#include "analyse.h"
#include "ratecontrol.h"
#include "macroblock.h"
+#include "me.h"
-#if VISUALIZE
+#ifdef HAVE_VISUALIZE
#include "common/visualize.h"
#endif
if( !f )
return;
/* Write the frame in display order */
- fseek( f, h->fdec->i_frame * h->param.i_height * h->param.i_width * 3/2, SEEK_SET );
+ fseek( f, (uint64_t)h->fdec->i_frame * h->param.i_height * h->param.i_width * 3/2, SEEK_SET );
for( i = 0; i < h->fdec->i_plane; i++ )
for( y = 0; y < h->param.i_height >> !!i; y++ )
fwrite( &h->fdec->plane[i][y*h->fdec->i_stride[i]], 1, h->param.i_width >> !!i, f );
x264_param_t *param = &h->param;
int i;
- /* First we fill all field */
+ /* First we fill all fields */
sh->sps = sps;
sh->pps = pps;
sh->i_redundant_pic_cnt = 0;
- if( !h->mb.b_direct_auto_read )
+ h->mb.b_direct_auto_write = h->param.analyse.i_direct_mv_pred == X264_DIRECT_PRED_AUTO
+ && h->param.i_bframe
+ && ( h->param.rc.b_stat_write || !h->param.rc.b_stat_read );
+
+ if( !h->mb.b_direct_auto_read && sh->i_type == SLICE_TYPE_B )
{
- if( h->mb.b_direct_auto_write )
- sh->b_direct_spatial_mv_pred = ( h->stat.i_direct_score[1] > h->stat.i_direct_score[0] );
+ if( h->fref1[0]->i_poc_l0ref0 == h->fref0[0]->i_poc )
+ {
+ if( h->mb.b_direct_auto_write )
+ sh->b_direct_spatial_mv_pred = ( h->stat.i_direct_score[1] > h->stat.i_direct_score[0] );
+ else
+ sh->b_direct_spatial_mv_pred = ( param->analyse.i_direct_mv_pred == X264_DIRECT_PRED_SPATIAL );
+ }
else
- sh->b_direct_spatial_mv_pred = ( param->analyse.i_direct_mv_pred == X264_DIRECT_PRED_SPATIAL );
+ {
+ h->mb.b_direct_auto_write = 0;
+ sh->b_direct_spatial_mv_pred = 1;
+ }
}
/* else b_direct_spatial_mv_pred was read from the 2pass statsfile */
}
}
- if( ( sh->pps->b_weighted_pred && ( sh->i_type == SLICE_TYPE_P || sh->i_type == SLICE_TYPE_SP ) ) ||
- ( sh->pps->b_weighted_bipred == 1 && sh->i_type == SLICE_TYPE_B ) )
+ if( sh->pps->b_weighted_pred && ( sh->i_type == SLICE_TYPE_P || sh->i_type == SLICE_TYPE_SP ) )
{
- /* FIXME */
+ /* pred_weight_table() */
+ bs_write_ue( s, sh->weight[0][0].i_denom );
+ bs_write_ue( s, sh->weight[0][1].i_denom );
+ for( i = 0; i < sh->i_num_ref_idx_l0_active; i++ )
+ {
+ int luma_weight_l0_flag = !!sh->weight[i][0].weightfn;
+ int chroma_weight_l0_flag = !!sh->weight[i][1].weightfn || !!sh->weight[i][2].weightfn;
+ bs_write1( s, luma_weight_l0_flag );
+ if( luma_weight_l0_flag )
+ {
+ bs_write_se( s, sh->weight[i][0].i_scale );
+ bs_write_se( s, sh->weight[i][0].i_offset );
+ }
+ bs_write1( s, chroma_weight_l0_flag );
+ if( chroma_weight_l0_flag )
+ {
+ int j;
+ for( j = 1; j < 3; j++ )
+ {
+ bs_write_se( s, sh->weight[i][j].i_scale );
+ bs_write_se( s, sh->weight[i][j].i_offset );
+ }
+ }
+ }
+ }
+ else if( sh->pps->b_weighted_bipred == 1 && sh->i_type == SLICE_TYPE_B )
+ {
+ /* TODO */
}
if( i_nal_ref_idc != 0 )
h->param.i_width, h->param.i_height );
return -1;
}
- if( h->param.i_csp != X264_CSP_I420 )
+ int i_csp = h->param.i_csp & X264_CSP_MASK;
+ if( i_csp != X264_CSP_I420 && i_csp != X264_CSP_YV12 )
{
- x264_log( h, X264_LOG_ERROR, "invalid CSP (only I420 supported)\n" );
+ x264_log( h, X264_LOG_ERROR, "invalid CSP (only I420/YV12 supported)\n" );
return -1;
}
if( h->param.i_threads == X264_THREADS_AUTO )
- h->param.i_threads = x264_cpu_num_processors() * 3/2;
+ h->param.i_threads = x264_cpu_num_processors() * (h->param.b_sliced_threads?2:3)/2;
h->param.i_threads = x264_clip3( h->param.i_threads, 1, X264_THREAD_MAX );
if( h->param.i_threads > 1 )
{
x264_log( h, X264_LOG_WARNING, "not compiled with pthread support!\n");
h->param.i_threads = 1;
#endif
+ /* Avoid absurdly small thread slices as they can reduce performance
+ * and VBV compliance. Capped at an arbitrary 4 rows per thread. */
+ if( h->param.b_sliced_threads )
+ {
+ int max_threads = (h->param.i_height+15)/16 / 4;
+ h->param.i_threads = X264_MIN( h->param.i_threads, max_threads );
+ }
}
+ else
+ h->param.b_sliced_threads = 0;
+ h->i_thread_frames = h->param.b_sliced_threads ? 1 : h->param.i_threads;
if( h->param.b_interlaced )
{
x264_log( h, X264_LOG_WARNING, "interlace + me=esa is not implemented\n" );
h->param.analyse.i_me_method = X264_ME_UMH;
}
- if( h->param.analyse.i_direct_mv_pred > X264_DIRECT_PRED_SPATIAL )
+ if( h->param.analyse.i_weighted_pred > 0 )
{
- x264_log( h, X264_LOG_WARNING, "interlace + direct=temporal is not implemented\n" );
- h->param.analyse.i_direct_mv_pred = X264_DIRECT_PRED_SPATIAL;
+ x264_log( h, X264_LOG_WARNING, "interlace + weightp is not implemented\n" );
+ h->param.analyse.i_weighted_pred = X264_WEIGHTP_NONE;
}
}
}
h->param.rc.i_qp_max = x264_clip3( h->param.rc.i_qp_max, 0, 51 );
h->param.rc.i_qp_min = x264_clip3( h->param.rc.i_qp_min, 0, h->param.rc.i_qp_max );
-
- if( ( h->param.i_width % 16 || h->param.i_height % 16 )
- && h->param.i_height != 1080 && !h->mb.b_lossless )
+ if( h->param.rc.i_vbv_buffer_size )
{
- // There's nothing special about 1080 in that the warning still applies to it,
- // but chances are the user can't help it if his content is already 1080p,
- // so there's no point in warning in that case.
