* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111, USA.
*****************************************************************************/
-#include <stdlib.h>
-#include <stdio.h>
-#include <string.h>
#include <math.h>
-#ifdef __WIN32__
-#include <windows.h>
-#define pthread_t HANDLE
-#define pthread_create(t,u,f,d) *(t)=CreateThread(NULL,0,f,d,0,NULL)
-#define pthread_join(t,s) { WaitForSingleObject(t,INFINITE); \
- CloseHandle(t); }
-#define HAVE_PTHREAD 1
-
-#elif defined(SYS_BEOS)
-#include <kernel/OS.h>
-#define pthread_t thread_id
-#define pthread_create(t,u,f,d) { *(t)=spawn_thread(f,"",10,d); \
- resume_thread(*(t)); }
-#define pthread_join(t,s) wait_for_thread(t,(long*)s)
-#define HAVE_PTHREAD 1
-
-#elif HAVE_PTHREAD
-#include <pthread.h>
-#endif
-
#include "common/common.h"
#include "common/cpu.h"
#endif
//#define DEBUG_MB_TYPE
-//#define DEBUG_DUMP_FRAME
-//#define DEBUG_BENCHMARK
-
-#ifdef DEBUG_BENCHMARK
-static int64_t i_mtime_encode_frame = 0;
-static int64_t i_mtime_analyse = 0;
-static int64_t i_mtime_encode = 0;
-static int64_t i_mtime_write = 0;
-static int64_t i_mtime_filter = 0;
-#define TIMER_START( d ) \
- { \
- int64_t d##start = x264_mdate();
-
-#define TIMER_STOP( d ) \
- d += x264_mdate() - d##start;\
- }
-#else
-#define TIMER_START( d )
-#define TIMER_STOP( d )
-#endif
#define NALU_OVERHEAD 5 // startcode + NAL type costs 5 bytes per frame
+static void x264_encoder_frame_end( x264_t *h, x264_t *thread_current,
+ x264_nal_t **pp_nal, int *pi_nal,
+ x264_picture_t *pic_out );
+
/****************************************************************************
*
******************************* x264 libs **********************************
return (float)(-10.0 * log( f_mse ) / log( 10.0 ));
}
-#ifdef DEBUG_DUMP_FRAME
-static void x264_frame_dump( x264_t *h, x264_frame_t *fr, char *name )
+static void x264_frame_dump( x264_t *h )
{
- FILE * f = fopen( name, "a" );
+ FILE *f = fopen( h->param.psz_dump_yuv, "r+b" );
int i, y;
-
- fseek( f, 0, SEEK_END );
-
- for( i = 0; i < fr->i_plane; i++ )
- {
- for( y = 0; y < h->param.i_height / ( i == 0 ? 1 : 2 ); y++ )
- {
- fwrite( &fr->plane[i][y*fr->i_stride[i]], 1, h->param.i_width / ( i == 0 ? 1 : 2 ), f );
- }
- }
+ 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 );
+ 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 );
fclose( f );
}
-#endif
/* Fill "default" values */
static void x264_slice_header_init( x264_t *h, x264_slice_header_t *sh,
x264_sps_t *sps, x264_pps_t *pps,
- int i_type, int i_idr_pic_id, int i_frame, int i_qp )
+ int i_idr_pic_id, int i_frame, int i_qp )
{
x264_param_t *param = &h->param;
int i;
sh->sps = sps;
sh->pps = pps;
- sh->i_type = i_type;
sh->i_first_mb = 0;
sh->i_last_mb = h->sps->i_mb_width * h->sps->i_mb_height;
sh->i_pps_id = pps->i_id;
sh->i_frame_num = i_frame;
- sh->b_field_pic = 0; /* Not field support for now */
- sh->b_bottom_field = 1; /* not yet used */
+ sh->b_mbaff = h->param.b_interlaced;
+ sh->b_field_pic = 0; /* no field support for now */
+ sh->b_bottom_field = 0; /* not yet used */
sh->i_idr_pic_id = i_idr_pic_id;
sh->i_redundant_pic_cnt = 0;
- sh->b_direct_spatial_mv_pred = ( param->analyse.i_direct_mv_pred == X264_DIRECT_PRED_SPATIAL );
+ if( !h->mb.b_direct_auto_read )
+ {
+ 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 b_direct_spatial_mv_pred was read from the 2pass statsfile */
sh->b_num_ref_idx_override = 0;
sh->i_num_ref_idx_l0_active = 1;
/* If effective qp <= 15, deblocking would have no effect anyway */
if( param->b_deblocking_filter
&& ( h->mb.b_variable_qp
- || 15 < i_qp + X264_MAX(param->i_deblocking_filter_alphac0, param->i_deblocking_filter_beta) ) )
+ || 15 < i_qp + 2 * X264_MAX(param->i_deblocking_filter_alphac0, param->i_deblocking_filter_beta) ) )
{
sh->i_disable_deblocking_filter_idc = 0;
}
{
int i;
- bs_write_ue( s, sh->i_first_mb );
+ if( sh->b_mbaff )
+ {
+ assert( sh->i_first_mb % (2*sh->sps->i_mb_width) == 0 );
+ bs_write_ue( s, sh->i_first_mb >> 1 );
+ }
+ else
+ bs_write_ue( s, sh->i_first_mb );
+
bs_write_ue( s, sh->i_type + 5 ); /* same type things */
bs_write_ue( s, sh->i_pps_id );
bs_write( s, sh->sps->i_log2_max_frame_num, sh->i_frame_num );
+ if( !sh->sps->b_frame_mbs_only )
+ {
+ bs_write1( s, sh->b_field_pic );
+ if ( sh->b_field_pic )
+ bs_write1( s, sh->b_bottom_field );
+ }
+
if( sh->i_idr_pic_id >= 0 ) /* NAL IDR */
{
bs_write_ue( s, sh->i_idr_pic_id );
}
}
+/* If we are within a reasonable distance of the end of the memory allocated for the bitstream, */
+/* reallocate, adding an arbitrary amount of space (100 kilobytes). */
+static void x264_bitstream_check_buffer( x264_t *h )
+{
+ if( ( h->param.b_cabac && (h->cabac.p_end - h->cabac.p < 2500) )
+ || ( h->out.bs.p_end - h->out.bs.p < 2500 ) )
+ {
+ uint8_t *bs_bak = h->out.p_bitstream;
+ intptr_t delta;
+ int i;
+
+ h->out.i_bitstream += 100000;
+ h->out.p_bitstream = x264_realloc( h->out.p_bitstream, h->out.i_bitstream );
+ delta = h->out.p_bitstream - bs_bak;
+
+ h->out.bs.p_start += delta;
+ h->out.bs.p += delta;
+ h->out.bs.p_end = h->out.p_bitstream + h->out.i_bitstream;
+
+ h->cabac.p_start += delta;
+ h->cabac.p += delta;
+ h->cabac.p_end = h->out.p_bitstream + h->out.i_bitstream;
+
+ for( i = 0; i <= h->out.i_nal; i++ )
+ h->out.nal[i].p_payload += delta;
+ }
+}
+
/****************************************************************************
*
****************************************************************************
static int x264_validate_parameters( x264_t *h )
{
+#ifdef HAVE_MMX
+ if( !(x264_cpu_detect() & X264_CPU_MMXEXT) )
+ {
+ x264_log( h, X264_LOG_ERROR, "your cpu does not support MMXEXT, but x264 was compiled with asm support\n");
+ x264_log( h, X264_LOG_ERROR, "to run x264, recompile without asm support (configure --disable-asm)\n");
+ return -1;
+ }
+#endif
if( h->param.i_width <= 0 || h->param.i_height <= 0 )
{
x264_log( h, X264_LOG_ERROR, "invalid width x height (%dx%d)\n",
return -1;
}
- h->param.i_threads = x264_clip3( h->param.i_threads, 1, X264_SLICE_MAX );
- h->param.i_threads = X264_MIN( h->param.i_threads, (h->param.i_height + 15) / 16 );
-#if !(HAVE_PTHREAD)
+ if( h->param.i_threads == 0 )
+ h->param.i_threads = x264_cpu_num_processors() * 3/2;
+ h->param.i_threads = x264_clip3( h->param.i_threads, 1, X264_THREAD_MAX );
if( h->param.i_threads > 1 )
{
+#ifndef HAVE_PTHREAD
x264_log( h, X264_LOG_WARNING, "not compiled with pthread support!\n");
- x264_log( h, X264_LOG_WARNING, "multislicing anyway, but you won't see any speed gain.\n" );
- }
+ h->param.i_threads = 1;
+#else
+ if( h->param.i_scenecut_threshold >= 0 )
+ h->param.b_pre_scenecut = 1;
#endif
+ }
+
+ if( h->param.b_interlaced )
+ {
+ if( h->param.analyse.i_me_method >= X264_ME_ESA )
+ {
+ 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 )
+ {
+ x264_log( h, X264_LOG_WARNING, "interlace + direct=temporal is not implemented\n" );
+ h->param.analyse.i_direct_mv_pred = X264_DIRECT_PRED_SPATIAL;
+ }
+ }
- if( h->param.rc.i_rf_constant > 0 )
