Index: libvpx/source/libvpx/vp8/encoder/rdopt.c |
diff --git a/libvpx/source/libvpx/vp8/encoder/rdopt.c b/libvpx/source/libvpx/vp8/encoder/rdopt.c |
index 992df71c29e83ff6924a90e34c889dcb8ddd1559..a125cc4819f2615bc11dc409aa32d46cdf64598d 100644 |
--- a/libvpx/source/libvpx/vp8/encoder/rdopt.c |
+++ b/libvpx/source/libvpx/vp8/encoder/rdopt.c |
@@ -31,7 +31,7 @@ |
#include "vp8/common/g_common.h" |
#include "variance.h" |
#include "mcomp.h" |
-#include "rdopt.h" |
+ |
#include "vpx_mem/vpx_mem.h" |
#include "dct.h" |
#include "vp8/common/systemdependent.h" |
@@ -46,8 +46,13 @@ |
extern void vp8cx_mb_init_quantizer(VP8_COMP *cpi, MACROBLOCK *x); |
extern void vp8_update_zbin_extra(VP8_COMP *cpi, MACROBLOCK *x); |
+ |
+#define RDCOST(RM,DM,R,D) ( ((128+(R)*(RM)) >> 8) + (DM)*(D) ) |
+ |
#define MAXF(a,b) (((a) > (b)) ? (a) : (b)) |
+ |
+ |
static const int auto_speed_thresh[17] = |
{ |
1000, |
@@ -134,8 +139,8 @@ const MV_REFERENCE_FRAME vp8_ref_frame_order[MAX_MODES] = |
}; |
static void fill_token_costs( |
- unsigned int c [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [MAX_ENTROPY_TOKENS], |
- const vp8_prob p [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [ENTROPY_NODES] |
+ unsigned int c [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens], |
+ const vp8_prob p [BLOCK_TYPES] [COEF_BANDS] [PREV_COEF_CONTEXTS] [vp8_coef_tokens-1] |
) |
{ |
int i, j, k; |
@@ -155,44 +160,45 @@ static int rd_iifactor [ 32 ] = { 4, 4, 3, 2, 1, 0, 0, 0, |
0, 0, 0, 0, 0, 0, 0, 0, |
}; |
+ |
/* values are now correlated to quantizer */ |
static int sad_per_bit16lut[QINDEX_RANGE] = |
{ |
- 2, 2, 2, 2, 2, 2, 2, 2, |
- 2, 2, 2, 2, 2, 2, 2, 2, |
- 3, 3, 3, 3, 3, 3, 3, 3, |
- 3, 3, 3, 3, 3, 3, 4, 4, |
- 4, 4, 4, 4, 4, 4, 4, 4, |
- 4, 4, 5, 5, 5, 5, 5, 5, |
5, 5, 5, 5, 5, 5, 6, 6, |
- 6, 6, 6, 6, 6, 6, 6, 6, |
- 6, 6, 7, 7, 7, 7, 7, 7, |
- 7, 7, 7, 7, 7, 7, 8, 8, |
- 8, 8, 8, 8, 8, 8, 8, 8, |
- 8, 8, 9, 9, 9, 9, 9, 9, |
- 9, 9, 9, 9, 9, 9, 10, 10, |
- 10, 10, 10, 10, 10, 10, 11, 11, |
- 11, 11, 11, 11, 12, 12, 12, 12, |
- 12, 12, 13, 13, 13, 13, 14, 14 |
+ 6, 6, 6, 6, 6, 7, 7, 7, |
+ 7, 7, 7, 7, 8, 8, 8, 8, |
+ 8, 8, 8, 8, 8, 8, 9, 9, |
+ 9, 9, 9, 9, 10, 10, 10, 10, |
+ 10, 10, 11, 11, 11, 11, 11, 11, |
+ 12, 12, 12, 12, 12, 12, 12, 13, |
+ 13, 13, 13, 13, 13, 14, 14, 14, |
+ 14, 14, 15, 15, 15, 15, 15, 15, |
+ 16, 16, 16, 16, 16, 16, 17, 17, |
+ 17, 17, 17, 17, 17, 18, 18, 18, |
+ 18, 18, 19, 19, 19, 19, 19, 19, |
+ 20, 20, 20, 21, 21, 21, 21, 22, |
+ 22, 22, 23, 23, 23, 24, 24, 24, |
+ 25, 25, 26, 26, 27, 27, 27, 28, |
+ 28, 28, 29, 29, 30, 30, 31, 31 |
}; |
static int sad_per_bit4lut[QINDEX_RANGE] = |
{ |
- 2, 2, 2, 2, 2, 2, 3, 3, |
- 3, 3, 3, 3, 3, 3, 3, 3, |
- 3, 3, 3, 3, 4, 4, 4, 4, |
- 4, 4, 4, 4, 4, 4, 5, 5, |
- 5, 5, 5, 5, 6, 6, 6, 6, |
- 6, 6, 6, 6, 6, 6, 6, 6, |
- 7, 7, 7, 7, 7, 7, 7, 7, |
+ 5, 5, 5, 5, 5, 5, 7, 7, |
7, 7, 7, 7, 7, 8, 8, 8, |
- 8, 8, 9, 9, 9, 9, 9, 9, |
- 10, 10, 10, 10, 10, 10, 10, 10, |
- 11, 11, 11, 11, 11, 11, 11, 11, |
- 12, 12, 12, 12, 12, 12, 12, 12, |
- 13, 13, 13, 13, 13, 13, 13, 14, |
- 14, 14, 14, 14, 15, 15, 15, 15, |
- 16, 16, 16, 16, 17, 17, 17, 18, |
- 18, 18, 19, 19, 19, 20, 20, 20, |
+ 8, 8, 8, 8, 10, 10, 10, 10, |
+ 10, 10, 10, 10, 10, 10, 11, 11, |
+ 11, 11, 11, 11, 13, 13, 13, 13, |
+ 13, 13, 14, 14, 14, 14, 14, 14, |
+ 16, 16, 16, 16, 16, 16, 16, 17, |
+ 17, 17, 17, 17, 17, 19, 19, 19, |
+ 19, 19, 20, 20, 20, 20, 20, 20, |
+ 22, 22, 22, 22, 22, 22, 23, 23, |
+ 23, 23, 23, 23, 23, 25, 25, 25, |
+ 25, 25, 26, 26, 26, 26, 26, 26, |
+ 28, 28, 28, 29, 29, 29, 29, 31, |
+ 31, 31, 32, 32, 32, 34, 34, 34, |
+ 35, 35, 37, 37, 38, 38, 38, 40, |
+ 40, 40, 41, 41, 43, 43, 44, 44, |
}; |
void vp8cx_initialize_me_consts(VP8_COMP *cpi, int QIndex) |
@@ -206,7 +212,7 @@ void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue) |
int q; |
int i; |
double capped_q = (Qvalue < 160) ? (double)Qvalue : 160.0; |
- double rdconst = 2.70; |
+ double rdconst = 3.00; |
vp8_clear_system_state(); //__asm emms; |
@@ -231,18 +237,20 @@ void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue) |
if (cpi->pass == 2 && (cpi->common.frame_type != KEY_FRAME)) |
{ |
- if (cpi->twopass.next_iiratio > 31) |
+ if (cpi->next_iiratio > 31) |
cpi->RDMULT += (cpi->RDMULT * rd_iifactor[31]) >> 4; |
else |
- cpi->RDMULT += |
- (cpi->RDMULT * rd_iifactor[cpi->twopass.next_iiratio]) >> 4; |
+ cpi->RDMULT += (cpi->RDMULT * rd_iifactor[cpi->next_iiratio]) >> 4; |
} |
- cpi->mb.errorperbit = (cpi->RDMULT / 110); |
+ cpi->mb.errorperbit = (cpi->RDMULT / 100); |
cpi->mb.errorperbit += (cpi->mb.errorperbit==0); |
vp8_set_speed_features(cpi); |
+ if (cpi->common.simpler_lpf) |
+ cpi->common.filter_type = SIMPLE_LOOPFILTER; |
+ |
q = (int)pow(Qvalue, 1.25); |
if (q < 8) |
@@ -297,6 +305,8 @@ void vp8_initialize_rd_consts(VP8_COMP *cpi, int Qvalue) |
void vp8_auto_select_speed(VP8_COMP *cpi) |
{ |
+ int used = cpi->oxcf.cpu_used; |
+ |
int milliseconds_for_compress = (int)(1000000 / cpi->oxcf.frame_rate); |
milliseconds_for_compress = milliseconds_for_compress * (16 - cpi->oxcf.cpu_used) / 16; |
@@ -317,10 +327,10 @@ void vp8_auto_select_speed(VP8_COMP *cpi) |
/* |
// this is done during parameter valid check |
- if( cpi->oxcf.cpu_used > 16) |
- cpi->oxcf.cpu_used = 16; |
- if( cpi->oxcf.cpu_used < -16) |
- cpi->oxcf.cpu_used = -16; |
+ if( used > 16) |
+ used = 16; |
+ if( used < -16) |
+ used = -16; |
*/ |
if (cpi->avg_pick_mode_time < milliseconds_for_compress && (cpi->avg_encode_time - cpi->avg_pick_mode_time) < milliseconds_for_compress) |
@@ -455,24 +465,22 @@ int VP8_UVSSE(MACROBLOCK *x, const vp8_variance_rtcd_vtable_t *rtcd) |
if ((mv_row | mv_col) & 7) |
{ |
- VARIANCE_INVOKE(rtcd, subpixvar8x8)(uptr, pre_stride, |
- mv_col & 7, mv_row & 7, upred_ptr, uv_stride, &sse2); |
- VARIANCE_INVOKE(rtcd, subpixvar8x8)(vptr, pre_stride, |
- mv_col & 7, mv_row & 7, vpred_ptr, uv_stride, &sse1); |
+ VARIANCE_INVOKE(rtcd, subpixvar8x8)(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, uv_stride, &sse2); |
+ VARIANCE_INVOKE(rtcd, subpixvar8x8)(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, uv_stride, &sse1); |
sse2 += sse1; |
} |
else |
{ |
- VARIANCE_INVOKE(rtcd, var8x8)(uptr, pre_stride, |
- upred_ptr, uv_stride, &sse2); |
- VARIANCE_INVOKE(rtcd, var8x8)(vptr, pre_stride, |
- vpred_ptr, uv_stride, &sse1); |
+ VARIANCE_INVOKE(rtcd, subpixvar8x8)(uptr, pre_stride, mv_col & 7, mv_row & 7, upred_ptr, uv_stride, &sse2); |
+ VARIANCE_INVOKE(rtcd, subpixvar8x8)(vptr, pre_stride, mv_col & 7, mv_row & 7, vpred_ptr, uv_stride, &sse1); |
sse2 += sse1; |
} |
+ |
return sse2; |
} |
+#if !(CONFIG_REALTIME_ONLY) |
static int cost_coeffs(MACROBLOCK *mb, BLOCKD *b, int type, ENTROPY_CONTEXT *a, ENTROPY_CONTEXT *l) |
{ |
int c = !type; /* start at coef 0, unless Y with Y2 */ |
