->partitions) {
         if (avctx->partitions & X264_PART_I4X4)
58f7833e
             x4->params.analyse.inter |= X264_ANALYSE_I4x4;
8a8720c1
         if (avctx->partitions & X264_PART_I8X8)
58f7833e
             x4->params.analyse.inter |= X264_ANALYSE_I8x8;
8a8720c1
         if (avctx->partitions & X264_PART_P8X8)
58f7833e
             x4->params.analyse.inter |= X264_ANALYSE_PSUB16x16;
8a8720c1
         if (avctx->partitions & X264_PART_P4X4)
58f7833e
             x4->params.analyse.inter |= X264_ANALYSE_PSUB8x8;
8a8720c1
         if (avctx->partitions & X264_PART_B8X8)
58f7833e
             x4->params.analyse.inter |= X264_ANALYSE_BSUB16x16;
     }
 
8a8720c1
     x4->params.analyse.i_direct_mv_pred  = avctx->directpred;
58f7833e
 
f543f636
     x4->params.analyse.b_weighted_bipred = avctx->flags2 & CODEC_FLAG2_WPRED;
58f7833e
 
8a8720c1
     if (avctx->me_method == ME_EPZS)
58f7833e
         x4->params.analyse.i_me_method = X264_ME_DIA;
8a8720c1
     else if (avctx->me_method == ME_HEX)
58f7833e
         x4->params.analyse.i_me_method = X264_ME_HEX;
8a8720c1
     else if (avctx->me_method == ME_UMH)
58f7833e
         x4->params.analyse.i_me_method = X264_ME_UMH;
8a8720c1
     else if (avctx->me_method == ME_FULL)
58f7833e
         x4->params.analyse.i_me_method = X264_ME_ESA;
8a8720c1
     else if (avctx->me_method == ME_TESA)
89223ef6
         x4->params.analyse.i_me_method = X264_ME_TESA;
58f7833e
     else x4->params.analyse.i_me_method = X264_ME_HEX;
 
dbb4f051
     x4->params.rc.i_aq_mode               = avctx->aq_mode;
     x4->params.rc.f_aq_strength           = avctx->aq_strength;
     x4->params.rc.i_lookahead             = avctx->rc_lookahead;
f3b3b489
 
dbb4f051
     x4->params.analyse.b_psy              = avctx->flags2 & CODEC_FLAG2_PSY;
     x4->params.analyse.f_psy_rd           = avctx->psy_rd;
     x4->params.analyse.f_psy_trellis      = avctx->psy_trellis;
f3b3b489
 
8a8720c1
     x4->params.analyse.i_me_range         = avctx->me_range;
     x4->params.analyse.i_subpel_refine    = avctx->me_subpel_quality;
58f7833e
 
f543f636
     x4->params.analyse.b_mixed_references = avctx->flags2 & CODEC_FLAG2_MIXED_REFS;
     x4->params.analyse.b_chroma_me        = avctx->me_cmp & FF_CMP_CHROMA;
     x4->params.analyse.b_transform_8x8    = avctx->flags2 & CODEC_FLAG2_8X8DCT;
     x4->params.analyse.b_fast_pskip       = avctx->flags2 & CODEC_FLAG2_FASTPSKIP;
58f7833e
 
8a8720c1
     x4->params.analyse.i_trellis          = avctx->trellis;
     x4->params.analyse.i_noise_reduction  = avctx->noise_reduction;
58f7833e
 
0140d3f0
     x4->params.rc.b_mb_tree               = !!(avctx->flags2 & CODEC_FLAG2_MBTREE);
     x4->params.rc.f_ip_factor             = 1 / fabs(avctx->i_quant_factor);
     x4->params.rc.f_pb_factor             = avctx->b_quant_factor;
     x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
 
20a16048
     if (!x4->preset)
         check_default_settings(avctx);
 
0140d3f0
     if (x4->preset || x4->tune) {
         if (x264_param_default_preset(&x4->params, x4->preset, x4->tune) < 0)
             return -1;
     }
 
     x4->params.pf_log               = X264_log;
     x4->params.p_log_private        = avctx;
     x4->params.i_log_level          = X264_LOG_DEBUG;
 
97dc86b7
     OPT_STR("weightp", x4->weightp);
 
0140d3f0
     x4->params.b_intra_refresh      = avctx->flags2 & CODEC_FLAG2_INTRA_REFRESH;
     x4->params.rc.i_bitrate         = avctx->bit_rate       / 1000;
     x4->params.rc.i_vbv_buffer_size = avctx->rc_buffer_size / 1000;
     x4->params.rc.i_vbv_max_bitrate = avctx->rc_max_rate    / 1000;
     x4->params.rc.b_stat_write      = avctx->flags & CODEC_FLAG_PASS1;
     if (avctx->flags & CODEC_FLAG_PASS2) {
         x4->params.rc.b_stat_read = 1;
     } else {
         if (avctx->crf) {
             x4->params.rc.i_rc_method   = X264_RC_CRF;
             x4->params.rc.f_rf_constant = avctx->crf;
             x4->params.rc.f_rf_constant_max = avctx->crf_max;
         } else if (avctx->cqp > -1) {
             x4->params.rc.i_rc_method   = X264_RC_CQP;
             x4->params.rc.i_qp_constant = avctx->cqp;
         }
     }
 
2c18893a
     OPT_STR("stats", x4->stats);
 
0140d3f0
     // if neither crf nor cqp modes are selected we have to enable the RC
     // we do it this way because we cannot check if the bitrate has been set
     if (!(avctx->crf || (avctx->cqp > -1)))
         x4->params.rc.i_rc_method = X264_RC_ABR;
 