- h->param.rc.i_qp_constant = h->param.rc.i_rf_constant;
- h->param.rc.i_rf_constant = x264_clip3( h->param.rc.i_rf_constant, 0, 51 );
+ if( h->param.rc.i_rc_method < 0 || h->param.rc.i_rc_method > 2 )
+ {
+ x264_log( h, X264_LOG_ERROR, "no ratecontrol method specified\n" );
+ return -1;
+ }
+ h->param.rc.f_rf_constant = x264_clip3f( h->param.rc.f_rf_constant, 0, 51 );
h->param.rc.i_qp_constant = x264_clip3( h->param.rc.i_qp_constant, 0, 51 );
- if( !h->param.rc.b_cbr && h->param.rc.i_qp_constant == 0 )
+ if( h->param.rc.i_rc_method == X264_RC_CRF )
+ h->param.rc.i_qp_constant = h->param.rc.f_rf_constant;
+ if( (h->param.rc.i_rc_method == X264_RC_CQP || h->param.rc.i_rc_method == X264_RC_CRF)
+ && h->param.rc.i_qp_constant == 0 )
{
h->mb.b_lossless = 1;
- h->param.analyse.b_transform_8x8 = 0;
h->param.i_cqm_preset = X264_CQM_FLAT;
h->param.psz_cqm_file = NULL;
+ h->param.rc.i_rc_method = X264_RC_CQP;
h->param.rc.f_ip_factor = 1;
h->param.rc.f_pb_factor = 1;
+ h->param.analyse.b_transform_8x8 = 0;
h->param.analyse.b_psnr = 0;
+ h->param.analyse.b_ssim = 0;
+ h->param.analyse.i_chroma_qp_offset = 0;
+ h->param.analyse.i_trellis = 0;
+ h->param.analyse.b_fast_pskip = 0;
+ h->param.analyse.i_noise_reduction = 0;
+ h->param.analyse.i_subpel_refine = x264_clip3( h->param.analyse.i_subpel_refine, 1, 6 );
+ }
+ if( h->param.rc.i_rc_method == X264_RC_CQP )
+ {
+ float qp_p = h->param.rc.i_qp_constant;
+ float qp_i = qp_p - 6*log(h->param.rc.f_ip_factor)/log(2);
+ float qp_b = qp_p + 6*log(h->param.rc.f_pb_factor)/log(2);
+ h->param.rc.i_qp_min = x264_clip3( (int)(X264_MIN3( qp_p, qp_i, qp_b )), 0, 51 );
+ h->param.rc.i_qp_max = x264_clip3( (int)(X264_MAX3( qp_p, qp_i, qp_b ) + .999), 0, 51 );
+ h->param.rc.i_aq_mode = 0;
}
- if( ( h->param.i_width % 16 || h->param.i_height % 16 ) && !h->mb.b_lossless )
+ if( ( h->param.i_width % 16 || h->param.i_height % 16 )
+ && h->param.i_height != 1080 && !h->mb.b_lossless )
{
+ // 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 );
h->param.i_bframe_bias = x264_clip3( h->param.i_bframe_bias, -90, 100 );
h->param.b_bframe_pyramid = h->param.b_bframe_pyramid && h->param.i_bframe > 1;
h->param.b_bframe_adaptive = h->param.b_bframe_adaptive && h->param.i_bframe > 0;
+ h->param.analyse.b_weighted_bipred = h->param.analyse.b_weighted_bipred && h->param.i_bframe > 0;
+ 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->param.i_scenecut_threshold < 0 )
+ h->param.b_pre_scenecut = 0;
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 );
+ h->param.analyse.i_luma_deadzone[1] = x264_clip3( h->param.analyse.i_luma_deadzone[1], 0, 32 );
- h->param.i_cabac_init_idc = x264_clip3( h->param.i_cabac_init_idc, -1, 2 );
- /* don't yet support merging of cabac stats */
- if( h->param.i_threads > 1 && h->param.i_cabac_init_idc == -1 )
- h->param.i_cabac_init_idc = 0;
+ h->param.i_cabac_init_idc = x264_clip3( h->param.i_cabac_init_idc, 0, 2 );
if( h->param.i_cqm_preset < X264_CQM_FLAT || h->param.i_cqm_preset > X264_CQM_CUSTOM )
h->param.i_cqm_preset = X264_CQM_FLAT;
if( h->param.analyse.i_me_method < X264_ME_DIA ||
- h->param.analyse.i_me_method > X264_ME_ESA )
+ h->param.analyse.i_me_method > X264_ME_TESA )
h->param.analyse.i_me_method = X264_ME_HEX;
if( h->param.analyse.i_me_range < 4 )
h->param.analyse.i_me_range = 4;
if( h->param.analyse.i_me_range > 16 && h->param.analyse.i_me_method <= X264_ME_HEX )
h->param.analyse.i_me_range = 16;
- h->param.analyse.i_subpel_refine = x264_clip3( h->param.analyse.i_subpel_refine, 1, 6 );
+ if( h->param.analyse.i_me_method == X264_ME_TESA &&
+ (h->mb.b_lossless || h->param.analyse.i_subpel_refine <= 1) )
+ h->param.analyse.i_me_method = X264_ME_ESA;
+ h->param.analyse.i_subpel_refine = x264_clip3( h->param.analyse.i_subpel_refine, 1, 7 );
+ h->param.analyse.b_bframe_rdo = h->param.analyse.b_bframe_rdo && h->param.analyse.i_subpel_refine >= 6;
+ h->param.analyse.b_mixed_references = h->param.analyse.b_mixed_references && h->param.i_frame_reference > 1;
+ h->param.analyse.inter &= X264_ANALYSE_PSUB16x16|X264_ANALYSE_PSUB8x8|X264_ANALYSE_BSUB16x16|
+ X264_ANALYSE_I4x4|X264_ANALYSE_I8x8;
+ h->param.analyse.intra &= X264_ANALYSE_I4x4|X264_ANALYSE_I8x8;
if( !(h->param.analyse.inter & X264_ANALYSE_PSUB16x16) )
h->param.analyse.inter &= ~X264_ANALYSE_PSUB8x8;
if( !h->param.analyse.b_transform_8x8 )
h->param.analyse.intra &= ~X264_ANALYSE_I8x8;
}
h->param.analyse.i_chroma_qp_offset = x264_clip3(h->param.analyse.i_chroma_qp_offset, -12, 12);
- h->param.analyse.i_mv_range = x264_clip3(h->param.analyse.i_mv_range, 32, 2048);
+ if( !h->param.b_cabac )
+ h->param.analyse.i_trellis = 0;
+ h->param.analyse.i_trellis = x264_clip3( h->param.analyse.i_trellis, 0, 2 );
+ h->param.rc.i_aq_mode = x264_clip3( h->param.rc.i_aq_mode, 0, 2 );
+ if( h->param.rc.f_aq_strength <= 0 )
+ h->param.rc.i_aq_mode = 0;
+ /* VAQ effectively replaces qcomp, so qcomp is raised towards 1 to compensate. */
+ if( h->param.rc.i_aq_mode == X264_AQ_GLOBAL )
+ h->param.rc.f_qcompress = x264_clip3f(h->param.rc.f_qcompress + h->param.rc.f_aq_strength / 0.7, 0, 1);
+ h->param.analyse.i_noise_reduction = x264_clip3( h->param.analyse.i_noise_reduction, 0, 1<<16 );
+
+ {
+ const x264_level_t *l = x264_levels;
+ while( l->level_idc != 0 && l->level_idc != h->param.i_level_idc )
+ l++;
+ if( l->level_idc == 0 )
+ {
+ x264_log( h, X264_LOG_ERROR, "invalid level_idc: %d\n", h->param.i_level_idc );
+ return -1;
+ }
+ if( h->param.analyse.i_mv_range <= 0 )
+ h->param.analyse.i_mv_range = l->mv_range;
+ else
+ h->param.analyse.i_mv_range = x264_clip3(h->param.analyse.i_mv_range, 32, 512);
+ if( h->param.analyse.i_direct_8x8_inference < 0 )
+ h->param.analyse.i_direct_8x8_inference = l->direct8x8;
+ }
+
+ if( h->param.i_threads > 1 )
+ {
+ int r = h->param.analyse.i_mv_range_thread;
+ int r2;
+ if( r <= 0 )
+ {
+ // half of the available space is reserved and divided evenly among the threads,
+ // 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;
+ r = max_range / 2;
+ }
+ r = X264_MAX( r, h->param.analyse.i_me_range );
+ r = X264_MIN( r, h->param.analyse.i_mv_range );
+ // round up to use the whole mb row
+ r2 = (r & ~15) + ((-X264_THREAD_HEIGHT) & 15);
+ if( r2 < r )
+ r2 += 16;
+ x264_log( h, X264_LOG_DEBUG, "using mv_range_thread = %d\n", r2 );
+ h->param.analyse.i_mv_range_thread = r2;
+ }
if( h->param.rc.f_qblur < 0 )
h->param.rc.f_qblur = 0;
if( h->param.rc.f_complexity_blur < 0 )
h->param.rc.f_complexity_blur = 0;
+ h->param.i_sps_id &= 31;
+
+ if( h->param.i_log_level < X264_LOG_INFO )
+ {
+ h->param.analyse.b_psnr = 0;
+ h->param.analyse.b_ssim = 0;
+ }
+
+ /* 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_deblocking_filter );
+ BOOLIFY( b_interlaced );
+ BOOLIFY( analyse.b_transform_8x8 );
+ BOOLIFY( analyse.i_direct_8x8_inference );
+ BOOLIFY( analyse.b_bidir_me );
+ BOOLIFY( analyse.b_chroma_me );
+ BOOLIFY( analyse.b_fast_pskip );
+ BOOLIFY( rc.b_stat_write );
+ BOOLIFY( rc.b_stat_read );
+#undef BOOLIFY