- x264_log( h, X264_LOG_WARNING,
- "width or height not divisible by 16 (%dx%d), compression will suffer.\n",
- h->param.i_width, h->param.i_height );
+ if( h->param.rc.i_rc_method == X264_RC_CQP )
+ {
+ x264_log( h, X264_LOG_WARNING, "VBV is incompatible with constant QP, ignored.\n" );
+ h->param.rc.i_vbv_max_bitrate = 0;
+ h->param.rc.i_vbv_buffer_size = 0;
+ }
+ else if( h->param.rc.i_vbv_max_bitrate == 0 )
+ {
+ if( h->param.rc.i_rc_method == X264_RC_ABR )
+ {
+ x264_log( h, X264_LOG_WARNING, "VBV maxrate unspecified, assuming CBR\n" );
+ h->param.rc.i_vbv_max_bitrate = h->param.rc.i_bitrate;
+ }
+ else
+ {
+ x264_log( h, X264_LOG_WARNING, "VBV bufsize set but maxrate unspecified, ignored\n" );
+ h->param.rc.i_vbv_buffer_size = 0;
+ }
+ }
+ else if( h->param.rc.i_vbv_max_bitrate < h->param.rc.i_bitrate &&
+ h->param.rc.i_rc_method == X264_RC_ABR )
+ {
+ x264_log( h, X264_LOG_WARNING, "max bitrate less than average bitrate, assuming CBR\n" );
+ h->param.rc.i_vbv_max_bitrate = h->param.rc.i_bitrate;
+ }
}
-
- int max_slices = (h->param.i_height+((16<<h->param.b_interlaced)-1))/(16<<h->param.b_interlaced);
- h->param.i_slice_count = x264_clip3( h->param.i_slice_count, 0, max_slices );
- h->param.i_slice_max_size = X264_MAX( h->param.i_slice_max_size, 0 );
- h->param.i_slice_max_mbs = X264_MAX( h->param.i_slice_max_mbs, 0 );
- if( h->param.b_interlaced && h->param.i_slice_max_size )
+ else if( h->param.rc.i_vbv_max_bitrate )
{
- x264_log( h, X264_LOG_WARNING, "interlaced + slice-max-size is not implemented\n" );
- h->param.i_slice_max_size = 0;
+ x264_log( h, X264_LOG_WARNING, "VBV maxrate specified, but no bufsize, ignored\n" );
+ h->param.rc.i_vbv_max_bitrate = 0;
}
- if( h->param.b_interlaced && h->param.i_slice_max_mbs )
+
+ int max_slices = (h->param.i_height+((16<<h->param.b_interlaced)-1))/(16<<h->param.b_interlaced);
+ if( h->param.b_sliced_threads )
+ h->param.i_slice_count = x264_clip3( h->param.i_threads, 0, max_slices );
+ else
{
- x264_log( h, X264_LOG_WARNING, "interlaced + slice-max-mbs is not implemented\n" );
- h->param.i_slice_max_mbs = 0;
+ h->param.i_slice_count = x264_clip3( h->param.i_slice_count, 0, max_slices );
+ h->param.i_slice_max_size = X264_MAX( h->param.i_slice_max_size, 0 );
+ h->param.i_slice_max_mbs = X264_MAX( h->param.i_slice_max_mbs, 0 );
+ if( h->param.b_interlaced && h->param.i_slice_max_size )
+ {
+ x264_log( h, X264_LOG_WARNING, "interlaced + slice-max-size is not implemented\n" );
+ h->param.i_slice_max_size = 0;
+ }
+ if( h->param.b_interlaced && h->param.i_slice_max_mbs )
+ {
+ x264_log( h, X264_LOG_WARNING, "interlaced + slice-max-mbs is not implemented\n" );
+ h->param.i_slice_max_mbs = 0;
+ }
+ if( h->param.i_slice_max_mbs || h->param.i_slice_max_size )
+ h->param.i_slice_count = 0;
}
- if( h->param.i_slice_max_mbs || h->param.i_slice_max_size )
- h->param.i_slice_count = 0;
h->param.i_frame_reference = x264_clip3( h->param.i_frame_reference, 1, 16 );
if( h->param.i_keyint_max <= 0 )
h->param.i_keyint_max = 1;
if( h->param.i_scenecut_threshold < 0 )
h->param.i_scenecut_threshold = 0;
- h->param.i_keyint_min = x264_clip3( h->param.i_keyint_min, 1, h->param.i_keyint_max/2+1 );
if( !h->param.analyse.i_subpel_refine && h->param.analyse.i_direct_mv_pred > X264_DIRECT_PRED_SPATIAL )
{
x264_log( h, X264_LOG_WARNING, "subme=0 + direct=temporal is not supported\n" );
}
h->param.i_bframe = x264_clip3( h->param.i_bframe, 0, X264_BFRAME_MAX );
if( h->param.i_keyint_max == 1 )
+ {
h->param.i_bframe = 0;
+ h->param.b_intra_refresh = 0;
+ }
h->param.i_bframe_bias = x264_clip3( h->param.i_bframe_bias, -90, 100 );
if( h->param.i_bframe <= 1 )
h->param.i_bframe_pyramid = X264_B_PYRAMID_NONE;
h->param.analyse.i_direct_mv_pred = 0;
h->param.analyse.b_weighted_bipred = 0;
}
+ if( h->param.b_intra_refresh && h->param.i_bframe_pyramid == X264_B_PYRAMID_NORMAL )
+ {
+ x264_log( h, X264_LOG_WARNING, "b-pyramid normal + intra-refresh is not supported\n" );
+ h->param.i_bframe_pyramid = X264_B_PYRAMID_STRICT;
+ }
+ if( h->param.b_intra_refresh && h->param.i_frame_reference > 1 )
+ {
+ x264_log( h, X264_LOG_WARNING, "ref > 1 + intra-refresh is not supported\n" );
+ h->param.i_frame_reference = 1;
+ }
+ h->param.i_keyint_min = x264_clip3( h->param.i_keyint_min, 1, h->param.i_keyint_max/2+1 );
h->param.rc.i_lookahead = x264_clip3( h->param.rc.i_lookahead, 0, X264_LOOKAHEAD_MAX );
{
int maxrate = X264_MAX( h->param.rc.i_vbv_max_bitrate, h->param.rc.i_bitrate );
h->param.rc.i_lookahead = X264_MIN( h->param.rc.i_lookahead, X264_MAX( h->param.i_keyint_max, bufsize*fps ) );
}
+ if( !h->param.i_timebase_num || !h->param.i_timebase_den )
+ {
+ h->param.i_timebase_num = h->param.i_fps_den;
+ h->param.i_timebase_den = h->param.i_fps_num;
+ }
+
h->param.rc.f_qcompress = x264_clip3f( h->param.rc.f_qcompress, 0.0, 1.0 );
if( !h->param.rc.i_lookahead || h->param.i_keyint_max == 1 || h->param.rc.f_qcompress == 1 )
h->param.rc.b_mb_tree = 0;
h->param.rc.i_lookahead = 0;
#ifdef HAVE_PTHREAD
if( h->param.i_sync_lookahead )
- h->param.i_sync_lookahead = x264_clip3( h->param.i_sync_lookahead, h->param.i_threads + h->param.i_bframe, X264_LOOKAHEAD_MAX );
- if( h->param.rc.b_stat_read || h->param.i_threads == 1 )
+ h->param.i_sync_lookahead = x264_clip3( h->param.i_sync_lookahead, h->i_thread_frames + h->param.i_bframe, X264_LOOKAHEAD_MAX );
+ if( h->param.rc.b_stat_read || h->i_thread_frames == 1 )
h->param.i_sync_lookahead = 0;
#else
h->param.i_sync_lookahead = 0;
#endif
- h->mb.b_direct_auto_write = h->param.analyse.i_direct_mv_pred == X264_DIRECT_PRED_AUTO
- && h->param.i_bframe
- && ( h->param.rc.b_stat_write || !h->param.rc.b_stat_read );
-
h->param.i_deblocking_filter_alphac0 = x264_clip3( h->param.i_deblocking_filter_alphac0, -6, 6 );
h->param.i_deblocking_filter_beta = x264_clip3( h->param.i_deblocking_filter_beta, -6, 6 );
h->param.analyse.i_luma_deadzone[0] = x264_clip3( h->param.analyse.i_luma_deadzone[0], 0, 32 );
/* Psy trellis has a similar effect. */
if( h->mb.i_psy_trellis )
h->param.analyse.i_chroma_qp_offset -= h->param.analyse.f_psy_trellis < 0.25 ? 1 : 2;
- else
- h->mb.i_psy_trellis = 0;
h->param.analyse.i_chroma_qp_offset = x264_clip3(h->param.analyse.i_chroma_qp_offset, -12, 12);
h->param.rc.i_aq_mode = x264_clip3( h->param.rc.i_aq_mode, 0, 2 );
h->param.rc.f_aq_strength = x264_clip3f( h->param.rc.f_aq_strength, 0, 3 );
h->param.rc.i_aq_mode = 1;
h->param.rc.f_aq_strength = 0;
}
- if( h->param.rc.b_mb_tree && h->param.i_bframe_pyramid )
- {
- x264_log( h, X264_LOG_WARNING, "b-pyramid + mb-tree is not supported\n" );
- h->param.i_bframe_pyramid = X264_B_PYRAMID_NONE;
- }
h->param.analyse.i_noise_reduction = x264_clip3( h->param.analyse.i_noise_reduction, 0, 1<<16 );
if( h->param.analyse.i_subpel_refine == 10 && (h->param.analyse.i_trellis != 2 || !h->param.rc.i_aq_mode) )
h->param.analyse.i_subpel_refine = 9;
h->param.analyse.i_mv_range = x264_clip3(h->param.analyse.i_mv_range, 32, 512 >> h->param.b_interlaced);
}
- if( h->param.i_threads > 1 )
+ h->param.analyse.i_weighted_pred = x264_clip3( h->param.analyse.i_weighted_pred, 0, X264_WEIGHTP_SMART );
+ if( !h->param.analyse.i_weighted_pred && h->param.rc.b_mb_tree && h->param.analyse.b_psy && !h->param.b_interlaced )
+ h->param.analyse.i_weighted_pred = X264_WEIGHTP_FAKE;
+
+ if( h->i_thread_frames > 1 )
{
int r = h->param.analyse.i_mv_range_thread;
int r2;
// the rest is allocated to whichever thread is far enough ahead to use it.