@@ -543,8 +551,8 @@ static void macro_block_yrd( MACROBLOCK *mb, |
BLOCK *beptr; |
int d; |
- ENCODEMB_INVOKE(rtcd, submby)( mb->src_diff, *(mb->block[0].base_src), |
- mb->e_mbd.predictor, mb->block[0].src_stride ); |
+ ENCODEMB_INVOKE(rtcd, submby)( mb->src_diff, mb->src.y_buffer, |
+ mb->e_mbd.predictor, mb->src.y_stride ); |
// Fdct and building the 2nd order block |
for (beptr = mb->block; beptr < mb->block + 16; beptr += 2) |
@@ -621,8 +629,7 @@ static int rd_pick_intra4x4block( |
rate = bmode_costs[mode]; |
- RECON_INVOKE(&cpi->rtcd.common->recon, intra4x4_predict) |
- (b, mode, b->predictor); |
+ vp8_predict_intra4x4(b, mode, b->predictor); |
ENCODEMB_INVOKE(IF_RTCD(&cpi->rtcd.encodemb), subb)(be, b, 16); |
x->vp8_short_fdct4x4(be->src_diff, be->coeff, 32); |
x->quantize_b(be, b); |
@@ -649,7 +656,8 @@ static int rd_pick_intra4x4block( |
vpx_memcpy(best_dqcoeff, b->dqcoeff, 32); |
} |
} |
- b->bmi.as_mode = (B_PREDICTION_MODE)(*best_mode); |
+ |
+ b->bmi.mode = (B_PREDICTION_MODE)(*best_mode); |
IDCT_INVOKE(IF_RTCD(&cpi->rtcd.common->idct), idct16)(best_dqcoeff, b->diff, 32); |
RECON_INVOKE(IF_RTCD(&cpi->rtcd.common->recon), recon)(best_predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride); |
@@ -657,15 +665,15 @@ static int rd_pick_intra4x4block( |
return best_rd; |
} |
-static int rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate, |
- int *rate_y, int *Distortion, int best_rd) |
+int vp8_rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate, |
+ int *rate_y, int *Distortion, int best_rd) |
{ |
MACROBLOCKD *const xd = &mb->e_mbd; |
int i; |
int cost = mb->mbmode_cost [xd->frame_type] [B_PRED]; |
int distortion = 0; |
int tot_rate_y = 0; |
- int64_t total_rd = 0; |
+ long long total_rd = 0; |
ENTROPY_CONTEXT_PLANES t_above, t_left; |
ENTROPY_CONTEXT *ta; |
ENTROPY_CONTEXT *tl; |
@@ -690,8 +698,8 @@ static int rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate, |
if (mb->e_mbd.frame_type == KEY_FRAME) |
{ |
- const B_PREDICTION_MODE A = above_block_mode(mic, i, mis); |
- const B_PREDICTION_MODE L = left_block_mode(mic, i); |
+ const B_PREDICTION_MODE A = vp8_above_bmi(mic, i, mis)->mode; |
+ const B_PREDICTION_MODE L = vp8_left_bmi(mic, i)->mode; |
bmode_costs = mb->bmode_costs[A][L]; |
} |
@@ -704,14 +712,13 @@ static int rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate, |
cost += r; |
distortion += d; |
tot_rate_y += ry; |
+ mic->bmi[i].mode = xd->block[i].bmi.mode = best_mode; |
- mic->bmi[i].as_mode = best_mode; |
- |
- if(total_rd >= (int64_t)best_rd) |
+ if(total_rd >= (long long)best_rd) |
break; |
} |
- if(total_rd >= (int64_t)best_rd) |
+ if(total_rd >= (long long)best_rd) |
return INT_MAX; |
*Rate = cost; |
@@ -720,13 +727,11 @@ static int rd_pick_intra4x4mby_modes(VP8_COMP *cpi, MACROBLOCK *mb, int *Rate, |
return RDCOST(mb->rdmult, mb->rddiv, cost, distortion); |
} |
- |
- |
-static int rd_pick_intra16x16mby_mode(VP8_COMP *cpi, |
- MACROBLOCK *x, |
- int *Rate, |
- int *rate_y, |
- int *Distortion) |
+int vp8_rd_pick_intra16x16mby_mode(VP8_COMP *cpi, |
+ MACROBLOCK *x, |
+ int *Rate, |
+ int *rate_y, |
+ int *Distortion) |
{ |
MB_PREDICTION_MODE mode; |
MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(mode_selected); |
@@ -798,7 +803,7 @@ static int vp8_rd_inter_uv(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *distort |
return RDCOST(x->rdmult, x->rddiv, *rate, *distortion); |
} |
-static void rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *rate_tokenonly, int *distortion) |
+int vp8_rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int *rate_tokenonly, int *distortion) |
{ |
MB_PREDICTION_MODE mode; |
MB_PREDICTION_MODE UNINITIALIZED_IS_SAFE(mode_selected); |
@@ -813,8 +818,7 @@ static void rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int |
int this_rd; |
x->e_mbd.mode_info_context->mbmi.uv_mode = mode; |
- RECON_INVOKE(&cpi->rtcd.common->recon, build_intra_predictors_mbuv) |
- (&x->e_mbd); |
+ vp8_build_intra_predictors_mbuv(&x->e_mbd); |
ENCODEMB_INVOKE(IF_RTCD(&cpi->rtcd.encodemb), submbuv)(x->src_diff, |
x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor, |
x->src.uv_stride); |
@@ -842,7 +846,9 @@ static void rd_pick_intra_mbuv_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate, int |
*distortion = d; |
x->e_mbd.mode_info_context->mbmi.uv_mode = mode_selected; |
+ return best_rd; |
} |
+#endif |
int vp8_cost_mv_ref(MB_PREDICTION_MODE m, const int near_mv_ref_ct[4]) |
{ |
@@ -853,17 +859,29 @@ int vp8_cost_mv_ref(MB_PREDICTION_MODE m, const int near_mv_ref_ct[4]) |
vp8_mv_ref_encoding_array - NEARESTMV + m); |
} |
-void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, int_mv *mv) |
+void vp8_set_mbmode_and_mvs(MACROBLOCK *x, MB_PREDICTION_MODE mb, MV *mv) |
{ |
+ int i; |
+ |
x->e_mbd.mode_info_context->mbmi.mode = mb; |
- x->e_mbd.mode_info_context->mbmi.mv.as_int = mv->as_int; |
+ x->e_mbd.mode_info_context->mbmi.mv.as_mv.row = mv->row; |
+ x->e_mbd.mode_info_context->mbmi.mv.as_mv.col = mv->col; |
+ |
+ for (i = 0; i < 16; i++) |
+ { |
+ B_MODE_INFO *bmi = &x->e_mbd.block[i].bmi; |
+ bmi->mode = (B_PREDICTION_MODE) mb; |
+ bmi->mv.as_mv.row = mv->row; |
+ bmi->mv.as_mv.col = mv->col; |
+ } |
} |
+#if !(CONFIG_REALTIME_ONLY) |
static int labels2mode( |
MACROBLOCK *x, |
int const *labelings, int which_label, |
B_PREDICTION_MODE this_mode, |
- int_mv *this_mv, int_mv *best_ref_mv, |
+ MV *this_mv, MV *best_ref_mv, |
int *mvcost[2] |
) |
{ |
@@ -904,13 +922,13 @@ static int labels2mode( |
thismvcost = vp8_mv_bit_cost(this_mv, best_ref_mv, mvcost, 102); |
break; |
case LEFT4X4: |
- this_mv->as_int = col ? d[-1].bmi.mv.as_int : left_block_mv(mic, i); |
+ *this_mv = col ? d[-1].bmi.mv.as_mv : vp8_left_bmi(mic, i)->mv.as_mv; |
break; |
case ABOVE4X4: |
- this_mv->as_int = row ? d[-4].bmi.mv.as_int : above_block_mv(mic, i, mis); |
+ *this_mv = row ? d[-4].bmi.mv.as_mv : vp8_above_bmi(mic, i, mis)->mv.as_mv; |
break; |
case ZERO4X4: |
- this_mv->as_int = 0; |
+ this_mv->row = this_mv->col = 0; |
break; |
default: |
break; |
@@ -918,22 +936,17 @@ static int labels2mode( |
if (m == ABOVE4X4) // replace above with left if same |
{ |
- int_mv left_mv; |
- |
- left_mv.as_int = col ? d[-1].bmi.mv.as_int : |
- left_block_mv(mic, i); |
+ const MV mv = col ? d[-1].bmi.mv.as_mv : vp8_left_bmi(mic, i)->mv.as_mv; |
- if (left_mv.as_int == this_mv->as_int) |
+ if (mv.row == this_mv->row && mv.col == this_mv->col) |
m = LEFT4X4; |
} |
cost = x->inter_bmode_costs[ m]; |
} |
- d->bmi.mv.as_int = this_mv->as_int; |
- |
- x->partition_info->bmi[i].mode = m; |
- x->partition_info->bmi[i].mv.as_int = this_mv->as_int; |
+ d->bmi.mode = m; |
+ d->bmi.mv.as_mv = *this_mv; |
} |
while (++i < 16); |
@@ -993,8 +1006,8 @@ static const unsigned int segmentation_to_sseshift[4] = {3, 3, 2, 0}; |