+
return 0;
}
+static void mbcmp_init( x264_t *h )
+{
+ int satd = !h->mb.b_lossless && h->param.analyse.i_subpel_refine > 1;
+ memcpy( h->pixf.mbcmp, satd ? h->pixf.satd : h->pixf.sad, sizeof(h->pixf.mbcmp) );
+ satd &= h->param.analyse.i_me_method == X264_ME_TESA;
+ memcpy( h->pixf.fpelcmp, satd ? h->pixf.satd : h->pixf.sad, sizeof(h->pixf.fpelcmp) );
+ memcpy( h->pixf.fpelcmp_x3, satd ? h->pixf.satd_x3 : h->pixf.sad_x3, sizeof(h->pixf.fpelcmp_x3) );
+ memcpy( h->pixf.fpelcmp_x4, satd ? h->pixf.satd_x4 : h->pixf.sad_x4, sizeof(h->pixf.fpelcmp_x4) );
+}
+
/****************************************************************************
* x264_encoder_open:
****************************************************************************/
x264_t *x264_encoder_open ( x264_param_t *param )
{
x264_t *h = x264_malloc( sizeof( x264_t ) );
+ char buf[1000], *p;
int i;
memset( h, 0, sizeof( x264_t ) );
{
int i_w = param->vui.i_sar_width;
int i_h = param->vui.i_sar_height;
- int a = i_w, b = i_h;
- while( b != 0 )
- {
- int t = a;
+ x264_reduce_fraction( &i_w, &i_h );
- a = b;
- b = t % b;
- }
-
- i_w /= a;
- i_h /= a;
while( i_w > 65535 || i_h > 65535 )
{
i_w /= 2;
h->param.vui.i_sar_height = 0;
if( i_w == 0 || i_h == 0 )
{
- x264_log( h, X264_LOG_ERROR, "cannot create valid sample aspect ratio\n" );
- }
- else if( i_w == i_h )
- {
- x264_log( h, X264_LOG_INFO, "no need for a SAR\n" );
+ x264_log( h, X264_LOG_WARNING, "cannot create valid sample aspect ratio\n" );
}
else
{
}
}
+ x264_reduce_fraction( &h->param.i_fps_num, &h->param.i_fps_den );
/* Init x264_t */
- h->out.i_nal = 0;
- h->out.i_bitstream = X264_MAX( 1000000, h->param.i_width * h->param.i_height * 1.7
- * ( h->param.rc.b_cbr ? pow( 0.5, h->param.rc.i_qp_min )
- : pow( 0.5, h->param.rc.i_qp_constant ) * X264_MAX( 1, h->param.rc.f_ip_factor )));
- h->out.p_bitstream = x264_malloc( h->out.i_bitstream );
-
h->i_frame = 0;
h->i_frame_num = 0;
h->i_idr_pic_id = 0;
h->sps = &h->sps_array[0];
- x264_sps_init( h->sps, 0, &h->param );
+ x264_sps_init( h->sps, h->param.i_sps_id, &h->param );
h->pps = &h->pps_array[0];
- x264_pps_init( h->pps, 0, &h->param, h->sps);
+ x264_pps_init( h->pps, h->param.i_sps_id, &h->param, h->sps);
+
+ x264_validate_levels( h );
- x264_cqm_init( h );
+ if( x264_cqm_init( h ) < 0 )
+ {
+ x264_free( h );
+ return NULL;
+ }
h->mb.i_mb_count = h->sps->i_mb_width * h->sps->i_mb_height;
/* Init frames. */
- h->frames.i_delay = h->param.i_bframe;
+ h->frames.i_delay = h->param.i_bframe + h->param.i_threads - 1;
h->frames.i_max_ref0 = h->param.i_frame_reference;
h->frames.i_max_ref1 = h->sps->vui.i_num_reorder_frames;
- h->frames.i_max_dpb = h->sps->vui.i_max_dec_frame_buffering + 1;
+ h->frames.i_max_dpb = h->sps->vui.i_max_dec_frame_buffering;
h->frames.b_have_lowres = !h->param.rc.b_stat_read
- && ( h->param.rc.b_cbr || h->param.rc.i_rf_constant || h->param.b_bframe_adaptive );
+ && ( h->param.rc.i_rc_method == X264_RC_ABR
+ || h->param.rc.i_rc_method == X264_RC_CRF
+ || h->param.b_bframe_adaptive
+ || h->param.b_pre_scenecut );
+ h->frames.b_have_lowres |= (h->param.rc.b_stat_read && h->param.rc.i_vbv_buffer_size > 0);
- for( i = 0; i < X264_BFRAME_MAX + 3; i++ )
- {
- h->frames.current[i] = NULL;
- h->frames.next[i] = NULL;
- h->frames.unused[i] = NULL;
- }
- for( i = 0; i < 1 + h->frames.i_delay; i++ )
- {
- h->frames.unused[i] = x264_frame_new( h );
- }
- for( i = 0; i < h->frames.i_max_dpb; i++ )
- {
- h->frames.reference[i] = x264_frame_new( h );
- }
- h->frames.reference[h->frames.i_max_dpb] = NULL;
h->frames.i_last_idr = - h->param.i_keyint_max;
h->frames.i_input = 0;
h->frames.last_nonb = NULL;
h->i_ref0 = 0;
h->i_ref1 = 0;
- h->fdec = h->frames.reference[0];
-
- /* init mb cache */
- x264_macroblock_cache_init( h );
-
- /* init cabac adaptive model */
- x264_cabac_model_init( &h->cabac );
+ x264_rdo_init( );
/* init CPU functions */
x264_predict_16x16_init( h->param.cpu, h->predict_16x16 );
x264_pixel_init( h->param.cpu, &h->pixf );
x264_dct_init( h->param.cpu, &h->dctf );
+ x264_zigzag_init( h->param.cpu, &h->zigzagf, h->param.b_interlaced );
x264_mc_init( h->param.cpu, &h->mc );
- x264_csp_init( h->param.cpu, h->param.i_csp, &h->csp );
x264_quant_init( h, h->param.cpu, &h->quantf );
+ x264_deblock_init( h->param.cpu, &h->loopf );
+ x264_dct_init_weights();
- memcpy( h->pixf.mbcmp,
- ( h->mb.b_lossless || h->param.analyse.i_subpel_refine <= 1 ) ? h->pixf.sad : h->pixf.satd,
- sizeof(h->pixf.mbcmp) );
+ mbcmp_init( h );
- /* rate control */
- if( x264_ratecontrol_new( h ) < 0 )
- return NULL;
+ p = buf + sprintf( buf, "using cpu capabilities:" );
+ 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) )
+ continue;
+ if( !strcmp(x264_cpu_names[i].name, "SSE3")
+ && (param->cpu & X264_CPU_SSSE3 || !(param->cpu & X264_CPU_CACHELINE_64)) )
+ continue;
+ if( (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 )
+ p += sprintf( p, " none!" );
+ x264_log( h, X264_LOG_INFO, "%s\n", buf );
- x264_log( h, X264_LOG_INFO, "using cpu capabilities %s%s%s%s%s%s\n",
- param->cpu&X264_CPU_MMX ? "MMX " : "",
- param->cpu&X264_CPU_MMXEXT ? "MMXEXT " : "",
- param->cpu&X264_CPU_SSE ? "SSE " : "",
- param->cpu&X264_CPU_SSE2 ? "SSE2 " : "",
- param->cpu&X264_CPU_3DNOW ? "3DNow! " : "",
- param->cpu&X264_CPU_ALTIVEC ? "Altivec " : "" );
+ h->out.i_nal = 0;
+ h->out.i_bitstream = X264_MAX( 1000000, h->param.i_width * h->param.i_height * 4
+ * ( h->param.rc.i_rc_method == X264_RC_ABR ? pow( 0.95, h->param.rc.i_qp_min )
+ : pow( 0.95, h->param.rc.i_qp_constant ) * X264_MAX( 1, h->param.rc.f_ip_factor )));
h->thread[0] = h;
- for( i = 1; i < param->i_threads; i++ )
+ h->i_thread_num = 0;
+ for( i = 1; i < h->param.i_threads; i++ )
h->thread[i] = x264_malloc( sizeof(x264_t) );
+ for( i = 0; i < h->param.i_threads; i++ )
+ {
+ if( i > 0 )
+ *h->thread[i] = *h;
+ h->thread[i]->fdec = x264_frame_pop_unused( h );
+ h->thread[i]->out.p_bitstream = x264_malloc( h->out.i_bitstream );
+ if( x264_macroblock_cache_init( h->thread[i] ) < 0 )
+ return NULL;
+ }
+
+ if( x264_ratecontrol_new( h ) < 0 )
+ return NULL;
+
+ if( h->param.psz_dump_yuv )
+ {
+ /* create or truncate the reconstructed video file */
+ FILE *f = fopen( h->param.psz_dump_yuv, "w" );
+ if( f )
+ fclose( f );
+ else
+ {
+ x264_log( h, X264_LOG_ERROR, "can't write to fdec.yuv\n" );
+ x264_free( h );
+ return NULL;
+ }
+ }
+
return h;
}
****************************************************************************/
int x264_encoder_reconfig( x264_t *h, x264_param_t *param )
{
- h->param.i_bframe_bias = param->i_bframe_bias;
- h->param.i_deblocking_filter_alphac0 = param->i_deblocking_filter_alphac0;
- h->param.i_deblocking_filter_beta = param->i_deblocking_filter_beta;
- h->param.analyse.i_me_method = param->analyse.i_me_method;
- h->param.analyse.i_me_range = param->analyse.i_me_range;
- h->param.analyse.i_subpel_refine = param->analyse.i_subpel_refine;
- h->param.analyse.intra = param->analyse.intra;
- h->param.analyse.inter = param->analyse.inter;