// reserving more space increases quality for some videos, but costs more time
// in thread synchronization.
- int max_range = (h->param.i_height + X264_THREAD_HEIGHT) / h->param.i_threads - X264_THREAD_HEIGHT;
+ int max_range = (h->param.i_height + X264_THREAD_HEIGHT) / h->i_thread_frames - X264_THREAD_HEIGHT;
r = max_range / 2;
}
r = X264_MAX( r, h->param.analyse.i_me_range );
/* ensure the booleans are 0 or 1 so they can be used in math */
#define BOOLIFY(x) h->param.x = !!h->param.x
BOOLIFY( b_cabac );
+ BOOLIFY( b_constrained_intra );
BOOLIFY( b_deblocking_filter );
+ BOOLIFY( b_deterministic );
+ BOOLIFY( b_sliced_threads );
BOOLIFY( b_interlaced );
+ BOOLIFY( b_intra_refresh );
+ BOOLIFY( b_visualize );
+ BOOLIFY( b_aud );
+ BOOLIFY( b_repeat_headers );
+ BOOLIFY( b_annexb );
BOOLIFY( analyse.b_transform_8x8 );
+ BOOLIFY( analyse.b_weighted_bipred );
BOOLIFY( analyse.b_chroma_me );
+ BOOLIFY( analyse.b_mixed_references );
BOOLIFY( analyse.b_fast_pskip );
+ BOOLIFY( analyse.b_dct_decimate );
+ BOOLIFY( analyse.b_psy );
+ BOOLIFY( analyse.b_psnr );
+ BOOLIFY( analyse.b_ssim );
BOOLIFY( rc.b_stat_write );
BOOLIFY( rc.b_stat_read );
BOOLIFY( rc.b_mb_tree );
if( h->param.rc.psz_stat_in )
h->param.rc.psz_stat_in = strdup( h->param.rc.psz_stat_in );
- x264_set_aspect_ratio( h, param, 1 );
+ x264_set_aspect_ratio( h, &h->param, 1 );
x264_reduce_fraction( &h->param.i_fps_num, &h->param.i_fps_den );
+ x264_reduce_fraction( &h->param.i_timebase_num, &h->param.i_timebase_den );
/* Init x264_t */
h->i_frame = -1;
h->i_frame_num = 0;
h->i_idr_pic_id = 0;
+ if( h->param.b_dts_compress )
+ {
+ /* h->i_dts_compress_multiplier == h->frames.i_bframe_delay + 1 */
+ h->i_dts_compress_multiplier = h->param.i_bframe ? (h->param.i_bframe_pyramid ? 3 : 2) : 1;
+ if( h->i_dts_compress_multiplier != 1 )
+ x264_log( h, X264_LOG_DEBUG, "DTS compresion changed timebase: %d/%d -> %d/%d\n",
+ h->param.i_timebase_num, h->param.i_timebase_den,
+ h->param.i_timebase_num, h->param.i_timebase_den * h->i_dts_compress_multiplier );
+ h->param.i_timebase_den *= h->i_dts_compress_multiplier;
+ }
+ else
+ h->i_dts_compress_multiplier = 1;
h->sps = &h->sps_array[0];
x264_sps_init( h->sps, h->param.i_sps_id, &h->param );
if( h->param.rc.b_mb_tree || h->param.rc.i_vbv_buffer_size )
h->frames.i_delay = X264_MAX( h->frames.i_delay, h->param.rc.i_lookahead );
i_slicetype_length = h->frames.i_delay;
- h->frames.i_delay += h->param.i_threads - 1;
+ h->frames.i_delay += h->i_thread_frames - 1;
h->frames.i_delay = X264_MIN( h->frames.i_delay, X264_LOOKAHEAD_MAX );
h->frames.i_delay += h->param.i_sync_lookahead;
+ h->frames.i_bframe_delay = h->param.i_bframe ? (h->param.i_bframe_pyramid ? 2 : 1) : 0;
h->frames.i_max_ref0 = h->param.i_frame_reference;
h->frames.i_max_ref1 = h->sps->vui.i_num_reorder_frames;
|| h->param.rc.i_rc_method == X264_RC_CRF
|| h->param.i_bframe_adaptive
|| h->param.i_scenecut_threshold
- || h->param.rc.b_mb_tree );
+ || h->param.rc.b_mb_tree
+ || h->param.analyse.i_weighted_pred == X264_WEIGHTP_SMART );
h->frames.b_have_lowres |= h->param.rc.b_stat_read && h->param.rc.i_vbv_buffer_size > 0;
h->frames.b_have_sub8x8_esa = !!(h->param.analyse.inter & X264_ANALYSE_PSUB8x8);
- h->frames.i_last_idr = - h->param.i_keyint_max;
+ h->frames.i_last_keyframe = - h->param.i_keyint_max;
h->frames.i_input = 0;
CHECKED_MALLOCZERO( h->frames.unused[0], (h->frames.i_delay + 3) * sizeof(x264_frame_t *) );
/* Allocate room for max refs plus a few extra just in case. */
- CHECKED_MALLOCZERO( h->frames.unused[1], (h->param.i_threads + 20) * sizeof(x264_frame_t *) );
+ CHECKED_MALLOCZERO( h->frames.unused[1], (h->i_thread_frames + 20) * sizeof(x264_frame_t *) );
CHECKED_MALLOCZERO( h->frames.current, (h->param.i_sync_lookahead + h->param.i_bframe
- + h->param.i_threads + 3) * sizeof(x264_frame_t *) );
-
+ + h->i_thread_frames + 3) * sizeof(x264_frame_t *) );
+ if( h->param.analyse.i_weighted_pred > 0 )
+ CHECKED_MALLOCZERO( h->frames.blank_unused, h->i_thread_frames * 4 * sizeof(x264_frame_t *) );
h->i_ref0 = 0;
h->i_ref1 = 0;
for( i=0; x264_cpu_names[i].flags; i++ )
{
if( !strcmp(x264_cpu_names[i].name, "SSE2")
- && param->cpu & (X264_CPU_SSE2_IS_FAST|X264_CPU_SSE2_IS_SLOW) )
+ && h->param.cpu & (X264_CPU_SSE2_IS_FAST|X264_CPU_SSE2_IS_SLOW) )
continue;
if( !strcmp(x264_cpu_names[i].name, "SSE3")
- && (param->cpu & X264_CPU_SSSE3 || !(param->cpu & X264_CPU_CACHELINE_64)) )
+ && (h->param.cpu & X264_CPU_SSSE3 || !(h->param.cpu & X264_CPU_CACHELINE_64)) )
continue;
if( !strcmp(x264_cpu_names[i].name, "SSE4.1")
- && (param->cpu & X264_CPU_SSE42) )
+ && (h->param.cpu & X264_CPU_SSE42) )
continue;
- if( (param->cpu & x264_cpu_names[i].flags) == x264_cpu_names[i].flags
+ if( (h->param.cpu & x264_cpu_names[i].flags) == x264_cpu_names[i].flags
&& (!i || x264_cpu_names[i].flags != x264_cpu_names[i-1].flags) )
p += sprintf( p, " %s", x264_cpu_names[i].name );
}
- if( !param->cpu )
+ if( !h->param.cpu )
p += sprintf( p, " none!" );
x264_log( h, X264_LOG_INFO, "%s\n", buf );
h->nal_buffer_size = h->out.i_bitstream * 3/2 + 4;
h->thread[0] = h;
- h->i_thread_num = 0;
for( i = 1; i < h->param.i_threads + !!h->param.i_sync_lookahead; i++ )
CHECKED_MALLOC( h->thread[i], sizeof(x264_t) );
+ if( x264_lookahead_init( h, i_slicetype_length ) )
+ goto fail;
+
for( i = 0; i < h->param.i_threads; i++ )
{
+ int init_nal_count = h->param.i_slice_count + 3;
+ int allocate_threadlocal_data = !h->param.b_sliced_threads || !i;
if( i > 0 )
*h->thread[i] = *h;
- h->thread[i]->fdec = x264_frame_pop_unused( h, 1 );
- if( !h->thread[i]->fdec )
- goto fail;
+
+ if( allocate_threadlocal_data )
+ {
+ h->thread[i]->fdec = x264_frame_pop_unused( h, 1 );
+ if( !h->thread[i]->fdec )
+ goto fail;
+ }
+ else
+ h->thread[i]->fdec = h->thread[0]->fdec;
+
CHECKED_MALLOC( h->thread[i]->out.p_bitstream, h->out.i_bitstream );
- /* Start each thread with room for 8 NAL units; it'll realloc later if needed. */
- CHECKED_MALLOC( h->thread[i]->out.nal, 8*sizeof(x264_nal_t) );
- h->thread[i]->out.i_nals_allocated = 8;
- if( x264_macroblock_cache_init( h->thread[i] ) < 0 )
+ /* Start each thread with room for init_nal_count NAL units; it'll realloc later if needed. */
+ CHECKED_MALLOC( h->thread[i]->out.nal, init_nal_count*sizeof(x264_nal_t) );