typedef struct |
{ |
- int_mv *ref_mv; |
- int_mv mvp; |
+ MV *ref_mv; |
+ MV *mvp; |
int segment_rd; |
int segment_num; |
@@ -1008,7 +1021,7 @@ typedef struct |
int mvthresh; |
int *mdcounts; |
- int_mv sv_mvp[4]; // save 4 mvp from 8x8 |
+ MV sv_mvp[4]; // save 4 mvp from 8x8 |
int sv_istep[2]; // save 2 initial step_param for 16x8/8x16 |
} BEST_SEG_INFO; |
@@ -1070,7 +1083,7 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
for (i = 0; i < label_count; i++) |
{ |
- int_mv mode_mv[B_MODE_COUNT]; |
+ MV mode_mv[B_MODE_COUNT]; |
int best_label_rd = INT_MAX; |
B_PREDICTION_MODE mode_selected = ZERO4X4; |
int bestlabelyrate = 0; |
@@ -1100,7 +1113,7 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
int n; |
int thissme; |
int bestsme = INT_MAX; |
- int_mv temp_mv; |
+ MV temp_mv; |
BLOCK *c; |
BLOCKD *e; |
@@ -1112,9 +1125,8 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
{ |
if (segmentation == BLOCK_8X16 || segmentation == BLOCK_16X8) |
{ |
- bsi->mvp.as_int = bsi->sv_mvp[i].as_int; |
- if (i==1 && segmentation == BLOCK_16X8) |
- bsi->mvp.as_int = bsi->sv_mvp[2].as_int; |
+ bsi->mvp = &bsi->sv_mvp[i]; |
+ if (i==1 && segmentation == BLOCK_16X8) bsi->mvp = &bsi->sv_mvp[2]; |
step_param = bsi->sv_istep[i]; |
} |
@@ -1122,9 +1134,8 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
// use previous block's result as next block's MV predictor. |
if (segmentation == BLOCK_4X4 && i>0) |
{ |
- bsi->mvp.as_int = x->e_mbd.block[i-1].bmi.mv.as_int; |
- if (i==4 || i==8 || i==12) |
- bsi->mvp.as_int = x->e_mbd.block[i-4].bmi.mv.as_int; |
+ bsi->mvp = &(x->e_mbd.block[i-1].bmi.mv.as_mv); |
+ if (i==4 || i==8 || i==12) bsi->mvp = &(x->e_mbd.block[i-4].bmi.mv.as_mv); |
step_param = 2; |
} |
} |
@@ -1133,10 +1144,6 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
{ |
int sadpb = x->sadperbit4; |
- int_mv mvp_full; |
- |
- mvp_full.as_mv.row = bsi->mvp.as_mv.row >>3; |
- mvp_full.as_mv.col = bsi->mvp.as_mv.col >>3; |
// find first label |
n = vp8_mbsplit_offset[segmentation][i]; |
@@ -1144,11 +1151,15 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
c = &x->block[n]; |
e = &x->e_mbd.block[n]; |
+ if (cpi->sf.search_method == HEX) |
+ bestsme = vp8_hex_search(x, c, e, bsi->ref_mv, |
+ &mode_mv[NEW4X4], step_param, sadpb, &num00, v_fn_ptr, x->mvsadcost, x->mvcost, bsi->ref_mv); |
+ |
+ else |
{ |
- bestsme = cpi->diamond_search_sad(x, c, e, &mvp_full, |
- &mode_mv[NEW4X4], step_param, |
- sadpb, &num00, v_fn_ptr, |
- x->mvcost, bsi->ref_mv); |
+ bestsme = cpi->diamond_search_sad(x, c, e, bsi->mvp, |
+ &mode_mv[NEW4X4], step_param, |
+ sadpb / 2, &num00, v_fn_ptr, x->mvsadcost, x->mvcost, bsi->ref_mv); |
n = num00; |
num00 = 0; |
@@ -1161,16 +1172,15 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
num00--; |
else |
{ |
- thissme = cpi->diamond_search_sad(x, c, e, |
- &mvp_full, &temp_mv, |
- step_param + n, sadpb, |
- &num00, v_fn_ptr, |
- x->mvcost, bsi->ref_mv); |
+ thissme = cpi->diamond_search_sad(x, c, e, bsi->mvp, |
+ &temp_mv, step_param + n, |
+ sadpb / 2, &num00, v_fn_ptr, x->mvsadcost, x->mvcost, bsi->ref_mv); |
if (thissme < bestsme) |
{ |
bestsme = thissme; |
- mode_mv[NEW4X4].as_int = temp_mv.as_int; |
+ mode_mv[NEW4X4].row = temp_mv.row; |
+ mode_mv[NEW4X4].col = temp_mv.col; |
} |
} |
} |
@@ -1181,34 +1191,30 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
// Should we do a full search (best quality only) |
if ((cpi->compressor_speed == 0) && (bestsme >> sseshift) > 4000) |
{ |
- /* Check if mvp_full is within the range. */ |
- vp8_clamp_mv(&mvp_full, x->mv_col_min, x->mv_col_max, x->mv_row_min, x->mv_row_max); |
- |
- thissme = cpi->full_search_sad(x, c, e, &mvp_full, |
- sadpb, 16, v_fn_ptr, |
- x->mvcost, bsi->ref_mv); |
+ thissme = cpi->full_search_sad(x, c, e, bsi->mvp, |
+ sadpb / 4, 16, v_fn_ptr, x->mvcost, x->mvsadcost,bsi->ref_mv); |
if (thissme < bestsme) |
{ |
bestsme = thissme; |
- mode_mv[NEW4X4].as_int = e->bmi.mv.as_int; |
+ mode_mv[NEW4X4] = e->bmi.mv.as_mv; |
} |
else |
{ |
// The full search result is actually worse so re-instate the previous best vector |
- e->bmi.mv.as_int = mode_mv[NEW4X4].as_int; |
+ e->bmi.mv.as_mv = mode_mv[NEW4X4]; |
} |
} |
} |
if (bestsme < INT_MAX) |
{ |
- int distortion; |
- unsigned int sse; |
- cpi->find_fractional_mv_step(x, c, e, &mode_mv[NEW4X4], |
- bsi->ref_mv, x->errorperbit, v_fn_ptr, x->mvcost, |
- &distortion, &sse); |
- |
+ if (!cpi->common.full_pixel) |
+ cpi->find_fractional_mv_step(x, c, e, &mode_mv[NEW4X4], |
+ bsi->ref_mv, x->errorperbit / 2, v_fn_ptr, x->mvcost); |
+ else |
+ vp8_skip_fractional_mv_step(x, c, e, &mode_mv[NEW4X4], |
+ bsi->ref_mv, x->errorperbit, v_fn_ptr, x->mvcost); |
} |
} /* NEW4X4 */ |
@@ -1216,8 +1222,8 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
bsi->ref_mv, x->mvcost); |
// Trap vectors that reach beyond the UMV borders |
- if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) || |
- ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max)) |
+ if (((mode_mv[this_mode].row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].row >> 3) > x->mv_row_max) || |
+ ((mode_mv[this_mode].col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].col >> 3) > x->mv_col_max)) |
{ |
continue; |
} |
@@ -1272,8 +1278,8 @@ static void rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, |
{ |
BLOCKD *bd = &x->e_mbd.block[i]; |
- bsi->mvs[i].as_mv = x->partition_info->bmi[i].mv.as_mv; |
- bsi->modes[i] = x->partition_info->bmi[i].mode; |
+ bsi->mvs[i].as_mv = bd->bmi.mv.as_mv; |
+ bsi->modes[i] = bd->bmi.mode; |
bsi->eobs[i] = bd->eob; |
} |
} |
@@ -1294,7 +1300,7 @@ void vp8_cal_step_param(int sr, int *sp) |
} |
static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
- int_mv *best_ref_mv, int best_rd, |
+ MV *best_ref_mv, int best_rd, |
int *mdcounts, int *returntotrate, |
int *returnyrate, int *returndistortion, |
int mvthresh) |
@@ -1306,7 +1312,7 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
bsi.segment_rd = best_rd; |
bsi.ref_mv = best_ref_mv; |
- bsi.mvp.as_int = best_ref_mv->as_int; |
+ bsi.mvp = best_ref_mv; |
bsi.mvthresh = mvthresh; |
bsi.mdcounts = mdcounts; |
@@ -1332,10 +1338,10 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
if (bsi.segment_rd < best_rd) |
{ |
- int col_min = (best_ref_mv->as_mv.col>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv->as_mv.col & 7)?1:0); |
- int row_min = (best_ref_mv->as_mv.row>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv->as_mv.row & 7)?1:0); |
- int col_max = (best_ref_mv->as_mv.col>>3) + MAX_FULL_PEL_VAL; |
- int row_max = (best_ref_mv->as_mv.row>>3) + MAX_FULL_PEL_VAL; |
+ int col_min = (best_ref_mv->col - MAX_FULL_PEL_VAL) >>3; |
+ int col_max = (best_ref_mv->col + MAX_FULL_PEL_VAL) >>3; |
+ int row_min = (best_ref_mv->row - MAX_FULL_PEL_VAL) >>3; |