- if( h->sps->b_direct8x8_inference && h->param.i_bframe
- && h->param.analyse.i_direct_mv_pred == X264_DIRECT_PRED_TEMPORAL )
- h->param.analyse.inter &= ~X264_ANALYSE_PSUB8x8;
-
- memcpy( h->pixf.mbcmp,
- ( h->mb.b_lossless || h->param.analyse.i_subpel_refine <= 1 ) ? h->pixf.sad : h->pixf.satd,
- sizeof(h->pixf.mbcmp) );
+#define COPY(var) h->param.var = param->var
+ COPY( i_frame_reference ); // but never uses more refs than initially specified
+ COPY( i_bframe_bias );
+ if( h->param.i_scenecut_threshold >= 0 && param->i_scenecut_threshold >= 0 )
+ COPY( i_scenecut_threshold ); // can't turn it on or off, only vary the threshold
+ COPY( b_deblocking_filter );
+ COPY( i_deblocking_filter_alphac0 );
+ COPY( i_deblocking_filter_beta );
+ COPY( analyse.intra );
+ COPY( analyse.inter );
+ COPY( analyse.i_direct_mv_pred );
+ COPY( analyse.i_me_method );
+ COPY( analyse.i_me_range );
+ COPY( analyse.i_noise_reduction );
+ COPY( analyse.i_subpel_refine );
+ COPY( analyse.i_trellis );
+ COPY( analyse.b_bidir_me );
+ COPY( analyse.b_bframe_rdo );
+ COPY( analyse.b_chroma_me );
+ COPY( analyse.b_dct_decimate );
+ COPY( analyse.b_fast_pskip );
+ COPY( analyse.b_mixed_references );
+ // can only twiddle these if they were enabled to begin with:
+ if( h->pps->b_transform_8x8_mode )
+ COPY( analyse.b_transform_8x8 );
+ if( h->frames.i_max_ref1 > 1 )
+ COPY( b_bframe_pyramid );
+#undef COPY
+
+ mbcmp_init( h );
return x264_validate_parameters( h );
}
nal->i_ref_idc = i_ref_idc;
nal->i_type = i_type;
- bs_align_0( &h->out.bs ); /* not needed */
-
nal->i_payload= 0;
- nal->p_payload= &h->out.p_bitstream[bs_pos( &h->out.bs) / 8];
+ nal->p_payload= &h->out.p_bitstream[bs_pos( &h->out.bs ) / 8];
}
static void x264_nal_end( x264_t *h )
{
x264_nal_t *nal = &h->out.nal[h->out.i_nal];
-
- bs_align_0( &h->out.bs ); /* not needed */
-
- nal->i_payload = &h->out.p_bitstream[bs_pos( &h->out.bs)/8] - nal->p_payload;
-
+ nal->i_payload = &h->out.p_bitstream[bs_pos( &h->out.bs ) / 8] - nal->p_payload;
h->out.i_nal++;
}
{
/* identify ourself */
x264_nal_start( h, NAL_SEI, NAL_PRIORITY_DISPOSABLE );
- x264_sei_version_write( &h->out.bs );
+ x264_sei_version_write( h, &h->out.bs );
x264_nal_end( h );
/* generate sequence parameters */
/* now set output*/
*pi_nal = h->out.i_nal;
*pp_nal = &h->out.nal[0];
+ h->out.i_nal = 0;
return 0;
}
-
-static void x264_frame_put( x264_frame_t *list[X264_BFRAME_MAX], x264_frame_t *frame )
-{
- int i = 0;
- while( list[i] ) i++;
- list[i] = frame;
-}
-
-static void x264_frame_push( x264_frame_t *list[X264_BFRAME_MAX], x264_frame_t *frame )
-{
- int i = 0;
- while( list[i] ) i++;
- while( i-- )
- list[i+1] = list[i];
- list[0] = frame;
-}
-
-static x264_frame_t *x264_frame_get( x264_frame_t *list[X264_BFRAME_MAX+1] )
-{
- x264_frame_t *frame = list[0];
- int i;
- for( i = 0; list[i]; i++ )
- list[i] = list[i+1];
- return frame;
-}
-
-static void x264_frame_sort( x264_frame_t *list[X264_BFRAME_MAX+1], int b_dts )
-{
- int i, b_ok;
- do {
- b_ok = 1;
- for( i = 0; list[i+1]; i++ )
- {
- int dtype = list[i]->i_type - list[i+1]->i_type;
- int dtime = list[i]->i_frame - list[i+1]->i_frame;
- int swap = b_dts ? dtype > 0 || ( dtype == 0 && dtime > 0 )
- : dtime > 0;
- if( swap )
- {
- x264_frame_t *tmp = list[i+1];
- list[i+1] = list[i];
- list[i] = tmp;
- b_ok = 0;
- }
- }
- } while( !b_ok );
-}
-#define x264_frame_sort_dts(list) x264_frame_sort(list, 1)
-#define x264_frame_sort_pts(list) x264_frame_sort(list, 0)
-
-static inline void x264_reference_build_list( x264_t *h, int i_poc, int i_slice_type )
+static inline void x264_reference_build_list( x264_t *h, int i_poc )
{
int i;
int b_ok;
/* build ref list 0/1 */
h->i_ref0 = 0;
h->i_ref1 = 0;
- for( i = 1; i < h->frames.i_max_dpb; i++ )
+ for( i = 0; h->frames.reference[i]; i++ )
{
- if( h->frames.reference[i]->i_poc >= 0 )
+ if( h->frames.reference[i]->i_poc < i_poc )
{
- if( h->frames.reference[i]->i_poc < i_poc )
- {
- h->fref0[h->i_ref0++] = h->frames.reference[i];
- }
- else if( h->frames.reference[i]->i_poc > i_poc )
- {
- h->fref1[h->i_ref1++] = h->frames.reference[i];
- }
+ h->fref0[h->i_ref0++] = h->frames.reference[i];
+ }
+ else if( h->frames.reference[i]->i_poc > i_poc )
+ {
+ h->fref1[h->i_ref1++] = h->frames.reference[i];
}
}
{
if( h->fref0[i]->i_poc < h->fref0[i+1]->i_poc )
{
- x264_frame_t *tmp = h->fref0[i+1];
-
- h->fref0[i+1] = h->fref0[i];
- h->fref0[i] = tmp;
+ XCHG( x264_frame_t*, h->fref0[i], h->fref0[i+1] );
b_ok = 0;
break;
}
{
if( h->fref1[i]->i_poc > h->fref1[i+1]->i_poc )
{
- x264_frame_t *tmp = h->fref1[i+1];
-
- h->fref1[i+1] = h->fref1[i];
- h->fref1[i] = tmp;
+ XCHG( x264_frame_t*, h->fref1[i], h->fref1[i+1] );
b_ok = 0;
break;
}
* We use POC, but check whether explicit reordering is needed */
h->b_ref_reorder[0] =
h->b_ref_reorder[1] = 0;
- if( i_slice_type == SLICE_TYPE_P )
+ if( h->sh.i_type == SLICE_TYPE_P )
{
for( i = 0; i < h->i_ref0 - 1; i++ )
if( h->fref0[i]->i_frame_num < h->fref0[i+1]->i_frame_num )
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, 16 - h->i_ref1 );
+ h->i_ref0 = X264_MIN( h->i_ref0, h->param.i_frame_reference ); // if reconfig() has lowered the limit
+ 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;
}
-static inline void x264_reference_update( x264_t *h )
+static void x264_fdec_filter_row( x264_t *h, int mb_y )
{
- int i;
+ /* mb_y is the mb to be encoded next, not the mb to be filtered here */
+ int b_hpel = h->fdec->b_kept_as_ref;
+ int b_deblock = !h->sh.i_disable_deblocking_filter_idc;
+ int b_end = mb_y == h->sps->i_mb_height;
+ int min_y = mb_y - (1 << h->sh.b_mbaff);
+ int max_y = b_end ? h->sps->i_mb_height : mb_y;
+ b_deblock &= b_hpel || h->param.psz_dump_yuv;
+ if( mb_y & h->sh.b_mbaff )
+ return;
+ if( min_y < 0 )
+ return;
+
+ if( !b_end )
+ {
+ int i, j;
+ for( j=0; j<=h->sh.b_mbaff; j++ )
+ for( i=0; i<3; i++ )
+ {
+ memcpy( h->mb.intra_border_backup[j][i],
+ h->fdec->plane[i] + ((mb_y*16 >> !!i) + j - 1 - h->sh.b_mbaff) * h->fdec->i_stride[i],
+ h->sps->i_mb_width*16 >> !!i );
+ }
+ }
+
+ if( b_deblock )
+ {
+ int y;
+ for( y = min_y; y < max_y; y += (1 << h->sh.b_mbaff) )
+ x264_frame_deblock_row( h, y );
+ }
+
+ if( b_hpel )
+ {
+ x264_frame_expand_border( h, h->fdec, min_y, b_end );
+ x264_frame_filter( h, h->fdec, min_y, b_end );
+ x264_frame_expand_border_filtered( h, h->fdec, min_y, b_end );
+ }
- /* apply deblocking filter to the current decoded picture */
- if( !h->sh.i_disable_deblocking_filter_idc )
+ if( h->param.i_threads > 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;
+
+ if( h->param.analyse.b_psnr )
{
- TIMER_START( i_mtime_filter );
- x264_frame_deblocking_filter( h, h->sh.i_type );
- TIMER_STOP( i_mtime_filter );
+ int i;
+ for( i=0; i<3; i++ )
+ h->stat.frame.i_ssd[i] +=
+ x264_pixel_ssd_wxh( &h->pixf,
+ h->fdec->plane[i] + (min_y>>!!i) * h->fdec->i_stride[i], h->fdec->i_stride[i],
+ h->fenc->plane[i] + (min_y>>!!i) * h->fenc->i_stride[i], h->fenc->i_stride[i],