+ h->thread[i]->out.i_nals_allocated = init_nal_count;
+
+ if( allocate_threadlocal_data && x264_macroblock_cache_init( h->thread[i] ) < 0 )
goto fail;
}
- if( x264_lookahead_init( h, i_slicetype_length ) )
- goto fail;
+ /* Allocate scratch buffer */
+ for( i = 0; i < h->param.i_threads + !!h->param.i_sync_lookahead; i++ )
+ {
+ int buf_hpel = (h->param.i_width+48) * sizeof(int16_t);
+ int buf_ssim = h->param.analyse.b_ssim * 8 * (h->param.i_width/4+3) * sizeof(int);
+ int me_range = X264_MIN(h->param.analyse.i_me_range, h->param.analyse.i_mv_range);
+ int buf_tesa = (h->param.analyse.i_me_method >= X264_ME_ESA) *
+ ((me_range*2+18) * sizeof(int16_t) + (me_range+4) * (me_range+1) * 4 * sizeof(mvsad_t));
+ int buf_mbtree = h->param.rc.b_mb_tree * ((h->sps->i_mb_width+3)&~3) * sizeof(int);
+ int buf_nnz = !h->param.b_cabac * h->pps->b_transform_8x8_mode * (h->sps->i_mb_width * 4 * 16 * sizeof(uint8_t));
+ int scratch_size = X264_MAX4( buf_hpel, buf_ssim, buf_tesa, X264_MAX( buf_mbtree, buf_nnz ) );
+ CHECKED_MALLOC( h->thread[i]->scratch_buffer, scratch_size );
+ }
if( x264_ratecontrol_new( h ) < 0 )
goto fail;
{
/* create or truncate the reconstructed video file */
FILE *f = fopen( h->param.psz_dump_yuv, "w" );
- if( f )
- fclose( f );
- else
+ if( !f )
{
- x264_log( h, X264_LOG_ERROR, "can't write to fdec.yuv\n" );
+ x264_log( h, X264_LOG_ERROR, "dump_yuv: can't write to %s\n", h->param.psz_dump_yuv );
goto fail;
}
+ else if( !x264_is_regular_file( f ) )
+ {
+ x264_log( h, X264_LOG_ERROR, "dump_yuv: incompatible with non-regular file %s\n", h->param.psz_dump_yuv );
+ goto fail;
+ }
+ fclose( f );
}
x264_log( h, X264_LOG_INFO, "profile %s, level %d.%d\n",
****************************************************************************/
int x264_encoder_reconfig( x264_t *h, x264_param_t *param )
{
- h = h->thread[h->i_thread_phase];
+ h = h->thread[h->thread[0]->i_thread_phase];
x264_set_aspect_ratio( h, param, 0 );
#define COPY(var) h->param.var = param->var
COPY( i_frame_reference ); // but never uses more refs than initially specified
COPY( b_deblocking_filter );
COPY( i_deblocking_filter_alphac0 );
COPY( i_deblocking_filter_beta );
- COPY( analyse.intra );
COPY( analyse.inter );
+ COPY( analyse.intra );
COPY( analyse.i_direct_mv_pred );
/* Scratch buffer prevents me_range from being increased for esa/tesa */
if( h->param.analyse.i_me_method < X264_ME_ESA || param->analyse.i_me_range < h->param.analyse.i_me_range )
COPY( i_slice_max_size );
COPY( i_slice_max_mbs );
COPY( i_slice_count );
+ /* VBV can't be turned on if it wasn't on to begin with */
+ if( h->param.rc.i_vbv_max_bitrate > 0 && h->param.rc.i_vbv_buffer_size > 0 &&
+ param->rc.i_vbv_max_bitrate > 0 && param->rc.i_vbv_buffer_size > 0 )
+ {
+ COPY( rc.i_vbv_max_bitrate );
+ COPY( rc.i_vbv_buffer_size );
+ COPY( rc.i_bitrate );
+ }
+ COPY( rc.f_rf_constant );
#undef COPY
mbcmp_init( h );
- return x264_validate_parameters( h );
+ int ret = x264_validate_parameters( h );
+
+ /* Supported reconfiguration options (1-pass only):
+ * vbv-maxrate
+ * vbv-bufsize
+ * crf
+ * bitrate (CBR only) */
+ if( !ret )
+ x264_ratecontrol_init_reconfigurable( h, 0 );
+
+ return ret;
+}
+
+/****************************************************************************
+ * x264_encoder_parameters:
+ ****************************************************************************/
+void x264_encoder_parameters( x264_t *h, x264_param_t *param )
+{
+ memcpy( param, &h->thread[h->i_thread_phase]->param, sizeof(x264_param_t) );
}
/* internal usage */
nal->i_payload= 0;
nal->p_payload= &h->out.p_bitstream[bs_pos( &h->out.bs ) / 8];
}
-static int x264_nal_end( x264_t *h )
+/* if number of allocated nals is not enough, re-allocate a larger one. */
+static int x264_nal_check_buffer( x264_t *h )
{
- x264_nal_t *nal = &h->out.nal[h->out.i_nal];
- nal->i_payload = &h->out.p_bitstream[bs_pos( &h->out.bs ) / 8] - nal->p_payload;
- h->out.i_nal++;
-
- /* if number of allocated nals is not enough, re-allocate a larger one. */
if( h->out.i_nal >= h->out.i_nals_allocated )
{
x264_nal_t *new_out = x264_malloc( sizeof(x264_nal_t) * (h->out.i_nals_allocated*2) );
}
return 0;
}
+static int x264_nal_end( x264_t *h )
+{
+ x264_nal_t *nal = &h->out.nal[h->out.i_nal];
+ nal->i_payload = &h->out.p_bitstream[bs_pos( &h->out.bs ) / 8] - nal->p_payload;
+ h->out.i_nal++;
+
+ return x264_nal_check_buffer( h );
+}
static int x264_encoder_encapsulate_nals( x264_t *h )
{
x264_pps_write( &h->out.bs, h->pps );
if( x264_nal_end( h ) )
return -1;
- bs_flush( &h->out.bs );
frame_size = x264_encoder_encapsulate_nals( h );
}
}
-static inline void x264_reference_build_list( x264_t *h, int i_poc )
+/* return -1 on failure, else return the index of the new reference frame */
+int x264_weighted_reference_duplicate( x264_t *h, int i_ref, const x264_weight_t *w )
+{
+ int i = h->i_ref0;
+ int j;
+ x264_frame_t *newframe;
+ if( i <= 1 ) /* empty list, definitely can't duplicate frame */
+ return -1;
+
+ /* Find a place to insert the duplicate in the reference list. */
+ for( j = 0; j < i; j++ )
+ if( h->fref0[i_ref]->i_frame != h->fref0[j]->i_frame )
+ {
+ /* found a place, after j, make sure there is not already a duplicate there */
+ if( j == i-1 || ( h->fref0[j+1] && h->fref0[i_ref]->i_frame != h->fref0[j+1]->i_frame ) )
+ break;
+ }
+
+ if( j == i ) /* No room in the reference list for the duplicate. */
+ return -1;
+ j++;
+
+ newframe = x264_frame_pop_blank_unused( h );
+
+ //FIXME: probably don't need to copy everything
+ *newframe = *h->fref0[i_ref];
+ newframe->i_reference_count = 1;
+ newframe->orig = h->fref0[i_ref];
+ newframe->b_duplicate = 1;
+ memcpy( h->fenc->weight[j], w, sizeof(h->fenc->weight[i]) );
+
+ /* shift the frames to make space for the dupe. */
+ h->b_ref_reorder[0] = 1;
+ if( h->i_ref0 < 16 )
+ ++h->i_ref0;
+ h->fref0[15] = NULL;
+ x264_frame_unshift( &h->fref0[j], newframe );
+
+ return j;
+}
+
+static void x264_weighted_pred_init( x264_t *h )
{
+ int i_ref;
int i;
- int b_ok;
+
+ /* for now no analysis and set all weights to nothing */
+ for( i_ref = 0; i_ref < h->i_ref0; i_ref++ )
+ h->fenc->weighted[i_ref] = h->fref0[i_ref]->filtered[0];
+
+ // FIXME: This only supports weighting of one reference frame
+ // and duplicates of that frame.