+ int row_max = (best_ref_mv->row + MAX_FULL_PEL_VAL) >>3; |
int tmp_col_min = x->mv_col_min; |
int tmp_col_max = x->mv_col_max; |
@@ -1353,18 +1359,18 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
x->mv_row_max = row_max; |
/* Get 8x8 result */ |
- bsi.sv_mvp[0].as_int = bsi.mvs[0].as_int; |
- bsi.sv_mvp[1].as_int = bsi.mvs[2].as_int; |
- bsi.sv_mvp[2].as_int = bsi.mvs[8].as_int; |
- bsi.sv_mvp[3].as_int = bsi.mvs[10].as_int; |
+ bsi.sv_mvp[0] = bsi.mvs[0].as_mv; |
+ bsi.sv_mvp[1] = bsi.mvs[2].as_mv; |
+ bsi.sv_mvp[2] = bsi.mvs[8].as_mv; |
+ bsi.sv_mvp[3] = bsi.mvs[10].as_mv; |
/* Use 8x8 result as 16x8/8x16's predictor MV. Adjust search range according to the closeness of 2 MV. */ |
/* block 8X16 */ |
{ |
- sr = MAXF((abs(bsi.sv_mvp[0].as_mv.row - bsi.sv_mvp[2].as_mv.row))>>3, (abs(bsi.sv_mvp[0].as_mv.col - bsi.sv_mvp[2].as_mv.col))>>3); |
+ sr = MAXF((abs(bsi.sv_mvp[0].row - bsi.sv_mvp[2].row))>>3, (abs(bsi.sv_mvp[0].col - bsi.sv_mvp[2].col))>>3); |
vp8_cal_step_param(sr, &bsi.sv_istep[0]); |
- sr = MAXF((abs(bsi.sv_mvp[1].as_mv.row - bsi.sv_mvp[3].as_mv.row))>>3, (abs(bsi.sv_mvp[1].as_mv.col - bsi.sv_mvp[3].as_mv.col))>>3); |
+ sr = MAXF((abs(bsi.sv_mvp[1].row - bsi.sv_mvp[3].row))>>3, (abs(bsi.sv_mvp[1].col - bsi.sv_mvp[3].col))>>3); |
vp8_cal_step_param(sr, &bsi.sv_istep[1]); |
rd_check_segment(cpi, x, &bsi, BLOCK_8X16); |
@@ -1372,10 +1378,10 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
/* block 16X8 */ |
{ |
- sr = MAXF((abs(bsi.sv_mvp[0].as_mv.row - bsi.sv_mvp[1].as_mv.row))>>3, (abs(bsi.sv_mvp[0].as_mv.col - bsi.sv_mvp[1].as_mv.col))>>3); |
+ sr = MAXF((abs(bsi.sv_mvp[0].row - bsi.sv_mvp[1].row))>>3, (abs(bsi.sv_mvp[0].col - bsi.sv_mvp[1].col))>>3); |
vp8_cal_step_param(sr, &bsi.sv_istep[0]); |
- sr = MAXF((abs(bsi.sv_mvp[2].as_mv.row - bsi.sv_mvp[3].as_mv.row))>>3, (abs(bsi.sv_mvp[2].as_mv.col - bsi.sv_mvp[3].as_mv.col))>>3); |
+ sr = MAXF((abs(bsi.sv_mvp[2].row - bsi.sv_mvp[3].row))>>3, (abs(bsi.sv_mvp[2].col - bsi.sv_mvp[3].col))>>3); |
vp8_cal_step_param(sr, &bsi.sv_istep[1]); |
rd_check_segment(cpi, x, &bsi, BLOCK_16X8); |
@@ -1385,7 +1391,7 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
/* Not skip 4x4 if speed=0 (good quality) */ |
if (cpi->sf.no_skip_block4x4_search || bsi.segment_num == BLOCK_8X8) /* || (sv_segment_rd8x8-bsi.segment_rd) < sv_segment_rd8x8>>5) */ |
{ |
- bsi.mvp.as_int = bsi.sv_mvp[0].as_int; |
+ bsi.mvp = &bsi.sv_mvp[0]; |
rd_check_segment(cpi, x, &bsi, BLOCK_4X4); |
} |
@@ -1402,7 +1408,8 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
{ |
BLOCKD *bd = &x->e_mbd.block[i]; |
- bd->bmi.mv.as_int = bsi.mvs[i].as_int; |
+ bd->bmi.mv.as_mv = bsi.mvs[i].as_mv; |
+ bd->bmi.mode = bsi.modes[i]; |
bd->eob = bsi.eobs[i]; |
} |
@@ -1420,66 +1427,94 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x, |
j = vp8_mbsplit_offset[bsi.segment_num][i]; |
- x->partition_info->bmi[i].mode = bsi.modes[j]; |
- x->partition_info->bmi[i].mv.as_mv = bsi.mvs[j].as_mv; |
+ x->partition_info->bmi[i].mode = x->e_mbd.block[j].bmi.mode; |
+ x->partition_info->bmi[i].mv.as_mv = x->e_mbd.block[j].bmi.mv.as_mv; |
} |
- /* |
- * used to set x->e_mbd.mode_info_context->mbmi.mv.as_int |
- */ |
- x->partition_info->bmi[15].mv.as_int = bsi.mvs[15].as_int; |
return bsi.segment_rd; |
} |
+#endif |
-static void insertsortmv(int arr[], int len) |
+static void swap(int *x,int *y) |
{ |
- int i, j, k; |
+ int tmp; |
- for ( i = 1 ; i <= len-1 ; i++ ) |
- { |
- for ( j = 0 ; j < i ; j++ ) |
- { |
- if ( arr[j] > arr[i] ) |
- { |
- int temp; |
- |
- temp = arr[i]; |
- |
- for ( k = i; k >j; k--) |
- arr[k] = arr[k - 1] ; |
- |
- arr[j] = temp ; |
- } |
- } |
- } |
+ tmp = *x; |
+ *x = *y; |
+ *y = tmp; |
} |
-static void insertsortsad(int arr[],int idx[], int len) |
+static void quicksortmv(int arr[],int left, int right) |
{ |
- int i, j, k; |
- |
- for ( i = 1 ; i <= len-1 ; i++ ) |
- { |
- for ( j = 0 ; j < i ; j++ ) |
- { |
- if ( arr[j] > arr[i] ) |
- { |
- int temp, tempi; |
- |
- temp = arr[i]; |
- tempi = idx[i]; |
- |
- for ( k = i; k >j; k--) |
- { |
- arr[k] = arr[k - 1] ; |
- idx[k] = idx[k - 1]; |
- } |
+ int lidx,ridx,pivot; |
+ |
+ lidx = left; |
+ ridx = right; |
+ |
+ if( left < right) |
+ { |
+ pivot = (left + right)/2; |
+ |
+ while(lidx <=pivot && ridx >=pivot) |
+ { |
+ while(arr[lidx] < arr[pivot] && lidx <= pivot) |
+ lidx++; |
+ while(arr[ridx] > arr[pivot] && ridx >= pivot) |
+ ridx--; |
+ swap(&arr[lidx], &arr[ridx]); |
+ lidx++; |
+ ridx--; |
+ if(lidx-1 == pivot) |
+ { |
+ ridx++; |
+ pivot = ridx; |
+ } |
+ else if(ridx+1 == pivot) |
+ { |
+ lidx--; |
+ pivot = lidx; |
+ } |
+ } |
+ quicksortmv(arr, left, pivot - 1); |
+ quicksortmv(arr, pivot + 1, right); |
+ } |
+} |
- arr[j] = temp ; |
- idx[j] = tempi; |
- } |
- } |
- } |
+static void quicksortsad(int arr[],int idx[], int left, int right) |
+{ |
+ int lidx,ridx,pivot; |
+ |
+ lidx = left; |
+ ridx = right; |
+ |
+ if( left < right) |
+ { |
+ pivot = (left + right)/2; |
+ |
+ while(lidx <=pivot && ridx >=pivot) |
+ { |
+ while(arr[lidx] < arr[pivot] && lidx <= pivot) |
+ lidx++; |
+ while(arr[ridx] > arr[pivot] && ridx >= pivot) |
+ ridx--; |
+ swap(&arr[lidx], &arr[ridx]); |
+ swap(&idx[lidx], &idx[ridx]); |
+ lidx++; |
+ ridx--; |
+ if(lidx-1 == pivot) |
+ { |
+ ridx++; |
+ pivot = ridx; |
+ } |
+ else if(ridx+1 == pivot) |
+ { |
+ lidx--; |
+ pivot = lidx; |
+ } |
+ } |
+ quicksortsad(arr, idx, left, pivot - 1); |
+ quicksortsad(arr, idx, pivot + 1, right); |
+ } |
} |
//The improved MV prediction |
@@ -1488,7 +1523,7 @@ void vp8_mv_pred |
VP8_COMP *cpi, |
MACROBLOCKD *xd, |
const MODE_INFO *here, |
- int_mv *mvp, |
+ MV *mvp, |
int refframe, |
int *ref_frame_sign_bias, |
int *sr, |
@@ -1615,8 +1650,8 @@ void vp8_mv_pred |
mvy[i] = near_mvs[i].as_mv.col; |
} |
- insertsortmv(mvx, vcnt); |
- insertsortmv(mvy, vcnt); |
+ quicksortmv (mvx, 0, vcnt-1); |
+ quicksortmv (mvy, 0, vcnt-1); |
mv.as_mv.row = mvx[vcnt/2]; |
mv.as_mv.col = mvy[vcnt/2]; |
@@ -1627,16 +1662,14 @@ void vp8_mv_pred |
} |
/* Set up return values */ |
- mvp->as_int = mv.as_int; |
- vp8_clamp_mv2(mvp, xd); |
+ *mvp = mv.as_mv; |
+ vp8_clamp_mv(mvp, xd); |
} |
void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffset, int near_sadidx[]) |
{ |
int near_sad[8] = {0}; // 0-cf above, 1-cf left, 2-cf aboveleft, 3-lf current, 4-lf above, 5-lf left, 6-lf right, 7-lf below |
- BLOCK *b = &x->block[0]; |
- unsigned char *src_y_ptr = *(b->base_src); |
//calculate sad for current frame 3 nearby MBs. |
if( xd->mb_to_top_edge==0 && xd->mb_to_left_edge ==0) |
@@ -1645,16 +1678,16 @@ void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffse |
}else if(xd->mb_to_top_edge==0) |
{ //only has left MB for sad calculation. |
near_sad[0] = near_sad[2] = INT_MAX; |