+ h->param.i_width >> !!i, (max_y-min_y) >> !!i );
+ }
+
+ if( h->param.analyse.b_ssim )
+ {
+ x264_emms();
+ /* offset by 2 pixels to avoid alignment of ssim blocks with dct blocks,
+ * and overlap by 4 */
+ min_y += min_y == 0 ? 2 : -6;
+ h->stat.frame.f_ssim +=
+ x264_pixel_ssim_wxh( &h->pixf,
+ h->fdec->plane[0] + 2+min_y*h->fdec->i_stride[0], h->fdec->i_stride[0],
+ h->fenc->plane[0] + 2+min_y*h->fenc->i_stride[0], h->fenc->i_stride[0],
+ h->param.i_width-2, max_y-min_y );
}
- /* expand border */
- x264_frame_expand_border( h->fdec );
+}
+
+static inline void x264_reference_update( x264_t *h )
+{
+ int i;
- /* create filtered images */
- x264_frame_filter( h->param.cpu, h->fdec );
+ if( h->fdec->i_frame >= 0 )
+ h->i_frame++;
- /* expand border of filtered images */
- x264_frame_expand_border_filtered( h->fdec );
+ if( !h->fdec->b_kept_as_ref )
+ {
+ if( h->param.i_threads > 1 )
+ {
+ x264_frame_push_unused( h, h->fdec );
+ h->fdec = x264_frame_pop_unused( h );
+ }
+ return;
+ }
/* move lowres copy of the image to the ref frame */
for( i = 0; i < 4; i++)
{
- uint8_t *tmp = h->fdec->lowres[i];
- h->fdec->lowres[i] = h->fenc->lowres[i];
- h->fenc->lowres[i] = tmp;
+ XCHG( uint8_t*, h->fdec->lowres[i], h->fenc->lowres[i] );
+ XCHG( uint8_t*, h->fdec->buffer_lowres[i], h->fenc->buffer_lowres[i] );
}
/* adaptive B decision needs a pointer, since it can't use the ref lists */
h->frames.last_nonb = h->fdec;
/* move frame in the buffer */
- h->fdec = h->frames.reference[h->frames.i_max_dpb-1];
- for( i = h->frames.i_max_dpb-1; i > 0; i-- )
- {
- h->frames.reference[i] = h->frames.reference[i-1];
- }
- h->frames.reference[0] = h->fdec;
+ x264_frame_push( h->frames.reference, h->fdec );
+ if( h->frames.reference[h->frames.i_max_dpb] )
+ x264_frame_push_unused( h, x264_frame_shift( h->frames.reference ) );
+ h->fdec = x264_frame_pop_unused( h );
}
static inline void x264_reference_reset( x264_t *h )
{
- int i;
-
- /* reset ref pictures */
- for( i = 1; i < h->frames.i_max_dpb; i++ )
- {
- h->frames.reference[i]->i_poc = -1;
- }
- h->frames.reference[0]->i_poc = 0;
+ while( h->frames.reference[0] )
+ x264_frame_push_unused( h, x264_frame_pop( h->frames.reference ) );
+ h->fdec->i_poc =
+ h->fenc->i_poc = 0;
}
-static inline void x264_slice_init( x264_t *h, int i_nal_type, int i_slice_type, int i_global_qp )
+static inline void x264_slice_init( x264_t *h, int i_nal_type, int i_global_qp )
{
/* ------------------------ Create slice header ----------------------- */
if( i_nal_type == NAL_SLICE_IDR )
{
- x264_slice_header_init( h, &h->sh, h->sps, h->pps, i_slice_type, h->i_idr_pic_id, h->i_frame_num - 1, i_global_qp );
+ x264_slice_header_init( h, &h->sh, h->sps, h->pps, h->i_idr_pic_id, h->i_frame_num, i_global_qp );
/* increment id */
h->i_idr_pic_id = ( h->i_idr_pic_id + 1 ) % 65536;
}
else
{
- x264_slice_header_init( h, &h->sh, h->sps, h->pps, i_slice_type, -1, h->i_frame_num - 1, i_global_qp );
+ x264_slice_header_init( h, &h->sh, h->sps, h->pps, -1, h->i_frame_num, i_global_qp );
/* always set the real higher num of ref frame used */
h->sh.b_num_ref_idx_override = 1;
/* Nothing to do ? */
}
- /* get adapative cabac model if needed */
- if( h->param.b_cabac )
- {
- if( h->param.i_cabac_init_idc == -1 )
- {
- h->sh.i_cabac_init_idc = x264_cabac_model_get( &h->cabac, i_slice_type );
- }
- }
-
x264_macroblock_slice_init( h );
}
-static int x264_slice_write( x264_t *h )
+static void x264_slice_write( x264_t *h )
{
int i_skip;
- int mb_xy;
+ int mb_xy, i_mb_x, i_mb_y;
int i;
/* init stats */
/* init cabac */
x264_cabac_context_init( &h->cabac, h->sh.i_type, h->sh.i_qp, h->sh.i_cabac_init_idc );
- x264_cabac_encode_init ( &h->cabac, &h->out.bs );
+ x264_cabac_encode_init ( &h->cabac, h->out.bs.p, h->out.bs.p_end );
}
h->mb.i_last_qp = h->sh.i_qp;
h->mb.i_last_dqp = 0;
- for( mb_xy = h->sh.i_first_mb, i_skip = 0; mb_xy < h->sh.i_last_mb; mb_xy++ )
+ i_mb_y = h->sh.i_first_mb / h->sps->i_mb_width;
+ i_mb_x = h->sh.i_first_mb % h->sps->i_mb_width;
+ i_skip = 0;
+
+ while( (mb_xy = i_mb_x + i_mb_y * h->sps->i_mb_width) < h->sh.i_last_mb )
{
- const int i_mb_y = mb_xy / h->sps->i_mb_width;
- const int i_mb_x = mb_xy % h->sps->i_mb_width;
+ int mb_spos = bs_pos(&h->out.bs) + x264_cabac_pos(&h->cabac);
- int mb_spos = bs_pos(&h->out.bs);
+ if( i_mb_x == 0 )
+ x264_fdec_filter_row( h, i_mb_y );
/* load cache */
x264_macroblock_cache_load( h, i_mb_x, i_mb_y );
* Slice I: choose I_4x4 or I_16x16 mode
* Slice P: choose between using P mode or intra (4x4 or 16x16)
* */
- TIMER_START( i_mtime_analyse );
x264_macroblock_analyse( h );
- TIMER_STOP( i_mtime_analyse );
- /* encode this macrobock -> be carefull it can change the mb type to P_SKIP if needed */
- TIMER_START( i_mtime_encode );
+ /* encode this macroblock -> be careful it can change the mb type to P_SKIP if needed */
x264_macroblock_encode( h );
- TIMER_STOP( i_mtime_encode );
- TIMER_START( i_mtime_write );
+ x264_bitstream_check_buffer( h );
+
if( h->param.b_cabac )
{
- if( mb_xy > h->sh.i_first_mb )
- x264_cabac_encode_terminal( &h->cabac, 0 );
+ if( mb_xy > h->sh.i_first_mb && !(h->sh.b_mbaff && (i_mb_y&1)) )
+ x264_cabac_encode_terminal( &h->cabac );
if( IS_SKIP( h->mb.i_type ) )
x264_cabac_mb_skip( h, 1 );
x264_macroblock_write_cavlc( h, &h->out.bs );
}
}
- TIMER_STOP( i_mtime_write );
#if VISUALIZE
if( h->param.b_visualize )
h->stat.frame.i_mb_count_8x8dct[1] += h->mb.b_transform_8x8;
}
- if( h->mb.b_variable_qp )
- x264_ratecontrol_mb(h, bs_pos(&h->out.bs) - mb_spos);
- }
+ x264_ratecontrol_mb( h, bs_pos(&h->out.bs) + x264_cabac_pos(&h->cabac) - mb_spos );
- if( h->param.b_cabac )
- {
- /* end of slice */
- x264_cabac_encode_terminal( &h->cabac, 1 );
- }
- else if( i_skip > 0 )
- {
- bs_write_ue( &h->out.bs, i_skip ); /* last skip run */
+ if( h->sh.b_mbaff )
+ {
+ i_mb_x += i_mb_y & 1;
+ i_mb_y ^= i_mb_x < h->sps->i_mb_width;
+ }
+ else
+ i_mb_x++;
+ if(i_mb_x == h->sps->i_mb_width)
+ {
+ i_mb_y++;
+ i_mb_x = 0;
+ }
}
if( h->param.b_cabac )
{
- x264_cabac_encode_flush( &h->cabac );
-
+ x264_cabac_encode_flush( h, &h->cabac );
+ h->out.bs.p = h->cabac.p;
}
else
{
+ if( i_skip > 0 )
+ bs_write_ue( &h->out.bs, i_skip ); /* last skip run */
/* rbsp_slice_trailing_bits */
bs_rbsp_trailing( &h->out.bs );
}
x264_nal_end( h );
+ x264_fdec_filter_row( h, h->sps->i_mb_height );
+
/* Compute misc bits */
h->stat.frame.i_misc_bits = bs_pos( &h->out.bs )
+ NALU_OVERHEAD * 8
- h->stat.frame.i_itex_bits
- h->stat.frame.i_ptex_bits
- h->stat.frame.i_hdr_bits;
+}
- return 0;
+static void x264_thread_sync_context( x264_t *dst, x264_t *src )
+{
+ x264_frame_t **f;
+ if( dst == src )
+ return;
+
+ // reference counting
+ for( f = src->frames.reference; *f; f++ )
+ (*f)->i_reference_count++;
+ for( f = dst->frames.reference; *f; f++ )
+ x264_frame_push_unused( src, *f );
+ src->fdec->i_reference_count++;
+ x264_frame_push_unused( src, dst->fdec );
+
+ // copy everything except the per-thread pointers and the constants.