+ h->fenc->i_lines_weighted = 0;
+
+ for( i_ref = 0; i_ref < (h->i_ref0 << h->sh.b_mbaff); i_ref++ )
+ for( i = 0; i < 3; i++ )
+ h->sh.weight[i_ref][i].weightfn = NULL;
+
+
+ if( h->sh.i_type != SLICE_TYPE_P || h->param.analyse.i_weighted_pred <= 0 )
+ return;
+
+ int i_padv = PADV << h->param.b_interlaced;
+ int denom = -1;
+ int weightluma = 0;
+ int buffer_next = 0;
+ int j;
+ //FIXME: when chroma support is added, move this into loop
+ h->sh.weight[0][1].weightfn = h->sh.weight[0][2].weightfn = NULL;
+ h->sh.weight[0][1].i_denom = h->sh.weight[0][2].i_denom = 0;
+ for( j = 0; j < h->i_ref0; j++ )
+ {
+ if( h->fenc->weight[j][0].weightfn )
+ {
+ h->sh.weight[j][0] = h->fenc->weight[j][0];
+ // if weight is useless, don't write it to stream
+ if( h->sh.weight[j][0].i_scale == 1<<h->sh.weight[j][0].i_denom && h->sh.weight[j][0].i_offset == 0 )
+ h->sh.weight[j][0].weightfn = NULL;
+ else
+ {
+ if( !weightluma )
+ {
+ weightluma = 1;
+ h->sh.weight[0][0].i_denom = denom = h->sh.weight[j][0].i_denom;
+ assert( x264_clip3( denom, 0, 7 ) == denom );
+ }
+ assert( h->sh.weight[j][0].i_denom == denom );
+ assert( x264_clip3( h->sh.weight[j][0].i_scale, 0, 127 ) == h->sh.weight[j][0].i_scale );
+ assert( x264_clip3( h->sh.weight[j][0].i_offset, -128, 127 ) == h->sh.weight[j][0].i_offset );
+ h->fenc->weighted[j] = h->mb.p_weight_buf[buffer_next++] +
+ h->fenc->i_stride[0] * i_padv + PADH;
+ }
+ }
+
+ //scale full resolution frame
+ if( h->sh.weight[j][0].weightfn && h->param.i_threads == 1 )
+ {
+ uint8_t *src = h->fref0[j]->filtered[0] - h->fref0[j]->i_stride[0]*i_padv - PADH;
+ uint8_t *dst = h->fenc->weighted[j] - h->fenc->i_stride[0]*i_padv - PADH;
+ int stride = h->fenc->i_stride[0];
+ int width = h->fenc->i_width[0] + PADH*2;
+ int height = h->fenc->i_lines[0] + i_padv*2;
+ x264_weight_scale_plane( h, dst, stride, src, stride, width, height, &h->sh.weight[j][0] );
+ h->fenc->i_lines_weighted = height;
+ }
+ }
+ if( !weightluma )
+ h->sh.weight[0][0].i_denom = 0;
+}
+
+static inline void x264_reference_build_list( x264_t *h, int i_poc )
+{
+ int i, b_ok;
/* build ref list 0/1 */
- h->i_ref0 = 0;
- h->i_ref1 = 0;
+ h->mb.pic.i_fref[0] = h->i_ref0 = 0;
+ h->mb.pic.i_fref[1] = h->i_ref1 = 0;
+ if( h->sh.i_type == SLICE_TYPE_I )
+ return;
+
for( i = 0; h->frames.reference[i]; i++ )
{
if( h->frames.reference[i]->i_poc < i_poc )
h->i_ref1 = X264_MIN( h->i_ref1, h->frames.i_max_ref1 );
h->i_ref0 = X264_MIN( h->i_ref0, h->frames.i_max_ref0 );
h->i_ref0 = X264_MIN( h->i_ref0, h->param.i_frame_reference ); // if reconfig() has lowered the limit
+
+ /* add duplicates */
+ if( h->fenc->i_type == X264_TYPE_P )
+ {
+ int idx = -1;
+ if( h->param.analyse.i_weighted_pred == X264_WEIGHTP_SMART )
+ {
+ x264_weight_t w[3];
+ w[1].weightfn = w[2].weightfn = NULL;
+ if( h->param.rc.b_stat_read )
+ x264_ratecontrol_set_weights( h, h->fenc );
+
+ if( !h->fenc->weight[0][0].weightfn )
+ {
+ h->fenc->weight[0][0].i_denom = 0;
+ SET_WEIGHT( w[0], 1, 1, 0, -1 );
+ idx = x264_weighted_reference_duplicate( h, 0, w );
+ }
+ else
+ {
+ if( h->fenc->weight[0][0].i_scale == 1<<h->fenc->weight[0][0].i_denom )
+ {
+ SET_WEIGHT( h->fenc->weight[0][0], 1, 1, 0, h->fenc->weight[0][0].i_offset );
+ }
+ x264_weighted_reference_duplicate( h, 0, weight_none );
+ if( h->fenc->weight[0][0].i_offset > -128 )
+ {
+ w[0] = h->fenc->weight[0][0];
+ w[0].i_offset--;
+ h->mc.weight_cache( h, &w[0] );
+ idx = x264_weighted_reference_duplicate( h, 0, w );
+ }
+ }
+ }
+ else if( h->param.analyse.i_weighted_pred == X264_WEIGHTP_BLIND )
+ {
+ //weighted offset=-1
+ x264_weight_t w[3];
+ SET_WEIGHT( w[0], 1, 1, 0, -1 );
+ h->fenc->weight[0][0].i_denom = 0;
+ w[1].weightfn = w[2].weightfn = NULL;
+ idx = x264_weighted_reference_duplicate( h, 0, w );
+ }
+ h->mb.ref_blind_dupe = idx;
+ }
+
assert( h->i_ref0 + h->i_ref1 <= 16 );
h->mb.pic.i_fref[0] = h->i_ref0;
h->mb.pic.i_fref[1] = h->i_ref1;
if( min_y < 0 )
return;
- if( !b_end )
+ if( !b_end && !h->param.b_sliced_threads )
{
int i, j;
for( j=0; j<=h->sh.b_mbaff; j++ )
}
}
- if( h->param.i_threads > 1 && h->fdec->b_kept_as_ref )
- {
+ if( h->i_thread_frames > 1 && h->fdec->b_kept_as_ref )
x264_frame_cond_broadcast( h->fdec, mb_y*16 + (b_end ? 10000 : -(X264_THREAD_HEIGHT << h->sh.b_mbaff)) );
- }
min_y = X264_MAX( min_y*16-8, 0 );
max_y = b_end ? h->param.i_height : mb_y*16-8;
int i, j;
if( !h->fdec->b_kept_as_ref )
{
- if( h->param.i_threads > 1 )
+ if( h->i_thread_frames > 1 )
{
x264_frame_push_unused( h, h->fdec );
h->fdec = x264_frame_pop_unused( h, 1 );
/* look for delay frames -- chain must only contain frames that are disposable */
for( i = 0; h->frames.current[i] && IS_DISPOSABLE( h->frames.current[i]->i_type ); i++ )
- b_hasdelayframe |= h->frames.current[i]->i_dts
+ b_hasdelayframe |= h->frames.current[i]->i_coded
!= h->frames.current[i]->i_frame + h->sps->vui.i_num_reorder_frames;
if( h->param.i_bframe_pyramid != X264_B_PYRAMID_STRICT && !b_hasdelayframe )
&& h->frames.reference[ref]->i_type == X264_TYPE_BREF )
{
int diff = h->i_frame_num - h->frames.reference[ref]->i_frame_num;
- h->sh.mmco[h->sh.i_mmco_command_count++].i_difference_of_pic_nums = diff;
+ h->sh.mmco[h->sh.i_mmco_command_count].i_difference_of_pic_nums = diff;
+ h->sh.mmco[h->sh.i_mmco_command_count++].i_poc = h->frames.reference[ref]->i_poc;
x264_frame_push_unused( h, x264_frame_pop( h->frames.reference ) );
h->b_ref_reorder[0] = 1;
break;
if( h->sps->i_poc_type == 0 )
{
h->sh.i_poc_lsb = h->fdec->i_poc & ( (1 << h->sps->i_log2_max_poc_lsb) - 1 );
- h->sh.i_delta_poc_bottom = 0; /* XXX won't work for field */
+ h->sh.i_delta_poc_bottom = 0;
}
else if( h->sps->i_poc_type == 1 )
{
/* Assume no more than 3 bytes of NALU escaping. */
int slice_max_size = h->param.i_slice_max_size > 0 ? (h->param.i_slice_max_size-3-NALU_OVERHEAD)*8 : INT_MAX;
int starting_bits = bs_pos(&h->out.bs);