- near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff); |
+ near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff); |
}else if(xd->mb_to_left_edge ==0) |
{ //only has left MB for sad calculation. |
near_sad[1] = near_sad[2] = INT_MAX; |
- near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff); |
+ near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff); |
}else |
{ |
- near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff); |
- near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff); |
- near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, 0x7fffffff); |
+ near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff); |
+ near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff); |
+ near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, 0x7fffffff); |
} |
if(cpi->common.last_frame_type != KEY_FRAME) |
@@ -1669,61 +1702,36 @@ void vp8_cal_sad(VP8_COMP *cpi, MACROBLOCKD *xd, MACROBLOCK *x, int recon_yoffse |
if(xd->mb_to_bottom_edge==0) near_sad[7] = INT_MAX; |
if(near_sad[4] != INT_MAX) |
- near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, 0x7fffffff); |
+ near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, 0x7fffffff); |
if(near_sad[5] != INT_MAX) |
- near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - 16, pre_y_stride, 0x7fffffff); |
- near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer, pre_y_stride, 0x7fffffff); |
+ near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - 16, pre_y_stride, 0x7fffffff); |
+ near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer, pre_y_stride, 0x7fffffff); |
if(near_sad[6] != INT_MAX) |
- near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + 16, pre_y_stride, 0x7fffffff); |
+ near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + 16, pre_y_stride, 0x7fffffff); |
if(near_sad[7] != INT_MAX) |
- near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, 0x7fffffff); |
+ near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, 0x7fffffff); |
} |
if(cpi->common.last_frame_type != KEY_FRAME) |
{ |
- insertsortsad(near_sad, near_sadidx, 8); |
+ quicksortsad(near_sad, near_sadidx, 0, 7); |
}else |
{ |
- insertsortsad(near_sad, near_sadidx, 3); |
+ quicksortsad(near_sad, near_sadidx, 0, 2); |
} |
} |
-static void rd_update_mvcount(VP8_COMP *cpi, MACROBLOCK *x, int_mv *best_ref_mv) |
-{ |
- if (x->e_mbd.mode_info_context->mbmi.mode == SPLITMV) |
- { |
- int i; |
- |
- for (i = 0; i < x->partition_info->count; i++) |
- { |
- if (x->partition_info->bmi[i].mode == NEW4X4) |
- { |
- cpi->MVcount[0][mv_max+((x->partition_info->bmi[i].mv.as_mv.row |
- - best_ref_mv->as_mv.row) >> 1)]++; |
- cpi->MVcount[1][mv_max+((x->partition_info->bmi[i].mv.as_mv.col |
- - best_ref_mv->as_mv.col) >> 1)]++; |
- } |
- } |
- } |
- else if (x->e_mbd.mode_info_context->mbmi.mode == NEWMV) |
- { |
- cpi->MVcount[0][mv_max+((x->e_mbd.mode_info_context->mbmi.mv.as_mv.row |
- - best_ref_mv->as_mv.row) >> 1)]++; |
- cpi->MVcount[1][mv_max+((x->e_mbd.mode_info_context->mbmi.mv.as_mv.col |
- - best_ref_mv->as_mv.col) >> 1)]++; |
- } |
-} |
- |
-void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra) |
+#if !(CONFIG_REALTIME_ONLY) |
+int vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int recon_uvoffset, int *returnrate, int *returndistortion, int *returnintra) |
{ |
BLOCK *b = &x->block[0]; |
BLOCKD *d = &x->e_mbd.block[0]; |
MACROBLOCKD *xd = &x->e_mbd; |
- union b_mode_info best_bmodes[16]; |
+ B_MODE_INFO best_bmodes[16]; |
MB_MODE_INFO best_mbmode; |
PARTITION_INFO best_partition; |
- int_mv best_ref_mv; |
- int_mv mode_mv[MB_MODE_COUNT]; |
+ MV best_ref_mv; |
+ MV mode_mv[MB_MODE_COUNT]; |
MB_PREDICTION_MODE this_mode; |
int num00; |
int best_mode_index = 0; |
@@ -1733,8 +1741,8 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
int mdcounts[4]; |
int rate; |
int distortion; |
- int best_rd = INT_MAX; |
- int best_intra_rd = INT_MAX; |
+ int best_rd = INT_MAX; // 1 << 30; |
+ int ref_frame_cost[MAX_REF_FRAMES]; |
int rate2, distortion2; |
int uv_intra_rate, uv_intra_distortion, uv_intra_rate_tokenonly; |
int rate_y, UNINITIALIZED_IS_SAFE(rate_uv); |
@@ -1747,14 +1755,17 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
//int intermodecost[MAX_MODES]; |
MB_PREDICTION_MODE uv_intra_mode; |
- int_mv mvp; |
+ |
+ int force_no_skip = 0; |
+ |
+ MV mvp; |
int near_sadidx[8] = {0, 1, 2, 3, 4, 5, 6, 7}; |
int saddone=0; |
int sr=0; //search range got from mv_pred(). It uses step_param levels. (0-7) |
- int_mv frame_nearest_mv[4]; |
- int_mv frame_near_mv[4]; |
- int_mv frame_best_ref_mv[4]; |
+ MV frame_nearest_mv[4]; |
+ MV frame_near_mv[4]; |
+ MV frame_best_ref_mv[4]; |
int frame_mdcounts[4][4]; |
int frame_lf_or_gf[4]; |
unsigned char *y_buffer[4]; |
@@ -1762,7 +1773,6 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
unsigned char *v_buffer[4]; |
vpx_memset(&best_mbmode, 0, sizeof(best_mbmode)); |
- vpx_memset(&best_bmodes, 0, sizeof(best_bmodes)); |
if (cpi->ref_frame_flags & VP8_LAST_FLAG) |
{ |
@@ -1811,10 +1821,36 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
x->skip = 0; |
+ ref_frame_cost[INTRA_FRAME] = vp8_cost_zero(cpi->prob_intra_coded); |
+ |
+ // Special case treatment when GF and ARF are not sensible options for reference |
+ if (cpi->ref_frame_flags == VP8_LAST_FLAG) |
+ { |
+ ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded) |
+ + vp8_cost_zero(255); |
+ ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded) |
+ + vp8_cost_one(255) |
+ + vp8_cost_zero(128); |
+ ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded) |
+ + vp8_cost_one(255) |
+ + vp8_cost_one(128); |
+ } |
+ else |
+ { |
+ ref_frame_cost[LAST_FRAME] = vp8_cost_one(cpi->prob_intra_coded) |
+ + vp8_cost_zero(cpi->prob_last_coded); |
+ ref_frame_cost[GOLDEN_FRAME] = vp8_cost_one(cpi->prob_intra_coded) |
+ + vp8_cost_one(cpi->prob_last_coded) |
+ + vp8_cost_zero(cpi->prob_gf_coded); |
+ ref_frame_cost[ALTREF_FRAME] = vp8_cost_one(cpi->prob_intra_coded) |
+ + vp8_cost_one(cpi->prob_last_coded) |
+ + vp8_cost_one(cpi->prob_gf_coded); |
+ } |
+ |
vpx_memset(mode_mv, 0, sizeof(mode_mv)); |
x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; |
- rd_pick_intra_mbuv_mode(cpi, x, &uv_intra_rate, &uv_intra_rate_tokenonly, &uv_intra_distortion); |
+ vp8_rd_pick_intra_mbuv_mode(cpi, x, &uv_intra_rate, &uv_intra_rate_tokenonly, &uv_intra_distortion); |
uv_intra_mode = x->e_mbd.mode_info_context->mbmi.uv_mode; |
for (mode_index = 0; mode_index < MAX_MODES; mode_index++) |
@@ -1824,6 +1860,8 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
int disable_skip = 0; |
int other_cost = 0; |
+ force_no_skip = 0; |
+ |
// Experimental debug code. |
// Record of rd values recorded for this MB. -1 indicates not measured |
//all_rds[mode_index] = -1; |