+ memcpy( &dst->i_frame, &src->i_frame, offsetof(x264_t, mb.type) - offsetof(x264_t, i_frame) );
+ memcpy( &dst->mb.i_type, &src->mb.i_type, offsetof(x264_t, rc) - offsetof(x264_t, mb.i_type) );
+ dst->stat = src->stat;
+}
+
+static void x264_thread_sync_stat( x264_t *dst, x264_t *src )
+{
+ if( dst == src )
+ return;
+ memcpy( &dst->stat.i_slice_count, &src->stat.i_slice_count, sizeof(dst->stat) - sizeof(dst->stat.frame) );
}
-static inline int x264_slices_write( x264_t *h )
+static int x264_slices_write( x264_t *h )
{
int i_frame_size;
x264_visualize_init( h );
#endif
- if( h->param.i_threads == 1 )
- {
- x264_slice_write( h );
- i_frame_size = h->out.nal[h->out.i_nal-1].i_payload;
- }
- else
- {
- int i_nal = h->out.i_nal;
- int i_bs_size = h->out.i_bitstream / h->param.i_threads;
- int i;
- /* duplicate contexts */
- for( i = 0; i < h->param.i_threads; i++ )
- {
- x264_t *t = h->thread[i];
- if( i > 0 )
- {
- memcpy( t, h, sizeof(x264_t) );
- t->out.p_bitstream += i*i_bs_size;
- bs_init( &t->out.bs, t->out.p_bitstream, i_bs_size );
- }
- t->sh.i_first_mb = (i * h->sps->i_mb_height / h->param.i_threads) * h->sps->i_mb_width;
- t->sh.i_last_mb = ((i+1) * h->sps->i_mb_height / h->param.i_threads) * h->sps->i_mb_width;
- t->out.i_nal = i_nal + i;
- }
-
- /* dispatch */
-#if HAVE_PTHREAD
- {
- pthread_t handles[X264_SLICE_MAX];
- void *status;
- for( i = 0; i < h->param.i_threads; i++ )
- pthread_create( &handles[i], NULL, (void*)x264_slice_write, (void*)h->thread[i] );
- for( i = 0; i < h->param.i_threads; i++ )
- pthread_join( handles[i], &status );
- }
-#else
- for( i = 0; i < h->param.i_threads; i++ )
- x264_slice_write( h->thread[i] );
-#endif
-
- /* merge contexts */
- i_frame_size = h->out.nal[i_nal].i_payload;
- for( i = 1; i < h->param.i_threads; i++ )
- {
- int j;
- x264_t *t = h->thread[i];
- h->out.nal[i_nal+i] = t->out.nal[i_nal+i];
- i_frame_size += t->out.nal[i_nal+i].i_payload;
- // all entries in stat.frame are ints
- for( j = 0; j < sizeof(h->stat.frame) / sizeof(int); j++ )
- ((int*)&h->stat.frame)[j] += ((int*)&t->stat.frame)[j];
- }
- h->out.i_nal = i_nal + h->param.i_threads;
- }
+ x264_stack_align( x264_slice_write, h );
+ i_frame_size = h->out.nal[h->out.i_nal-1].i_payload;
#if VISUALIZE
if( h->param.b_visualize )
}
#endif
- return i_frame_size;
+ h->out.i_frame_size = i_frame_size;
+ return 0;
}
/****************************************************************************
x264_picture_t *pic_in,
x264_picture_t *pic_out )
{
- x264_frame_t *frame_psnr = h->fdec; /* just to keep the current decoded frame for psnr calculation */
+ x264_t *thread_current, *thread_prev, *thread_oldest;
int i_nal_type;
int i_nal_ref_idc;
- int i_slice_type;
- int i_frame_size;
-
- int i;
int i_global_qp;
- char psz_message[80];
+ if( h->param.i_threads > 1)
+ {
+ int i = ++h->i_thread_phase;
+ int t = h->param.i_threads;
+ thread_current = h->thread[ i%t ];
+ thread_prev = h->thread[ (i-1)%t ];
+ thread_oldest = h->thread[ (i+1)%t ];
+ x264_thread_sync_context( thread_current, thread_prev );
+ x264_thread_sync_ratecontrol( thread_current, thread_prev, thread_oldest );
+ h = thread_current;
+// fprintf(stderr, "current: %p prev: %p oldest: %p \n", thread_current, thread_prev, thread_oldest);
+ }
+ else
+ {
+ thread_current =
+ thread_prev =
+ thread_oldest = h;
+ }
+
+ // ok to call this before encoding any frames, since the initial values of fdec have b_kept_as_ref=0
+ x264_reference_update( h );
+ h->fdec->i_lines_completed = -1;
/* no data out */
*pi_nal = 0;
*pp_nal = NULL;
-
/* ------------------- Setup new frame from picture -------------------- */
- TIMER_START( i_mtime_encode_frame );
if( pic_in != NULL )
{
/* 1: Copy the picture to a frame and move it to a buffer */
- x264_frame_t *fenc = x264_frame_get( h->frames.unused );
+ x264_frame_t *fenc = x264_frame_pop_unused( h );
- x264_frame_copy_picture( h, fenc, pic_in );
+ if( x264_frame_copy_picture( h, fenc, pic_in ) < 0 )
+ return -1;
- if( h->param.i_width % 16 || h->param.i_height % 16 )
+ if( h->param.i_width != 16 * h->sps->i_mb_width ||
+ h->param.i_height != 16 * h->sps->i_mb_height )
x264_frame_expand_border_mod16( h, fenc );
fenc->i_frame = h->frames.i_input++;
- x264_frame_put( h->frames.next, fenc );
+ x264_frame_push( h->frames.next, fenc );
if( h->frames.b_have_lowres )
- x264_frame_init_lowres( h->param.cpu, fenc );
+ x264_frame_init_lowres( h, fenc );
- if( h->frames.i_input <= h->frames.i_delay )
+ if( h->frames.i_input <= h->frames.i_delay + 1 - h->param.i_threads )
{
/* Nothing yet to encode */
/* waiting for filling bframe buffer */
int bframes = 0;
/* 2: Select frame types */
if( h->frames.next[0] == NULL )
+ {
+ x264_encoder_frame_end( thread_oldest, thread_current, pp_nal, pi_nal, pic_out );
return 0;
+ }
x264_slicetype_decide( h );
/* 3: move some B-frames and 1 non-B to encode queue */
while( IS_X264_TYPE_B( h->frames.next[bframes]->i_type ) )
bframes++;
- x264_frame_put( h->frames.current, x264_frame_get( &h->frames.next[bframes] ) );
+ x264_frame_push( h->frames.current, x264_frame_shift( &h->frames.next[bframes] ) );
/* FIXME: when max B-frames > 3, BREF may no longer be centered after GOP closing */
if( h->param.b_bframe_pyramid && bframes > 1 )
{
- x264_frame_t *mid = x264_frame_get( &h->frames.next[bframes/2] );
+ x264_frame_t *mid = x264_frame_shift( &h->frames.next[bframes/2] );
mid->i_type = X264_TYPE_BREF;
- x264_frame_put( h->frames.current, mid );
+ x264_frame_push( h->frames.current, mid );
bframes--;
}
while( bframes-- )
- x264_frame_put( h->frames.current, x264_frame_get( h->frames.next ) );
+ x264_frame_push( h->frames.current, x264_frame_shift( h->frames.next ) );
}
- TIMER_STOP( i_mtime_encode_frame );
/* ------------------- Get frame to be encoded ------------------------- */
/* 4: get picture to encode */
- h->fenc = x264_frame_get( h->frames.current );
+ h->fenc = x264_frame_shift( h->frames.current );
if( h->fenc == NULL )
{
/* Nothing yet to encode (ex: waiting for I/P with B frames) */
}
/* ------------------- Setup frame context ----------------------------- */
- /* 5: Init data dependant of frame type */
- TIMER_START( i_mtime_encode_frame );
+ /* 5: Init data dependent of frame type */
if( h->fenc->i_type == X264_TYPE_IDR )
{
/* reset ref pictures */
i_nal_type = NAL_SLICE_IDR;
i_nal_ref_idc = NAL_PRIORITY_HIGHEST;
- i_slice_type = SLICE_TYPE_I;
+ h->sh.i_type = SLICE_TYPE_I;
}
else if( h->fenc->i_type == X264_TYPE_I )
{
i_nal_type = NAL_SLICE;
i_nal_ref_idc = NAL_PRIORITY_HIGH; /* Not completely true but for now it is (as all I/P are kept as ref)*/
- i_slice_type = SLICE_TYPE_I;
+ h->sh.i_type = SLICE_TYPE_I;
}
else if( h->fenc->i_type == X264_TYPE_P )
{
i_nal_type = NAL_SLICE;
i_nal_ref_idc = NAL_PRIORITY_HIGH; /* Not completely true but for now it is (as all I/P are kept as ref)*/
- i_slice_type = SLICE_TYPE_P;
+ h->sh.i_type = SLICE_TYPE_P;
}
else if( h->fenc->i_type == X264_TYPE_BREF )
{
i_nal_type = NAL_SLICE;
i_nal_ref_idc = NAL_PRIORITY_HIGH; /* maybe add MMCO to forget it? -> low */
- i_slice_type = SLICE_TYPE_B;
+ h->sh.i_type = SLICE_TYPE_B;
}
else /* B frame */
{
i_nal_type = NAL_SLICE;
i_nal_ref_idc = NAL_PRIORITY_DISPOSABLE;
- i_slice_type = SLICE_TYPE_B;
+ h->sh.i_type = SLICE_TYPE_B;
}
h->fdec->i_poc =
h->fdec->i_type = h->fenc->i_type;
h->fdec->i_frame = h->fenc->i_frame;
h->fenc->b_kept_as_ref =
- h->fdec->b_kept_as_ref = i_nal_ref_idc != NAL_PRIORITY_DISPOSABLE;
+ h->fdec->b_kept_as_ref = i_nal_ref_idc != NAL_PRIORITY_DISPOSABLE && h->param.i_keyint_max > 1;
/* ------------------- Init ----------------------------- */
/* build ref list 0/1 */
- x264_reference_build_list( h, h->fdec->i_poc, i_slice_type );
+ x264_reference_build_list( h, h->fdec->i_poc );
/* Init the rate control */
- x264_ratecontrol_start( h, i_slice_type, h->fenc->i_qpplus1 );
+ x264_ratecontrol_start( h, h->fenc->i_qpplus1 );
i_global_qp = x264_ratecontrol_qp( h );