2c855cea
     if (avctx->rc_buffer_size && avctx->rc_initial_buffer_occupancy &&
8a8720c1
         (avctx->rc_initial_buffer_occupancy <= avctx->rc_buffer_size)) {
58f7833e
         x4->params.rc.f_vbv_buffer_init =
8a8720c1
             (float)avctx->rc_initial_buffer_occupancy / avctx->rc_buffer_size;
2c855cea
     }
58f7833e
 
f92b0084
     OPT_STR("level", x4->level);
 
d1f9621d
     if(x4->x264opts){
         const char *p= x4->x264opts;
         while(p){
             char param[256]={0}, val[256]={0};
             sscanf(p, "%255[^:=]=%255[^:]", param, val);
             OPT_STR(param, val);
             p= strchr(p, ':');
             p+=!!p;
         }
     }
 
37c0a443
     if (x4->fastfirstpass)
         x264_param_apply_fastfirstpass(&x4->params);
 
     if (x4->profile)
         if (x264_param_apply_profile(&x4->params, x4->profile) < 0)
             return -1;
 
0140d3f0
     x4->params.i_width          = avctx->width;
     x4->params.i_height         = avctx->height;
     x4->params.vui.i_sar_width  = avctx->sample_aspect_ratio.num;
     x4->params.vui.i_sar_height = avctx->sample_aspect_ratio.den;
     x4->params.i_fps_num = x4->params.i_timebase_den = avctx->time_base.den;
     x4->params.i_fps_den = x4->params.i_timebase_num = avctx->time_base.num;
58f7833e
 
f543f636
     x4->params.analyse.b_psnr = avctx->flags & CODEC_FLAG_PSNR;
d749da7e
     x4->params.analyse.b_ssim = avctx->flags2 & CODEC_FLAG2_SSIM;
58f7833e
 
f543f636
     x4->params.b_aud          = avctx->flags2 & CODEC_FLAG2_AUD;
58f7833e
 
8a8720c1
     x4->params.i_threads      = avctx->thread_count;
568d4b81
 
f543f636
     x4->params.b_interlaced   = avctx->flags & CODEC_FLAG_INTERLACED_DCT;
d9e5c9b7
 
d39b33c6
 //    x4->params.b_open_gop     = !(avctx->flags & CODEC_FLAG_CLOSED_GOP);
e25c6710
 
2aa72ecc
     x4->params.i_slice_count  = avctx->slices;
 
3cf8db94
     x4->params.vui.b_fullrange = avctx->pix_fmt == PIX_FMT_YUVJ420P;
 
8a8720c1
     if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER)
806011f2
         x4->params.b_repeat_headers = 0;
 
0140d3f0
     // update AVCodecContext with x264 parameters
c225615b
     avctx->has_b_frames = x4->params.i_bframe ?
         x4->params.i_bframe_pyramid ? 2 : 1 : 0;
0140d3f0
     avctx->bit_rate = x4->params.rc.i_bitrate*1000;
     avctx->crf = x4->params.rc.f_rf_constant;
 
6662ec29
     x4->enc = x264_encoder_open(&x4->params);
8a8720c1
     if (!x4->enc)
6662ec29
         return -1;
 
     avctx->coded_frame = &x4->out_pic;
 
8a8720c1
     if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
4f59b684
         x264_nal_t *nal;
f3b3b489
         int nnal, s, i;
806011f2
 
2d3beede
         s = x264_encoder_headers(x4->enc, &nal, &nnal);
806011f2
 
f3b3b489
         for (i = 0; i < nnal; i++)
             if (nal[i].i_type == NAL_SEI)
                 av_log(avctx, AV_LOG_INFO, "%s\n", nal[i].p_payload+25);
 
8a8720c1
         avctx->extradata      = av_malloc(s);
fe022ce2
         avctx->extradata_size = encode_nals(avctx, avctx->extradata, s, nal, nnal, 1);
806011f2
     }
 
6662ec29
     return 0;
 }
 
0140d3f0
 #define OFFSET(x) offsetof(X264Context,x)
 #define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
 
 static const AVOption options[] = {
458f20bc
     {"preset", "Set the encoding preset", OFFSET(preset), FF_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
     {"tune", "Tune the encoding params", OFFSET(tune), FF_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
     {"fastfirstpass", "Use fast settings when encoding first pass", OFFSET(fastfirstpass), FF_OPT_TYPE_INT, {.dbl=1}, 0, 1, VE},
     {"profile", "Set profile restrictions", OFFSET(profile), FF_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
     {"level", "Specify level (as defined by Annex A)", OFFSET(level), FF_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
     {"passlogfile", "Filename for 2 pass stats", OFFSET(stats), FF_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
     {"wpredp", "Weighted prediction for P-frames", OFFSET(weightp), FF_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
     {"x264opts", "x264 options", OFFSET(x264opts), FF_OPT_TYPE_STRING, {.str=NULL}, 0, 0, VE},
0140d3f0
     { NULL },
 };
 
 static const AVClass class = { "libx264", av_default_item_name, options, LIBAVUTIL_VERSION_INT };
 
e7e2df27
 AVCodec ff_libx264_encoder = {
8a8720c1
     .name           = "libx264",
72415b2a
     .type           = AVMEDIA_TYPE_VIDEO,
8a8720c1
     .id             = CODEC_ID_H264,
6662ec29
     .priv_data_size = sizeof(X264Context),
8a8720c1
     .init           = X264_init,
     .encode         = X264_frame,
     .close          = X264_close,
     .capabilities   = CODEC_CAP_DELAY,
3cf8db94
     .pix_fmts       = (const enum PixelFormat[]) { PIX_FMT_YUV420P, PIX_FMT_YUVJ420P, PIX_FMT_NONE },
8a8720c1
     .long_name      = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
0140d3f0
     .priv_class     = &class,
6662ec29
 };