@@ -1867,6 +1905,28 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
lf_or_gf = frame_lf_or_gf[x->e_mbd.mode_info_context->mbmi.ref_frame]; |
} |
+ if(x->e_mbd.mode_info_context->mbmi.mode == NEWMV) |
+ { |
+ if(!saddone) |
+ { |
+ vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] ); |
+ saddone = 1; |
+ } |
+ |
+ vp8_mv_pred(cpi, &x->e_mbd, x->e_mbd.mode_info_context, &mvp, |
+ x->e_mbd.mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias, &sr, &near_sadidx[0]); |
+ |
+ /* adjust mvp to make sure it is within MV range */ |
+ if(mvp.row > best_ref_mv.row + MAX_FULL_PEL_VAL) |
+ mvp.row = best_ref_mv.row + MAX_FULL_PEL_VAL; |
+ else if(mvp.row < best_ref_mv.row - MAX_FULL_PEL_VAL) |
+ mvp.row = best_ref_mv.row - MAX_FULL_PEL_VAL; |
+ if(mvp.col > best_ref_mv.col + MAX_FULL_PEL_VAL) |
+ mvp.col = best_ref_mv.col + MAX_FULL_PEL_VAL; |
+ else if(mvp.col < best_ref_mv.col - MAX_FULL_PEL_VAL) |
+ mvp.col = best_ref_mv.col - MAX_FULL_PEL_VAL; |
+ } |
+ |
// Check to see if the testing frequency for this mode is at its max |
// If so then prevent it from being tested and increase the threshold for its testing |
if (cpi->mode_test_hit_counts[mode_index] && (cpi->mode_check_freq[mode_index] > 1)) |
@@ -1918,7 +1978,7 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
int tmp_rd; |
// Note the rate value returned here includes the cost of coding the BPRED mode : x->mbmode_cost[x->e_mbd.frame_type][BPRED]; |
- tmp_rd = rd_pick_intra4x4mby_modes(cpi, x, &rate, &rate_y, &distortion, best_yrd); |
+ tmp_rd = vp8_rd_pick_intra4x4mby_modes(cpi, x, &rate, &rate_y, &distortion, best_yrd); |
rate2 += rate; |
distortion2 += distortion; |
@@ -1985,197 +2045,228 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
break; |
case NEWMV: |
- { |
- int thissme; |
- int bestsme = INT_MAX; |
- int step_param = cpi->sf.first_step; |
- int further_steps; |
- int n; |
- int do_refine=1; /* If last step (1-away) of n-step search doesn't pick the center point as the best match, |
- we will do a final 1-away diamond refining search */ |
- |
- int sadpb = x->sadperbit16; |
- int_mv mvp_full; |
- |
- int col_min = (best_ref_mv.as_mv.col>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv.as_mv.col & 7)?1:0); |
- int row_min = (best_ref_mv.as_mv.row>>3) - MAX_FULL_PEL_VAL + ((best_ref_mv.as_mv.row & 7)?1:0); |
- int col_max = (best_ref_mv.as_mv.col>>3) + MAX_FULL_PEL_VAL; |
- int row_max = (best_ref_mv.as_mv.row>>3) + MAX_FULL_PEL_VAL; |
- int tmp_col_min = x->mv_col_min; |
- int tmp_col_max = x->mv_col_max; |
- int tmp_row_min = x->mv_row_min; |
- int tmp_row_max = x->mv_row_max; |
+ // Decrement full search counter |
+ if (cpi->check_freq[lf_or_gf] > 0) |
+ cpi->check_freq[lf_or_gf] --; |
- if(!saddone) |
{ |
- vp8_cal_sad(cpi,xd,x, recon_yoffset ,&near_sadidx[0] ); |
- saddone = 1; |
- } |
+ int thissme; |
+ int bestsme = INT_MAX; |
+ int step_param = cpi->sf.first_step; |
+ int search_range; |
+ int further_steps; |
+ int n; |
- vp8_mv_pred(cpi, &x->e_mbd, x->e_mbd.mode_info_context, &mvp, |
- x->e_mbd.mode_info_context->mbmi.ref_frame, cpi->common.ref_frame_sign_bias, &sr, &near_sadidx[0]); |
+ int col_min = (best_ref_mv.col - MAX_FULL_PEL_VAL) >>3; |
+ int col_max = (best_ref_mv.col + MAX_FULL_PEL_VAL) >>3; |
+ int row_min = (best_ref_mv.row - MAX_FULL_PEL_VAL) >>3; |
+ int row_max = (best_ref_mv.row + MAX_FULL_PEL_VAL) >>3; |
+ |
+ int tmp_col_min = x->mv_col_min; |
+ int tmp_col_max = x->mv_col_max; |
+ int tmp_row_min = x->mv_row_min; |
+ int tmp_row_max = x->mv_row_max; |
+ |
+ // Get intersection of UMV window and valid MV window to reduce # of checks in diamond search. |
+ if (x->mv_col_min < col_min ) |
+ x->mv_col_min = col_min; |
+ if (x->mv_col_max > col_max ) |
+ x->mv_col_max = col_max; |
+ if (x->mv_row_min < row_min ) |
+ x->mv_row_min = row_min; |
+ if (x->mv_row_max > row_max ) |
+ x->mv_row_max = row_max; |
+ |
+ //adjust search range according to sr from mv prediction |
+ if(sr > step_param) |
+ step_param = sr; |
+ |
+ // Work out how long a search we should do |
+ search_range = MAXF(abs(best_ref_mv.col), abs(best_ref_mv.row)) >> 3; |
+ |
+ if (search_range >= x->vector_range) |
+ x->vector_range = search_range; |
+ else if (x->vector_range > cpi->sf.min_fs_radius) |
+ x->vector_range--; |
+ |
+ // Initial step/diamond search |
+ { |
+ int sadpb = x->sadperbit16; |
- mvp_full.as_mv.col = mvp.as_mv.col>>3; |
- mvp_full.as_mv.row = mvp.as_mv.row>>3; |
+ if (cpi->sf.search_method == HEX) |
+ { |
+ bestsme = vp8_hex_search(x, b, d, &best_ref_mv, &d->bmi.mv.as_mv, step_param, sadpb/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); |
+ mode_mv[NEWMV].row = d->bmi.mv.as_mv.row; |
+ mode_mv[NEWMV].col = d->bmi.mv.as_mv.col; |
+ } |
+ else |
+ { |
+ bestsme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param, sadpb / 2/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb < 9 |
+ mode_mv[NEWMV].row = d->bmi.mv.as_mv.row; |
+ mode_mv[NEWMV].col = d->bmi.mv.as_mv.col; |
- // Get intersection of UMV window and valid MV window to reduce # of checks in diamond search. |
- if (x->mv_col_min < col_min ) |
- x->mv_col_min = col_min; |
- if (x->mv_col_max > col_max ) |
- x->mv_col_max = col_max; |
- if (x->mv_row_min < row_min ) |
- x->mv_row_min = row_min; |
- if (x->mv_row_max > row_max ) |
- x->mv_row_max = row_max; |
+ // Further step/diamond searches as necessary |
+ n = 0; |
+ further_steps = (cpi->sf.max_step_search_steps - 1) - step_param; |
- //adjust search range according to sr from mv prediction |
- if(sr > step_param) |
- step_param = sr; |
+ n = num00; |
+ num00 = 0; |
- // Initial step/diamond search |
- { |
- bestsme = cpi->diamond_search_sad(x, b, d, &mvp_full, &d->bmi.mv, |
- step_param, sadpb, &num00, |
- &cpi->fn_ptr[BLOCK_16X16], |
- x->mvcost, &best_ref_mv); |
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int; |
+ while (n < further_steps) |
+ { |
+ n++; |
- // Further step/diamond searches as necessary |
- n = 0; |
- further_steps = (cpi->sf.max_step_search_steps - 1) - step_param; |
+ if (num00) |
+ num00--; |
+ else |
+ { |
+ thissme = cpi->diamond_search_sad(x, b, d, &mvp, &d->bmi.mv.as_mv, step_param + n, sadpb / 4/*x->errorperbit*/, &num00, &cpi->fn_ptr[BLOCK_16X16], x->mvsadcost, x->mvcost, &best_ref_mv); //sadpb = 9 |
- n = num00; |
- num00 = 0; |
+ if (thissme < bestsme) |
+ { |
+ bestsme = thissme; |
+ mode_mv[NEWMV].row = d->bmi.mv.as_mv.row; |
+ mode_mv[NEWMV].col = d->bmi.mv.as_mv.col; |
+ } |
+ else |
+ { |
+ d->bmi.mv.as_mv.row = mode_mv[NEWMV].row; |
+ d->bmi.mv.as_mv.col = mode_mv[NEWMV].col; |
+ } |
+ } |
+ } |
+ } |
- /* If there won't be more n-step search, check to see if refining search is needed. */ |
- if (n > further_steps) |
- do_refine = 0; |
+ } |
- while (n < further_steps) |
+ // Should we do a full search |