pic_out->i_qpplus1 =
h->fdec->i_qpplus1 = i_global_qp + 1;
- if( i_slice_type == SLICE_TYPE_B )
+ if( h->sh.i_type == SLICE_TYPE_B )
x264_macroblock_bipred_init( h );
- if( h->fenc->b_kept_as_ref )
- h->i_frame_num++;
-
/* ------------------------ Create slice header ----------------------- */
- x264_slice_init( h, i_nal_type, i_slice_type, i_global_qp );
+ x264_slice_init( h, i_nal_type, i_global_qp );
+
+ if( i_nal_ref_idc != NAL_PRIORITY_DISPOSABLE )
+ h->i_frame_num++;
/* ---------------------- Write the bitstream -------------------------- */
/* Init bitstream context */
if(h->param.b_aud){
int pic_type;
- if(i_slice_type == SLICE_TYPE_I)
+ if(h->sh.i_type == SLICE_TYPE_I)
pic_type = 0;
- else if(i_slice_type == SLICE_TYPE_P)
+ else if(h->sh.i_type == SLICE_TYPE_P)
pic_type = 1;
- else if(i_slice_type == SLICE_TYPE_B)
+ else if(h->sh.i_type == SLICE_TYPE_B)
pic_type = 2;
else
pic_type = 7;
h->i_nal_ref_idc = i_nal_ref_idc;
/* Write SPS and PPS */
- if( i_nal_type == NAL_SLICE_IDR )
+ if( i_nal_type == NAL_SLICE_IDR && h->param.b_repeat_headers )
{
if( h->fenc->i_frame == 0 )
{
/* identify ourself */
x264_nal_start( h, NAL_SEI, NAL_PRIORITY_DISPOSABLE );
- x264_sei_version_write( &h->out.bs );
+ x264_sei_version_write( h, &h->out.bs );
x264_nal_end( h );
}
}
/* Write frame */
- i_frame_size = x264_slices_write( h );
+ if( h->param.i_threads > 1 )
+ {
+ x264_pthread_create( &h->thread_handle, NULL, (void*)x264_slices_write, h );
+ h->b_thread_active = 1;
+ }
+ else
+ x264_slices_write( h );
/* restore CPU state (before using float again) */
- x264_cpu_restore( h->param.cpu );
+ x264_emms();
- if( i_slice_type == SLICE_TYPE_P && !h->param.rc.b_stat_read
- && h->param.i_scenecut_threshold >= 0 )
+ if( h->sh.i_type == SLICE_TYPE_P && !h->param.rc.b_stat_read
+ && h->param.i_scenecut_threshold >= 0
+ && !h->param.b_pre_scenecut )
{
const int *mbs = h->stat.frame.i_mb_count;
int i_mb_i = mbs[I_16x16] + mbs[I_8x8] + mbs[I_4x4];
/* macroblock_analyse() doesn't further analyse skipped mbs,
* so we have to guess their cost */
- if( i_mb_s < i_mb )
- i_intra_cost = i_intra_cost * i_mb / (i_mb - i_mb_s);
+ if( h->stat.frame.i_mbs_analysed > 0 )
+ i_intra_cost = i_intra_cost * i_mb / h->stat.frame.i_mbs_analysed;
if( i_gop_size < h->param.i_keyint_min / 4 )
f_bias = f_thresh_min / 4;
f_bias = X264_MIN( f_bias, 1.0 );
/* Bad P will be reencoded as I */
- if( i_mb_s < i_mb &&
+ if( h->stat.frame.i_mbs_analysed > 0 &&
i_inter_cost >= (1.0 - f_bias) * i_intra_cost )
{
int b;
- x264_log( h, X264_LOG_DEBUG, "scene cut at %d Icost:%.0f Pcost:%.0f ratio:%.3f bias=%.3f lastIDR:%d (I:%d P:%d S:%d)\n",
+ x264_log( h, X264_LOG_DEBUG, "scene cut at %d Icost:%.0f Pcost:%.0f ratio:%.4f bias:%.4f gop:%d (imb:%d pmb:%d smb:%d)\n",
h->fenc->i_frame,
(double)i_intra_cost, (double)i_inter_cost,
- (double)i_inter_cost / i_intra_cost,
+ 1. - (double)i_inter_cost / i_intra_cost,
f_bias, i_gop_size,
i_mb_i, i_mb_p, i_mb_s );
if( h->param.b_bframe_adaptive || b > 1 )
h->fenc->i_type = X264_TYPE_AUTO;
x264_frame_sort_pts( h->frames.current );
- x264_frame_push( h->frames.next, h->fenc );
+ x264_frame_unshift( h->frames.next, h->fenc );
h->fenc = h->frames.current[b-1];
h->frames.current[b-1] = NULL;
h->fenc->i_type = X264_TYPE_P;
/* Do IDR if needed */
else if( i_gop_size >= h->param.i_keyint_min )
{
- x264_frame_t *tmp;
-
/* Reset */
h->i_frame_num = 0;
h->fenc->i_poc = 0;
/* Put enqueued frames back in the pool */
- while( (tmp = x264_frame_get( h->frames.current ) ) != NULL )
- x264_frame_put( h->frames.next, tmp );
+ while( h->frames.current[0] )
+ x264_frame_push( h->frames.next, x264_frame_shift( h->frames.current ) );
x264_frame_sort_pts( h->frames.next );
}
else
}
}
+ x264_encoder_frame_end( thread_oldest, thread_current, pp_nal, pi_nal, pic_out );
+ return 0;
+}
+
+static void x264_encoder_frame_end( x264_t *h, x264_t *thread_current,
+ x264_nal_t **pp_nal, int *pi_nal,
+ x264_picture_t *pic_out )
+{
+ int i;
+ char psz_message[80];
+
+ if( h->b_thread_active )
+ {
+ x264_pthread_join( h->thread_handle, NULL );
+ h->b_thread_active = 0;
+ }
+ if( !h->out.i_nal )
+ {
+ pic_out->i_type = X264_TYPE_AUTO;
+ return;
+ }
+
+ x264_frame_push_unused( thread_current, h->fenc );
+
/* End bitstream, set output */
*pi_nal = h->out.i_nal;
*pp_nal = h->out.nal;
+ h->out.i_nal = 0;
/* Set output picture properties */
- if( i_slice_type == SLICE_TYPE_I )
- pic_out->i_type = i_nal_type == NAL_SLICE_IDR ? X264_TYPE_IDR : X264_TYPE_I;
- else if( i_slice_type == SLICE_TYPE_P )
+ if( h->sh.i_type == SLICE_TYPE_I )
+ pic_out->i_type = h->i_nal_type == NAL_SLICE_IDR ? X264_TYPE_IDR : X264_TYPE_I;
+ else if( h->sh.i_type == SLICE_TYPE_P )
pic_out->i_type = X264_TYPE_P;
else
pic_out->i_type = X264_TYPE_B;
}
/* ---------------------- Update encoder state ------------------------- */
- /* update cabac */
- if( h->param.b_cabac && h->param.i_cabac_init_idc == -1 )
- {
- x264_cabac_model_update( &h->cabac, i_slice_type, h->sh.i_qp );
- }
-
- /* handle references */
- if( i_nal_ref_idc != NAL_PRIORITY_DISPOSABLE )
- {
- x264_reference_update( h );
- }
-
- /* increase frame count */
- h->i_frame++;
-
- /* restore CPU state (before using float again) */
- /* XXX: not needed? (done above) */
- x264_cpu_restore( h->param.cpu );
/* update rc */
- x264_ratecontrol_end( h, i_frame_size * 8 );
+ x264_emms();
+ x264_ratecontrol_end( h, h->out.i_frame_size * 8 );
- x264_frame_put( h->frames.unused, h->fenc );
+ /* restore CPU state (before using float again) */
+ x264_emms();
- TIMER_STOP( i_mtime_encode_frame );
+ x264_noise_reduction_update( thread_current );
/* ---------------------- Compute/Print statistics --------------------- */
+ x264_thread_sync_stat( h, h->thread[0] );
+
/* Slice stat */
- h->stat.i_slice_count[i_slice_type]++;
- h->stat.i_slice_size[i_slice_type] += i_frame_size + NALU_OVERHEAD;
- h->stat.i_slice_qp[i_slice_type] += i_global_qp;
+ h->stat.i_slice_count[h->sh.i_type]++;
+ h->stat.i_slice_size[h->sh.i_type] += h->out.i_frame_size + NALU_OVERHEAD;
+ h->stat.f_slice_qp[h->sh.i_type] += h->fdec->f_qp_avg_aq;
- for( i = 0; i < 19; i++ )
+ for( i = 0; i < X264_MBTYPE_MAX; i++ )
h->stat.i_mb_count[h->sh.i_type][i] += h->stat.frame.i_mb_count[i];
for( i = 0; i < 2; i++ )
h->stat.i_mb_count_8x8dct[i] += h->stat.frame.i_mb_count_8x8dct[i];
{
for( i = 0; i < 7; i++ )
h->stat.i_mb_count_size[h->sh.i_type][i] += h->stat.frame.i_mb_count_size[i];
- for( i = 0; i < 16; i++ )
+ for( i = 0; i < 32; i++ )
h->stat.i_mb_count_ref[h->sh.i_type][i] += h->stat.frame.i_mb_count_ref[i];
}
+ if( h->sh.i_type == SLICE_TYPE_B )
+ {
+ h->stat.i_direct_frames[ h->sh.b_direct_spatial_mv_pred ] ++;
+ if( h->mb.b_direct_auto_write )
+ {
+ //FIXME somewhat arbitrary time constants
+ if( h->stat.i_direct_score[0] + h->stat.i_direct_score[1] > h->mb.i_mb_count )
+ {
+ for( i = 0; i < 2; i++ )
+ h->stat.i_direct_score[i] = h->stat.i_direct_score[i] * 9/10;
+ }
+ for( i = 0; i < 2; i++ )
+ h->stat.i_direct_score[i] += h->stat.frame.i_direct_score[i];
+ }
+ }
+ psz_message[0] = '\0';
if( h->param.analyse.b_psnr )
{
- int64_t i_sqe_y, i_sqe_u, i_sqe_v;
-
- /* PSNR */
- i_sqe_y = x264_pixel_ssd_wxh( &h->pixf, frame_psnr->plane[0], frame_psnr->i_stride[0], h->fenc->plane[0], h->fenc->i_stride[0], h->param.i_width, h->param.i_height );
- i_sqe_u = x264_pixel_ssd_wxh( &h->pixf, frame_psnr->plane[1], frame_psnr->i_stride[1], h->fenc->plane[1], h->fenc->i_stride[1], h->param.i_width/2, h->param.i_height/2);
- i_sqe_v = x264_pixel_ssd_wxh( &h->pixf, frame_psnr->plane[2], frame_psnr->i_stride[2], h->fenc->plane[2], h->fenc->i_stride[2], h->param.i_width/2, h->param.i_height/2);
- x264_cpu_restore( h->param.cpu );
+ int64_t sqe[3] = {
+ h->stat.frame.i_ssd[0],
+ h->stat.frame.i_ssd[1],
+ h->stat.frame.i_ssd[2],
+ };
- h->stat.i_sqe_global[i_slice_type] += i_sqe_y + i_sqe_u + i_sqe_v;