+ bs_realign( &h->out.bs );
/* Slice */
x264_nal_start( h, h->i_nal_type, h->i_nal_ref_idc );
/* Slice header */
+ x264_macroblock_thread_init( h );
x264_slice_header_write( &h->out.bs, &h->sh, h->i_nal_ref_idc );
if( h->param.b_cabac )
{
}
}
- if( i_mb_x == 0 && !h->mb.b_reencode_mb )
+ if( i_mb_x == 0 && !h->mb.b_reencode_mb && !h->param.b_sliced_threads )
x264_fdec_filter_row( h, i_mb_y );
/* load cache */
bs_write_ue( &h->out.bs, i_skip ); /* skip run */
i_skip = 0;
}
- x264_macroblock_write_cavlc( h, &h->out.bs );
+ x264_macroblock_write_cavlc( h );
}
}
else
h->mb.b_reencode_mb = 0;
-#if VISUALIZE
+#ifdef HAVE_VISUALIZE
if( h->param.b_visualize )
x264_visualize_mb( h );
#endif
+ (h->out.i_nal*NALU_OVERHEAD * 8)
- h->stat.frame.i_tex_bits
- h->stat.frame.i_mv_bits;
- x264_fdec_filter_row( h, h->sps->i_mb_height );
+ if( !h->param.b_sliced_threads )
+ x264_fdec_filter_row( h, h->sps->i_mb_height );
}
return 0;
static void x264_thread_sync_context( x264_t *dst, x264_t *src )
{
- x264_frame_t **f;
if( dst == src )
return;
// reference counting
+ x264_frame_t **f;
for( f = src->frames.reference; *f; f++ )
(*f)->i_reference_count++;
for( f = dst->frames.reference; *f; f++ )
static void *x264_slices_write( x264_t *h )
{
int i_slice_num = 0;
+ int last_thread_mb = h->sh.i_last_mb;
if( h->param.i_sync_lookahead )
x264_lower_thread_priority( 10 );
x264_cpu_mask_misalign_sse();
#endif
-#if VISUALIZE
+#ifdef HAVE_VISUALIZE
if( h->param.b_visualize )
if( x264_visualize_init( h ) )
return (void *)-1;
/* init stats */
memset( &h->stat.frame, 0, sizeof(h->stat.frame) );
h->mb.b_reencode_mb = 0;
- while( h->sh.i_first_mb < h->mb.i_mb_count )
+ while( h->sh.i_first_mb <= last_thread_mb )
{
- h->sh.i_last_mb = h->mb.i_mb_count - 1;
+ h->sh.i_last_mb = last_thread_mb;
if( h->param.i_slice_max_mbs )
h->sh.i_last_mb = h->sh.i_first_mb + h->param.i_slice_max_mbs - 1;
- else if( h->param.i_slice_count )
+ else if( h->param.i_slice_count && !h->param.b_sliced_threads )
{
- x264_emms();
- i_slice_num++;
- double height = h->sps->i_mb_height >> h->param.b_interlaced;
+ int height = h->sps->i_mb_height >> h->param.b_interlaced;
int width = h->sps->i_mb_width << h->param.b_interlaced;
- h->sh.i_last_mb = (int)(height * i_slice_num / h->param.i_slice_count + 0.5) * width - 1;
+ i_slice_num++;
+ h->sh.i_last_mb = (height * i_slice_num + h->param.i_slice_count/2) / h->param.i_slice_count * width - 1;
}
- h->sh.i_last_mb = X264_MIN( h->sh.i_last_mb, h->mb.i_mb_count - 1 );
+ h->sh.i_last_mb = X264_MIN( h->sh.i_last_mb, last_thread_mb );
if( x264_stack_align( x264_slice_write, h ) )
return (void *)-1;
h->sh.i_first_mb = h->sh.i_last_mb + 1;
}
-#if VISUALIZE
+#ifdef HAVE_VISUALIZE
if( h->param.b_visualize )
{
x264_visualize_show( h );
return (void *)0;
}
+static int x264_threaded_slices_write( x264_t *h )
+{
+ int i, j;
+ void *ret = NULL;
+ /* set first/last mb and sync contexts */
+ for( i = 0; i < h->param.i_threads; i++ )
+ {
+ x264_t *t = h->thread[i];
+ if( i )
+ {
+ t->param = h->param;
+ memcpy( &t->i_frame, &h->i_frame, offsetof(x264_t, rc) - offsetof(x264_t, i_frame) );
+ }
+ int height = h->sps->i_mb_height >> h->param.b_interlaced;
+ t->i_threadslice_start = ((height * i + h->param.i_slice_count/2) / h->param.i_threads) << h->param.b_interlaced;
+ t->i_threadslice_end = ((height * (i+1) + h->param.i_slice_count/2) / h->param.i_threads) << h->param.b_interlaced;
+ t->sh.i_first_mb = t->i_threadslice_start * h->sps->i_mb_width;
+ t->sh.i_last_mb = t->i_threadslice_end * h->sps->i_mb_width - 1;
+ }
+
+ x264_analyse_weight_frame( h, h->sps->i_mb_height*16 + 16 );
+
+ x264_threads_distribute_ratecontrol( h );
+
+ /* dispatch */
+ for( i = 0; i < h->param.i_threads; i++ )
+ {
+ if( x264_pthread_create( &h->thread[i]->thread_handle, NULL, (void*)x264_slices_write, (void*)h->thread[i] ) )
+ return -1;
+ h->thread[i]->b_thread_active = 1;
+ }
+ for( i = 0; i < h->param.i_threads; i++ )
+ {
+ x264_pthread_join( h->thread[i]->thread_handle, &ret );
+ h->thread[i]->b_thread_active = 0;
+ if( (intptr_t)ret )
+ return (intptr_t)ret;
+ }
+
+ /* deblocking and hpel filtering */
+ for( i = 0; i <= h->sps->i_mb_height; i++ )
+ x264_fdec_filter_row( h, i );
+
+ x264_threads_merge_ratecontrol( h );
+
+ for( i = 1; i < h->param.i_threads; i++ )
+ {
+ x264_t *t = h->thread[i];
+ for( j = 0; j < t->out.i_nal; j++ )
+ {
+ h->out.nal[h->out.i_nal] = t->out.nal[j];
+ h->out.i_nal++;
+ x264_nal_check_buffer( h );
+ }
+ /* All entries in stat.frame are ints except for ssd/ssim,
+ * which are only calculated in the main thread. */
+ for( j = 0; j < (offsetof(x264_t,stat.frame.i_ssd) - offsetof(x264_t,stat.frame.i_mv_bits)) / sizeof(int); j++ )
+ ((int*)&h->stat.frame)[j] += ((int*)&t->stat.frame)[j];
+ }
+
+ return 0;
+}
+
/****************************************************************************
* x264_encoder_encode:
* XXX: i_poc : is the poc of the current given picture
x264_picture_t *pic_out )
{
x264_t *thread_current, *thread_prev, *thread_oldest;
- int i_nal_type;
- int i_nal_ref_idc;
-
- int i_global_qp;
+ int i_nal_type, i_nal_ref_idc, i_global_qp, i;
- if( h->param.i_threads > 1)
+ if( h->i_thread_frames > 1 )
{
thread_prev = h->thread[ h->i_thread_phase ];
- h->i_thread_phase = (h->i_thread_phase + 1) % h->param.i_threads;
+ h->i_thread_phase = (h->i_thread_phase + 1) % h->i_thread_frames;
thread_current = h->thread[ h->i_thread_phase ];
- thread_oldest = h->thread[ (h->i_thread_phase + 1) % h->param.i_threads ];
+ thread_oldest = h->thread[ (h->i_thread_phase + 1) % h->i_thread_frames ];
x264_thread_sync_context( thread_current, thread_prev );
x264_thread_sync_ratecontrol( thread_current, thread_prev, thread_oldest );
h = thread_current;
fenc->i_frame = h->frames.i_input++;
+ if( h->frames.i_bframe_delay && fenc->i_frame == h->frames.i_bframe_delay )
+ h->frames.i_bframe_delay_time = fenc->i_pts;