+ if (!cpi->check_freq[lf_or_gf] || cpi->do_full[lf_or_gf]) |
{ |
- n++; |
+ int thissme; |
+ int full_flag_thresh = 0; |
+ MV full_mvp; |
+ |
+ full_mvp.row = d->bmi.mv.as_mv.row <<3; // use diamond search result as full search staring point |
+ full_mvp.col = d->bmi.mv.as_mv.col <<3; |
+ |
+ // Update x->vector_range based on best vector found in step search |
+ search_range = MAXF(abs((mvp.row>>3) - d->bmi.mv.as_mv.row), abs((mvp.col>>3) - d->bmi.mv.as_mv.col)); |
+ //search_range *= 1.4; //didn't improve PSNR |
- if (num00) |
- num00--; |
+ if (search_range > x->vector_range) |
+ x->vector_range = search_range; |
else |
+ search_range = x->vector_range; |
+ |
+ // Apply limits |
+ search_range = (search_range > cpi->sf.max_fs_radius) ? cpi->sf.max_fs_radius : search_range; |
+ |
+ //add this to reduce full search range. |
+ if(sr<=3 && search_range > 8) search_range = 8; |
+ |
{ |
- thissme = cpi->diamond_search_sad(x, b, d, &mvp_full, |
- &d->bmi.mv, step_param + n, sadpb, &num00, |
- &cpi->fn_ptr[BLOCK_16X16], x->mvcost, |
- &best_ref_mv); |
+ int sadpb = x->sadperbit16 >> 2; |
+ thissme = cpi->full_search_sad(x, b, d, &full_mvp, sadpb, search_range, &cpi->fn_ptr[BLOCK_16X16], x->mvcost, x->mvsadcost,&best_ref_mv); |
+ } |
- /* check to see if refining search is needed. */ |
- if (num00 > (further_steps-n)) |
- do_refine = 0; |
+ // Barrier threshold to initiating full search |
+ // full_flag_thresh = 10 + (thissme >> 7); |
+ if ((thissme + full_flag_thresh) < bestsme) |
+ { |
+ cpi->do_full[lf_or_gf] ++; |
+ bestsme = thissme; |
+ } |
+ else if (thissme < bestsme) |
+ bestsme = thissme; |
+ else |
+ { |
+ cpi->do_full[lf_or_gf] = cpi->do_full[lf_or_gf] >> 1; |
+ cpi->check_freq[lf_or_gf] = cpi->sf.full_freq[lf_or_gf]; |
- if (thissme < bestsme) |
- { |
- bestsme = thissme; |
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int; |
- } |
- else |
- { |
- d->bmi.mv.as_int = mode_mv[NEWMV].as_int; |
- } |
+ // The full search result is actually worse so re-instate the previous best vector |
+ d->bmi.mv.as_mv.row = mode_mv[NEWMV].row; |
+ d->bmi.mv.as_mv.col = mode_mv[NEWMV].col; |
} |
} |
- } |
- |
- /* final 1-away diamond refining search */ |
- if (do_refine == 1) |
- { |
- int search_range; |
- //It seems not a good way to set search_range. Need further investigation. |
- //search_range = MAXF(abs((mvp.row>>3) - d->bmi.mv.as_mv.row), abs((mvp.col>>3) - d->bmi.mv.as_mv.col)); |
- search_range = 8; |
+ x->mv_col_min = tmp_col_min; |
+ x->mv_col_max = tmp_col_max; |
+ x->mv_row_min = tmp_row_min; |
+ x->mv_row_max = tmp_row_max; |
- //thissme = cpi->full_search_sad(x, b, d, &d->bmi.mv.as_mv, sadpb, search_range, &cpi->fn_ptr[BLOCK_16X16], x->mvcost, &best_ref_mv); |
- thissme = cpi->refining_search_sad(x, b, d, &d->bmi.mv, sadpb, |
- search_range, &cpi->fn_ptr[BLOCK_16X16], |
- x->mvcost, &best_ref_mv); |
+ if (bestsme < INT_MAX) |
+ // cpi->find_fractional_mv_step(x,b,d,&d->bmi.mv.as_mv,&best_ref_mv,x->errorperbit/2,cpi->fn_ptr.svf,cpi->fn_ptr.vf,x->mvcost); // normal mvc=11 |
+ cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv.as_mv, &best_ref_mv, x->errorperbit / 4, &cpi->fn_ptr[BLOCK_16X16], x->mvcost); |
- if (thissme < bestsme) |
- { |
- bestsme = thissme; |
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int; |
- } |
- else |
- { |
- d->bmi.mv.as_int = mode_mv[NEWMV].as_int; |
- } |
- } |
+ mode_mv[NEWMV].row = d->bmi.mv.as_mv.row; |
+ mode_mv[NEWMV].col = d->bmi.mv.as_mv.col; |
- x->mv_col_min = tmp_col_min; |
- x->mv_col_max = tmp_col_max; |
- x->mv_row_min = tmp_row_min; |
- x->mv_row_max = tmp_row_max; |
+ // Add the new motion vector cost to our rolling cost variable |
+ rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, x->mvcost, 96); |
- if (bestsme < INT_MAX) |
- { |
- int dis; /* TODO: use dis in distortion calculation later. */ |
- unsigned int sse; |
- cpi->find_fractional_mv_step(x, b, d, &d->bmi.mv, &best_ref_mv, |
- x->errorperbit, |
- &cpi->fn_ptr[BLOCK_16X16], |
- x->mvcost, &dis, &sse); |
} |
- mode_mv[NEWMV].as_int = d->bmi.mv.as_int; |
- |
- // Add the new motion vector cost to our rolling cost variable |
- rate2 += vp8_mv_bit_cost(&mode_mv[NEWMV], &best_ref_mv, x->mvcost, 96); |
- } |
- |
case NEARESTMV: |
case NEARMV: |
+ |
// Clip "next_nearest" so that it does not extend to far out of image |
- vp8_clamp_mv2(&mode_mv[this_mode], xd); |
+ if (mode_mv[this_mode].col < (xd->mb_to_left_edge - LEFT_TOP_MARGIN)) |
+ mode_mv[this_mode].col = xd->mb_to_left_edge - LEFT_TOP_MARGIN; |
+ else if (mode_mv[this_mode].col > xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN) |
+ mode_mv[this_mode].col = xd->mb_to_right_edge + RIGHT_BOTTOM_MARGIN; |
+ |
+ if (mode_mv[this_mode].row < (xd->mb_to_top_edge - LEFT_TOP_MARGIN)) |
+ mode_mv[this_mode].row = xd->mb_to_top_edge - LEFT_TOP_MARGIN; |
+ else if (mode_mv[this_mode].row > xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN) |
+ mode_mv[this_mode].row = xd->mb_to_bottom_edge + RIGHT_BOTTOM_MARGIN; |
// Do not bother proceeding if the vector (from newmv,nearest or near) is 0,0 as this should then be coded using the zeromv mode. |
- if (((this_mode == NEARMV) || (this_mode == NEARESTMV)) && (mode_mv[this_mode].as_int == 0)) |
+ if (((this_mode == NEARMV) || (this_mode == NEARESTMV)) && |
+ ((mode_mv[this_mode].row == 0) && (mode_mv[this_mode].col == 0))) |
continue; |
case ZEROMV: |
+ mv_selected: |
+ |
// Trap vectors that reach beyond the UMV borders |
// Note that ALL New MV, Nearest MV Near MV and Zero MV code drops through to this point |
// because of the lack of break statements in the previous two cases. |
- if (((mode_mv[this_mode].as_mv.row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].as_mv.row >> 3) > x->mv_row_max) || |
- ((mode_mv[this_mode].as_mv.col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].as_mv.col >> 3) > x->mv_col_max)) |
+ if (((mode_mv[this_mode].row >> 3) < x->mv_row_min) || ((mode_mv[this_mode].row >> 3) > x->mv_row_max) || |
+ ((mode_mv[this_mode].col >> 3) < x->mv_col_min) || ((mode_mv[this_mode].col >> 3) > x->mv_col_max)) |
continue; |
vp8_set_mbmode_and_mvs(x, this_mode, &mode_mv[this_mode]); |
- vp8_build_inter16x16_predictors_mby(&x->e_mbd); |
+ vp8_build_inter_predictors_mby(&x->e_mbd); |
if (cpi->active_map_enabled && x->active_ptr[0] == 0) { |
x->skip = 1; |
} |
else if (x->encode_breakout) |
{ |
- unsigned int sse; |
- unsigned int var; |
+ int sum, sse; |
int threshold = (xd->block[0].dequant[1] |
* xd->block[0].dequant[1] >>4); |
if(threshold < x->encode_breakout) |
threshold = x->encode_breakout; |
- var = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x16) |
- (*(b->base_src), b->src_stride, |
- x->e_mbd.predictor, 16, &sse); |
+ VARIANCE_INVOKE(&cpi->rtcd.variance, get16x16var) |