- h->stat.f_psnr_average[i_slice_type] += x264_psnr( i_sqe_y + i_sqe_u + i_sqe_v, 3 * h->param.i_width * h->param.i_height / 2 );
- h->stat.f_psnr_mean_y[i_slice_type] += x264_psnr( i_sqe_y, h->param.i_width * h->param.i_height );
- h->stat.f_psnr_mean_u[i_slice_type] += x264_psnr( i_sqe_u, h->param.i_width * h->param.i_height / 4 );
- h->stat.f_psnr_mean_v[i_slice_type] += x264_psnr( i_sqe_v, h->param.i_width * h->param.i_height / 4 );
+ h->stat.i_sqe_global[h->sh.i_type] += sqe[0] + sqe[1] + sqe[2];
+ h->stat.f_psnr_average[h->sh.i_type] += x264_psnr( sqe[0] + sqe[1] + sqe[2], 3 * h->param.i_width * h->param.i_height / 2 );
+ h->stat.f_psnr_mean_y[h->sh.i_type] += x264_psnr( sqe[0], h->param.i_width * h->param.i_height );
+ h->stat.f_psnr_mean_u[h->sh.i_type] += x264_psnr( sqe[1], h->param.i_width * h->param.i_height / 4 );
+ h->stat.f_psnr_mean_v[h->sh.i_type] += x264_psnr( sqe[2], h->param.i_width * h->param.i_height / 4 );
- snprintf( psz_message, 80, " PSNR Y:%2.2f U:%2.2f V:%2.2f",
- x264_psnr( i_sqe_y, h->param.i_width * h->param.i_height ),
- x264_psnr( i_sqe_u, h->param.i_width * h->param.i_height / 4),
- x264_psnr( i_sqe_v, h->param.i_width * h->param.i_height / 4) );
- psz_message[79] = '\0';
+ snprintf( psz_message, 80, " PSNR Y:%5.2f U:%5.2f V:%5.2f",
+ x264_psnr( sqe[0], h->param.i_width * h->param.i_height ),
+ x264_psnr( sqe[1], h->param.i_width * h->param.i_height / 4),
+ x264_psnr( sqe[2], h->param.i_width * h->param.i_height / 4) );
}
- else
+
+ if( h->param.analyse.b_ssim )
{
- psz_message[0] = '\0';
+ double ssim_y = h->stat.frame.f_ssim
+ / (((h->param.i_width-6)>>2) * ((h->param.i_height-6)>>2));
+ h->stat.f_ssim_mean_y[h->sh.i_type] += ssim_y;
+ snprintf( psz_message + strlen(psz_message), 80 - strlen(psz_message),
+ " SSIM Y:%.5f", ssim_y );
}
+ psz_message[79] = '\0';
x264_log( h, X264_LOG_DEBUG,
- "frame=%4d QP=%i NAL=%d Slice:%c Poc:%-3d I:%-4d P:%-4d SKIP:%-4d size=%d bytes%s\n",
- h->i_frame - 1,
- i_global_qp,
- i_nal_ref_idc,
- i_slice_type == SLICE_TYPE_I ? 'I' : (i_slice_type == SLICE_TYPE_P ? 'P' : 'B' ),
- frame_psnr->i_poc,
+ "frame=%4d QP=%.2f NAL=%d Slice:%c Poc:%-3d I:%-4d P:%-4d SKIP:%-4d size=%d bytes%s\n",
+ h->i_frame,
+ h->fdec->f_qp_avg_aq,
+ h->i_nal_ref_idc,
+ h->sh.i_type == SLICE_TYPE_I ? 'I' : (h->sh.i_type == SLICE_TYPE_P ? 'P' : 'B' ),
+ h->fdec->i_poc,
h->stat.frame.i_mb_count_i,
h->stat.frame.i_mb_count_p,
h->stat.frame.i_mb_count_skip,
- i_frame_size,
+ h->out.i_frame_size,
psz_message );
+ // keep stats all in one place
+ x264_thread_sync_stat( h->thread[0], h );
+ // for the use of the next frame
+ x264_thread_sync_stat( thread_current, h );
#ifdef DEBUG_MB_TYPE
{
int mb_xy;
for( mb_xy = 0; mb_xy < h->sps->i_mb_width * h->sps->i_mb_height; mb_xy++ )
{
- if( h->mb.type[mb_xy] < 19 && h->mb.type[mb_xy] >= 0 )
+ if( h->mb.type[mb_xy] < X264_MBTYPE_MAX && h->mb.type[mb_xy] >= 0 )
fprintf( stderr, "%c ", mb_chars[ h->mb.type[mb_xy] ] );
else
fprintf( stderr, "? " );
}
#endif
-#ifdef DEBUG_DUMP_FRAME
- /* Dump reconstructed frame */
- x264_frame_dump( h, frame_psnr, "fdec.yuv" );
-#endif
- return 0;
+ if( h->param.psz_dump_yuv )
+ x264_frame_dump( h );
}
/****************************************************************************
****************************************************************************/
void x264_encoder_close ( x264_t *h )
{
-#ifdef DEBUG_BENCHMARK
- int64_t i_mtime_total = i_mtime_analyse + i_mtime_encode + i_mtime_write + i_mtime_filter + 1;
-#endif
int64_t i_yuv_size = 3 * h->param.i_width * h->param.i_height / 2;
int i;
-#ifdef DEBUG_BENCHMARK
- x264_log( h, X264_LOG_INFO,
- "analyse=%d(%lldms) encode=%d(%lldms) write=%d(%lldms) filter=%d(%lldms)\n",
- (int)(100*i_mtime_analyse/i_mtime_total), i_mtime_analyse/1000,
- (int)(100*i_mtime_encode/i_mtime_total), i_mtime_encode/1000,
- (int)(100*i_mtime_write/i_mtime_total), i_mtime_write/1000,
- (int)(100*i_mtime_filter/i_mtime_total), i_mtime_filter/1000 );
-#endif
+ for( i=0; i<h->param.i_threads; i++ )
+ {
+ // don't strictly have to wait for the other threads, but it's simpler than canceling them
+ if( h->thread[i]->b_thread_active )
+ x264_pthread_join( h->thread[i]->thread_handle, NULL );
+ }
/* Slices used and PSNR */
for( i=0; i<5; i++ )
"slice %s:%-5d Avg QP:%5.2f size:%6.0f PSNR Mean Y:%5.2f U:%5.2f V:%5.2f Avg:%5.2f Global:%5.2f\n",
slice_name[i_slice],
i_count,
- (double)h->stat.i_slice_qp[i_slice] / i_count,
+ h->stat.f_slice_qp[i_slice] / i_count,
(double)h->stat.i_slice_size[i_slice] / i_count,
h->stat.f_psnr_mean_y[i_slice] / i_count, h->stat.f_psnr_mean_u[i_slice] / i_count, h->stat.f_psnr_mean_v[i_slice] / i_count,
h->stat.f_psnr_average[i_slice] / i_count,
"slice %s:%-5d Avg QP:%5.2f size:%6.0f\n",
slice_name[i_slice],
i_count,
- (double)h->stat.i_slice_qp[i_slice] / i_count,
+ h->stat.f_slice_qp[i_slice] / i_count,
(double)h->stat.i_slice_size[i_slice] / i_count );
}
}
i_mb_count[B_SKIP] / i_count );
}
+ x264_ratecontrol_summary( h );
+
if( h->stat.i_slice_count[SLICE_TYPE_I] + h->stat.i_slice_count[SLICE_TYPE_P] + h->stat.i_slice_count[SLICE_TYPE_B] > 0 )
{
const int i_count = h->stat.i_slice_count[SLICE_TYPE_I] +
#define SUM3b(p,o) (p[SLICE_TYPE_I][o] + p[SLICE_TYPE_P][o] + p[SLICE_TYPE_B][o])
float f_bitrate = fps * SUM3(h->stat.i_slice_size) / i_count / 125;
- if( h->param.analyse.b_transform_8x8 )
+ if( h->pps->b_transform_8x8_mode )
{
int64_t i_i8x8 = SUM3b( h->stat.i_mb_count, I_8x8 );
int64_t i_intra = i_i8x8 + SUM3b( h->stat.i_mb_count, I_4x4 )
100. * h->stat.i_mb_count_8x8dct[1] / h->stat.i_mb_count_8x8dct[0] );
}
- if( h->param.i_frame_reference > 1 )
+ if( h->param.analyse.i_direct_mv_pred == X264_DIRECT_PRED_AUTO
+ && h->stat.i_slice_count[SLICE_TYPE_B] )
+ {
+ x264_log( h, X264_LOG_INFO, "direct mvs spatial:%.1f%% temporal:%.1f%%\n",
+ h->stat.i_direct_frames[1] * 100. / h->stat.i_slice_count[SLICE_TYPE_B],
+ h->stat.i_direct_frames[0] * 100. / h->stat.i_slice_count[SLICE_TYPE_B] );
+ }
+
+ if( h->frames.i_max_ref0 > 1 )
{
int i_slice;
for( i_slice = 0; i_slice < 2; i_slice++ )
char *p = buf;
int64_t i_den = 0;
int i_max = 0;
- for( i = 0; i < h->param.i_frame_reference; i++ )
+ for( i = 0; i < h->frames.i_max_ref0 << h->param.b_interlaced; i++ )
if( h->stat.i_mb_count_ref[i_slice][i] )
{
i_den += h->stat.i_mb_count_ref[i_slice][i];
}
}
+ if( h->param.analyse.b_ssim )
+ {
+ x264_log( h, X264_LOG_INFO,
+ "SSIM Mean Y:%.7f\n",
+ SUM3( h->stat.f_ssim_mean_y ) / i_count );
+ }
if( h->param.analyse.b_psnr )
+ {
x264_log( h, X264_LOG_INFO,
"PSNR Mean Y:%6.3f U:%6.3f V:%6.3f Avg:%6.3f Global:%6.3f kb/s:%.2f\n",
SUM3( h->stat.f_psnr_mean_y ) / i_count,
SUM3( h->stat.f_psnr_average ) / i_count,
x264_psnr( SUM3( h->stat.i_sqe_global ), i_count * i_yuv_size ),
f_bitrate );
+ }
else
x264_log( h, X264_LOG_INFO, "kb/s:%.1f\n", f_bitrate );
}
/* frames */
- for( i = 0; i < X264_BFRAME_MAX + 3; i++ )
- {
- if( h->frames.current[i] ) x264_frame_delete( h->frames.current[i] );
- if( h->frames.next[i] ) x264_frame_delete( h->frames.next[i] );
- if( h->frames.unused[i] ) x264_frame_delete( h->frames.unused[i] );
- }
- /* ref frames */
- for( i = 0; i < h->frames.i_max_dpb; i++ )
- {
+ for( i = 0; h->frames.current[i]; i++ )
+ x264_frame_delete( h->frames.current[i] );
+ for( i = 0; h->frames.next[i]; i++ )
+ x264_frame_delete( h->frames.next[i] );
+ for( i = 0; h->frames.unused[i]; i++ )
+ x264_frame_delete( h->frames.unused[i] );
+ for( i = 0; h->frames.reference[i]; i++ )
x264_frame_delete( h->frames.reference[i] );
- }
/* rc */
x264_ratecontrol_delete( h );
if( h->param.rc.psz_rc_eq )
free( h->param.rc.psz_rc_eq );
- x264_macroblock_cache_end( h );
- x264_free( h->out.p_bitstream );
- for( i = 1; i < h->param.i_threads; i++ )
+ x264_cqm_delete( h );
+ for( i = h->param.i_threads - 1; i >= 0; i-- )
+ {
+ x264_macroblock_cache_end( h->thread[i] );
+ x264_free( h->thread[i]->out.p_bitstream );
x264_free( h->thread[i] );
- x264_free( h );
+ }
}
-
projects / x264 / blobdiff