+
if( h->frames.b_have_lowres )
+ {
+ if( h->param.analyse.i_weighted_pred == X264_WEIGHTP_FAKE || h->param.analyse.i_weighted_pred == X264_WEIGHTP_SMART )
+ x264_weight_plane_analyse( h, fenc );
x264_frame_init_lowres( h, fenc );
+ }
if( h->param.rc.b_mb_tree && h->param.rc.b_stat_read )
{
/* 2: Place the frame into the queue for its slice type decision */
x264_lookahead_put_frame( h, fenc );
- if( h->frames.i_input <= h->frames.i_delay + 1 - h->param.i_threads )
+ if( h->frames.i_input <= h->frames.i_delay + 1 - h->i_thread_frames )
{
/* Nothing yet to encode, waiting for filling of buffers */
pic_out->i_type = X264_TYPE_AUTO;
else
{
/* signal kills for lookahead thread */
+ x264_pthread_mutex_lock( &h->lookahead->ifbuf.mutex );
h->lookahead->b_exit_thread = 1;
x264_pthread_cond_broadcast( &h->lookahead->ifbuf.cv_fill );
+ x264_pthread_mutex_unlock( &h->lookahead->ifbuf.mutex );
}
h->i_frame++;
h->fenc->param->param_free( h->fenc->param );
}
- if( h->fenc->i_type == X264_TYPE_IDR )
+ if( h->fenc->b_keyframe )
{
- h->frames.i_last_idr = h->fenc->i_frame;
- h->i_frame_num = 0;
+ h->frames.i_last_keyframe = h->fenc->i_frame;
+ if( h->fenc->i_type == X264_TYPE_IDR )
+ h->i_frame_num = 0;
}
- h->sh.i_mmco_command_count = 0;
+ h->sh.i_mmco_command_count =
h->sh.i_mmco_remove_from_end = 0;
h->b_ref_reorder[0] =
h->b_ref_reorder[1] = 0;
}
h->fdec->i_poc =
- h->fenc->i_poc = 2 * (h->fenc->i_frame - h->frames.i_last_idr);
+ h->fenc->i_poc = 2 * (h->fenc->i_frame - h->frames.i_last_keyframe);
h->fdec->i_type = h->fenc->i_type;
h->fdec->i_frame = h->fenc->i_frame;
h->fenc->b_kept_as_ref =
/* ---------------------- Write the bitstream -------------------------- */
/* Init bitstream context */
- h->out.i_nal = 0;
- bs_init( &h->out.bs, h->out.p_bitstream, h->out.i_bitstream );
+ if( h->param.b_sliced_threads )
+ {
+ for( i = 0; i < h->param.i_threads; i++ )
+ {
+ bs_init( &h->thread[i]->out.bs, h->thread[i]->out.p_bitstream, h->thread[i]->out.i_bitstream );
+ h->thread[i]->out.i_nal = 0;
+ }
+ }
+ else
+ {
+ bs_init( &h->out.bs, h->out.p_bitstream, h->out.i_bitstream );
+ h->out.i_nal = 0;
+ }
if( h->param.b_aud )
{
int overhead = NALU_OVERHEAD;
+ if( h->param.b_intra_refresh && h->fenc->i_type == X264_TYPE_P )
+ {
+ int pocdiff = (h->fdec->i_poc - h->fref0[0]->i_poc)/2;
+ float increment = X264_MAX( ((float)h->sps->i_mb_width-1) / h->param.i_keyint_max, 1 );
+ int max_position = (int)(increment * h->param.i_keyint_max);
+ if( IS_X264_TYPE_I( h->fref0[0]->i_type ) )
+ h->fdec->f_pir_position = 0;
+ else
+ {
+ h->fdec->f_pir_position = h->fref0[0]->f_pir_position;
+ if( h->fdec->f_pir_position+0.5 >= max_position )
+ {
+ h->fdec->f_pir_position = 0;
+ h->fenc->b_keyframe = 1;
+ }
+ }
+ h->fdec->i_pir_start_col = h->fdec->f_pir_position+0.5;
+ h->fdec->f_pir_position += increment * pocdiff;
+ h->fdec->i_pir_end_col = h->fdec->f_pir_position+0.5;
+ }
+
/* Write SPS and PPS */
- if( i_nal_type == NAL_SLICE_IDR && h->param.b_repeat_headers )
+ if( h->fenc->b_keyframe )
{
- if( h->fenc->i_frame == 0 )
+ if( h->param.b_repeat_headers )
{
- /* identify ourself */
- x264_nal_start( h, NAL_SEI, NAL_PRIORITY_DISPOSABLE );
- if( x264_sei_version_write( h, &h->out.bs ) )
+ if( h->fenc->i_frame == 0 )
+ {
+ /* identify ourself */
+ x264_nal_start( h, NAL_SEI, NAL_PRIORITY_DISPOSABLE );
+ if( x264_sei_version_write( h, &h->out.bs ) )
+ return -1;
+ if( x264_nal_end( h ) )
+ return -1;
+ overhead += h->out.nal[h->out.i_nal-1].i_payload + NALU_OVERHEAD;
+ }
+
+ /* generate sequence parameters */
+ x264_nal_start( h, NAL_SPS, NAL_PRIORITY_HIGHEST );
+ x264_sps_write( &h->out.bs, h->sps );
+ if( x264_nal_end( h ) )
return -1;
+ overhead += h->out.nal[h->out.i_nal-1].i_payload + NALU_OVERHEAD;
+
+ /* generate picture parameters */
+ x264_nal_start( h, NAL_PPS, NAL_PRIORITY_HIGHEST );
+ x264_pps_write( &h->out.bs, h->pps );
if( x264_nal_end( h ) )
return -1;
overhead += h->out.nal[h->out.i_nal-1].i_payload + NALU_OVERHEAD;
}
- /* generate sequence parameters */
- x264_nal_start( h, NAL_SPS, NAL_PRIORITY_HIGHEST );
- x264_sps_write( &h->out.bs, h->sps );
- if( x264_nal_end( h ) )
- return -1;
- overhead += h->out.nal[h->out.i_nal-1].i_payload + NALU_OVERHEAD;
-
- /* generate picture parameters */
- x264_nal_start( h, NAL_PPS, NAL_PRIORITY_HIGHEST );
- x264_pps_write( &h->out.bs, h->pps );
- if( x264_nal_end( h ) )
- return -1;
- overhead += h->out.nal[h->out.i_nal-1].i_payload + NALU_OVERHEAD;
+ if( h->fenc->i_type != X264_TYPE_IDR )
+ {
+ int time_to_recovery = X264_MIN( h->sps->i_mb_width - 1, h->param.i_keyint_max ) + h->param.i_bframe;
+ x264_nal_start( h, NAL_SEI, NAL_PRIORITY_DISPOSABLE );
+ x264_sei_recovery_point_write( h, &h->out.bs, time_to_recovery );
+ x264_nal_end( h );
+ overhead += h->out.nal[h->out.i_nal-1].i_payload + NALU_OVERHEAD;
+ }
}
/* Init the rate control */
x264_reference_check_reorder( h );
}
+ if( h->i_ref0 )
+ h->fdec->i_poc_l0ref0 = h->fref0[0]->i_poc;
+
if( h->sh.i_type == SLICE_TYPE_B )
x264_macroblock_bipred_init( h );
+ /*------------------------- Weights -------------------------------------*/
+ x264_weighted_pred_init( h );
+
/* ------------------------ Create slice header ----------------------- */
x264_slice_init( h, i_nal_type, i_global_qp );
h->i_frame_num++;
/* Write frame */
- if( h->param.i_threads > 1 )
+ h->i_threadslice_start = 0;
+ h->i_threadslice_end = h->sps->i_mb_height;
+ if( h->i_thread_frames > 1 )
{
if( x264_pthread_create( &h->thread_handle, NULL, (void*)x264_slices_write, h ) )
return -1;
h->b_thread_active = 1;
}
+ else if( h->param.b_sliced_threads )
+ {
+ if( x264_threaded_slices_write( h ) )
+ return -1;
+ }
else
if( (intptr_t)x264_slices_write( h ) )
return -1;
{
void *ret = NULL;
x264_pthread_join( h->thread_handle, &ret );
+ h->b_thread_active = 0;
if( (intptr_t)ret )
return (intptr_t)ret;
- h->b_thread_active = 0;
}
if( !h->out.i_nal )
{
pic_out->i_type = X264_TYPE_P;
else
pic_out->i_type = X264_TYPE_B;
- pic_out->i_pts = h->fenc->i_pts;
+
+ pic_out->b_keyframe = h->fenc->b_keyframe;
+
+ pic_out->i_pts = h->fenc->i_pts *= h->i_dts_compress_multiplier;
+ if( h->frames.i_bframe_delay )
+ {
+ int64_t *i_prev_dts = thread_current->frames.i_prev_dts;
+ if( h->i_frame <= h->frames.i_bframe_delay )
+ {
+ if( h->i_dts_compress_multiplier == 1 )
+ pic_out->i_dts = h->fenc->i_reordered_pts - h->frames.i_bframe_delay_time;
+ else
+ {
+ /* DTS compression */
+ if( h->i_frame == 1 )
+ thread_current->frames.i_init_delta = h->fenc->i_reordered_pts * h->i_dts_compress_multiplier;
+ pic_out->i_dts = h->i_frame * thread_current->frames.i_init_delta / h->i_dts_compress_multiplier;
+ }
+ }
+ else
+ pic_out->i_dts = i_prev_dts[ (h->i_frame - h->frames.i_bframe_delay) % h->frames.i_bframe_delay ];
+ i_prev_dts[ h->i_frame % h->frames.i_bframe_delay ] = h->fenc->i_reordered_pts * h->i_dts_compress_multiplier;
+ }
+ else
+ pic_out->i_dts = h->fenc->i_reordered_pts;
+ assert( pic_out->i_pts >= pic_out->i_dts );
pic_out->img.i_plane = h->fdec->i_plane;
for(i = 0; i < 3; i++)
for( i = 0; i < 32; i++ )
h->stat.i_mb_count_ref[h->sh.i_type][i_list][i] += h->stat.frame.i_mb_count_ref[i_list][i];
if( h->sh.i_type == SLICE_TYPE_P )
+ {
h->stat.i_consecutive_bframes[h->fdec->i_frame - h->fref0[0]->i_frame - 1]++;
+ if( h->param.analyse.i_weighted_pred == X264_WEIGHTP_SMART )
+ {
+ for( i = 0; i < 3; i++ )
+ for( j = 0; j < h->i_ref0; j++ )
+ if( h->sh.weight[0][i].i_denom != 0 )
+ {
+ h->stat.i_wpred[i]++;
+ break;
+ }
+ }
+ }
if( h->sh.i_type == SLICE_TYPE_B )
{
h->stat.i_direct_frames[ h->sh.b_direct_spatial_mv_pred ] ++;
}
#endif
+ /* Remove duplicates, must be done near the end as breaks h->fref0 array
+ * by freeing some of its pointers. */
+ for( i = 0; i < h->i_ref0; i++ )
+ if( h->fref0[i] && h->fref0[i]->b_duplicate )
+ {
+ x264_frame_push_blank_unused( h, h->fref0[i] );
+ h->fref0[i] = 0;
+ }
+
if( h->param.psz_dump_yuv )
x264_frame_dump( h );
x264_lookahead_delete( h );
- for( i=0; i<h->param.i_threads; i++ )
+ if( h->param.i_threads > 1 )
{
// don't strictly have to wait for the other threads, but it's simpler than canceling them
- if( h->thread[i]->b_thread_active )
+ for( i = 0; i < h->param.i_threads; i++ )
+ if( h->thread[i]->b_thread_active )
+ x264_pthread_join( h->thread[i]->thread_handle, NULL );
+ if( h->i_thread_frames > 1 )
{
- x264_pthread_join( h->thread[i]->thread_handle, NULL );
- assert( h->thread[i]->fenc->i_reference_count == 1 );
- x264_frame_delete( h->thread[i]->fenc );
- }
- }
-
- if( h->param.i_threads > 1 )
- {
- x264_t *thread_prev;
+ for( i = 0; i < h->i_thread_frames; i++ )
+ {
+ if( h->thread[i]->b_thread_active )
+ {
+ assert( h->thread[i]->fenc->i_reference_count == 1 );
+ x264_frame_delete( h->thread[i]->fenc );
+ }
+ }
- thread_prev = h->thread[h->i_thread_phase];
- x264_thread_sync_ratecontrol( h, thread_prev, h );
- x264_thread_sync_ratecontrol( thread_prev, thread_prev, h );
- h->i_frame = thread_prev->i_frame + 1 - h->param.i_threads;
+ x264_t *thread_prev = h->thread[h->i_thread_phase];
+ x264_thread_sync_ratecontrol( h, thread_prev, h );
+ x264_thread_sync_ratecontrol( thread_prev, thread_prev, h );
+ h->i_frame = thread_prev->i_frame + 1 - h->i_thread_frames;
+ }
}
h->i_frame++;
x264_log( h, X264_LOG_INFO, "8x8 transform intra:%.1f%%%s\n", 100. * i_i8x8 / i_intra, buf );
}
- if( h->param.analyse.i_direct_mv_pred == X264_DIRECT_PRED_AUTO
+ if( (h->param.analyse.i_direct_mv_pred == X264_DIRECT_PRED_AUTO ||
+ (h->stat.i_direct_frames[0] && h->stat.i_direct_frames[1]))
&& h->stat.i_frame_count[SLICE_TYPE_B] )
{
x264_log( h, X264_LOG_INFO, "direct mvs spatial:%.1f%% temporal:%.1f%%\n",
fixed_pred_modes[i][8] * 100.0 / sum_pred_modes[i] );
}
+ if( h->param.analyse.i_weighted_pred == X264_WEIGHTP_SMART && h->stat.i_frame_count[SLICE_TYPE_P] > 0 )
+ x264_log( h, X264_LOG_INFO, "Weighted P-Frames: Y:%.1f%%\n",
+ h->stat.i_wpred[0] * 100.0 / h->stat.i_frame_count[SLICE_TYPE_P] );
+
for( i_list = 0; i_list < 2; i_list++ )
{
int i_slice;
x264_free( h->nal_buffer );
x264_analyse_free_costs( h );
- if( h->param.i_threads > 1)
+ if( h->i_thread_frames > 1)
h = h->thread[h->i_thread_phase];
/* frames */
x264_frame_delete_list( h->frames.unused[0] );
x264_frame_delete_list( h->frames.unused[1] );
x264_frame_delete_list( h->frames.current );
+ x264_frame_delete_list( h->frames.blank_unused );
h = h->thread[0];
{
x264_frame_t **frame;
- for( frame = h->thread[i]->frames.reference; *frame; frame++ )
+ if( !h->param.b_sliced_threads || i == 0 )
{
+ for( frame = h->thread[i]->frames.reference; *frame; frame++ )
+ {
+ assert( (*frame)->i_reference_count > 0 );
+ (*frame)->i_reference_count--;
+ if( (*frame)->i_reference_count == 0 )
+ x264_frame_delete( *frame );
+ }
+ frame = &h->thread[i]->fdec;
assert( (*frame)->i_reference_count > 0 );
(*frame)->i_reference_count--;
if( (*frame)->i_reference_count == 0 )
x264_frame_delete( *frame );
+ x264_macroblock_cache_end( h->thread[i] );
}
- frame = &h->thread[i]->fdec;
- assert( (*frame)->i_reference_count > 0 );
- (*frame)->i_reference_count--;
- if( (*frame)->i_reference_count == 0 )
- x264_frame_delete( *frame );
-
- x264_macroblock_cache_end( h->thread[i] );
+ x264_free( h->thread[i]->scratch_buffer );
x264_free( h->thread[i]->out.p_bitstream );
x264_free( h->thread[i]->out.nal);
x264_free( h->thread[i] );
{
int delayed_frames = 0;
int i;
- for( i=0; i<h->param.i_threads; i++ )
- delayed_frames += h->thread[i]->b_thread_active;
- h = h->thread[h->i_thread_phase];
+ if( h->i_thread_frames > 1 )
+ {
+ for( i=0; i<h->i_thread_frames; i++ )
+ delayed_frames += h->thread[i]->b_thread_active;
+ h = h->thread[h->i_thread_phase];
+ }
for( i=0; h->frames.current[i]; i++ )
delayed_frames++;
x264_pthread_mutex_lock( &h->lookahead->ofbuf.mutex );