+ (x->src.y_buffer, x->src.y_stride, |
+ x->e_mbd.predictor, 16, (unsigned int *)(&sse), &sum); |
if (sse < threshold) |
{ |
- unsigned int q2dc = xd->block[24].dequant[0]; |
+ // Check u and v to make sure skip is ok |
+ int sse2 = 0; |
/* If theres is no codeable 2nd order dc |
or a very small uniform pixel change change */ |
- if ((sse - var < q2dc * q2dc >>4) || |
- (sse /2 > var && sse-var < 64)) |
+ if (abs(sum) < (xd->block[24].dequant[0]<<2)|| |
+ ((sum * sum>>8) > sse && abs(sum) <128)) |
{ |
- // Check u and v to make sure skip is ok |
- int sse2= VP8_UVSSE(x, IF_RTCD(&cpi->rtcd.variance)); |
+ sse2 = VP8_UVSSE(x, IF_RTCD(&cpi->rtcd.variance)); |
+ |
if (sse2 * 2 < threshold) |
{ |
x->skip = 1; |
@@ -2187,7 +2278,8 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
distortion_uv = sse2; |
disable_skip = 1; |
- this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2); |
+ this_rd = RDCOST(x->rdmult, x->rddiv, rate2, |
+ distortion2); |
break; |
} |
@@ -2225,11 +2317,8 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
rate2 += other_cost; |
} |
- /* Estimate the reference frame signaling cost and add it |
- * to the rolling cost variable. |
- */ |
- rate2 += |
- x->e_mbd.ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame]; |
+ // Estimate the reference frame signaling cost and add it to the rolling cost variable. |
+ rate2 += ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame]; |
if (!disable_skip) |
{ |
@@ -2272,12 +2361,9 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
//all_rates[mode_index] = rate2; |
//all_dist[mode_index] = distortion2; |
- // Keep record of best intra distortion |
- if ((x->e_mbd.mode_info_context->mbmi.ref_frame == INTRA_FRAME) && |
- (this_rd < best_intra_rd) ) |
+ if ((x->e_mbd.mode_info_context->mbmi.ref_frame == INTRA_FRAME) && (this_rd < *returnintra)) |
{ |
- best_intra_rd = this_rd; |
- *returnintra = distortion2 ; |
+ *returnintra = this_rd ; |
} |
// Did this mode help.. i.i is it the new best mode |
@@ -2285,16 +2371,14 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
{ |
// Note index of best mode so far |
best_mode_index = mode_index; |
+ x->e_mbd.mode_info_context->mbmi.force_no_skip = force_no_skip; |
if (this_mode <= B_PRED) |
{ |
x->e_mbd.mode_info_context->mbmi.uv_mode = uv_intra_mode; |
- /* required for left and above block mv */ |
- x->e_mbd.mode_info_context->mbmi.mv.as_int = 0; |
} |
- other_cost += |
- x->e_mbd.ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame]; |
+ other_cost += ref_frame_cost[x->e_mbd.mode_info_context->mbmi.ref_frame]; |
/* Calculate the final y RD estimate for this mode */ |
best_yrd = RDCOST(x->rdmult, x->rddiv, (rate2-rate_uv-other_cost), |
@@ -2306,12 +2390,10 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
vpx_memcpy(&best_mbmode, &x->e_mbd.mode_info_context->mbmi, sizeof(MB_MODE_INFO)); |
vpx_memcpy(&best_partition, x->partition_info, sizeof(PARTITION_INFO)); |
- if ((this_mode == B_PRED) || (this_mode == SPLITMV)) |
- for (i = 0; i < 16; i++) |
- { |
- best_bmodes[i] = x->e_mbd.block[i].bmi; |
- } |
- |
+ for (i = 0; i < 16; i++) |
+ { |
+ vpx_memcpy(&best_bmodes[i], &x->e_mbd.block[i].bmi, sizeof(B_MODE_INFO)); |
+ } |
// Testing this mode gave rise to an improvement in best error score. Lower threshold a bit for next time |
cpi->rd_thresh_mult[mode_index] = (cpi->rd_thresh_mult[mode_index] >= (MIN_THRESHMULT + 2)) ? cpi->rd_thresh_mult[mode_index] - 2 : MIN_THRESHMULT; |
@@ -2360,81 +2442,68 @@ void vp8_rd_pick_inter_mode(VP8_COMP *cpi, MACROBLOCK *x, int recon_yoffset, int |
} |
- // Note how often each mode chosen as best |
- cpi->mode_chosen_counts[best_mode_index] ++; |
- |
- |
- if (cpi->is_src_frame_alt_ref && |
- (best_mbmode.mode != ZEROMV || best_mbmode.ref_frame != ALTREF_FRAME)) |
+ // If we have chosen new mv or split then decay the full search check count more quickly. |
+ if ((vp8_mode_order[best_mode_index] == NEWMV) || (vp8_mode_order[best_mode_index] == SPLITMV)) |
{ |
- x->e_mbd.mode_info_context->mbmi.mode = ZEROMV; |
- x->e_mbd.mode_info_context->mbmi.ref_frame = ALTREF_FRAME; |
- x->e_mbd.mode_info_context->mbmi.mv.as_int = 0; |
- x->e_mbd.mode_info_context->mbmi.uv_mode = DC_PRED; |
- x->e_mbd.mode_info_context->mbmi.mb_skip_coeff = |
- (cpi->common.mb_no_coeff_skip) ? 1 : 0; |
- x->e_mbd.mode_info_context->mbmi.partitioning = 0; |
+ int lf_or_gf = (vp8_ref_frame_order[best_mode_index] == LAST_FRAME) ? 0 : 1; |
- return; |
+ if (cpi->check_freq[lf_or_gf] && !cpi->do_full[lf_or_gf]) |
+ { |
+ cpi->check_freq[lf_or_gf] --; |
+ } |
} |
+ // Keep a record of best mode index that we chose |
+ cpi->last_best_mode_index = best_mode_index; |
- // macroblock modes |
- vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); |
+ // Note how often each mode chosen as best |
+ cpi->mode_chosen_counts[best_mode_index] ++; |
- if (best_mbmode.mode == B_PRED) |
- { |
- for (i = 0; i < 16; i++) |
- x->e_mbd.block[i].bmi.as_mode = best_bmodes[i].as_mode; |
- } |
- if (best_mbmode.mode == SPLITMV) |
+ if (cpi->is_src_frame_alt_ref && (best_mbmode.mode != ZEROMV || best_mbmode.ref_frame != ALTREF_FRAME)) |
{ |
- for (i = 0; i < 16; i++) |
- x->e_mbd.block[i].bmi.mv.as_int = best_bmodes[i].mv.as_int; |
- |
+ best_mbmode.mode = ZEROMV; |
+ best_mbmode.ref_frame = ALTREF_FRAME; |
+ best_mbmode.mv.as_int = 0; |
+ best_mbmode.uv_mode = 0; |
+ best_mbmode.mb_skip_coeff = (cpi->common.mb_no_coeff_skip) ? 1 : 0; |
+ best_mbmode.partitioning = 0; |
+ best_mbmode.dc_diff = 0; |
+ |
+ vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); |
vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO)); |
- x->e_mbd.mode_info_context->mbmi.mv.as_int = |
- x->partition_info->bmi[15].mv.as_int; |
- } |
- |
- rd_update_mvcount(cpi, x, &frame_best_ref_mv[xd->mode_info_context->mbmi.ref_frame]); |
- |
-} |
- |
-void vp8_rd_pick_intra_mode(VP8_COMP *cpi, MACROBLOCK *x, int *rate_) |
-{ |
- int error4x4, error16x16; |
- int rate4x4, rate16x16 = 0, rateuv; |
- int dist4x4, dist16x16, distuv; |
- int rate; |
- int rate4x4_tokenonly = 0; |
- int rate16x16_tokenonly = 0; |
- int rateuv_tokenonly = 0; |
- |
- x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME; |
+ for (i = 0; i < 16; i++) |
+ { |
+ vpx_memset(&x->e_mbd.block[i].bmi, 0, sizeof(B_MODE_INFO)); |
+ } |
- rd_pick_intra_mbuv_mode(cpi, x, &rateuv, &rateuv_tokenonly, &distuv); |
- rate = rateuv; |
+ x->e_mbd.mode_info_context->mbmi.mv.as_int = 0; |
- error16x16 = rd_pick_intra16x16mby_mode(cpi, x, |
- &rate16x16, &rate16x16_tokenonly, |
- &dist16x16); |
+ return best_rd; |
+ } |
- error4x4 = rd_pick_intra4x4mby_modes(cpi, x, |
- &rate4x4, &rate4x4_tokenonly, |
- &dist4x4, error16x16); |
- if (error4x4 < error16x16) |
+ if(best_mbmode.mode <= B_PRED) |
{ |
- x->e_mbd.mode_info_context->mbmi.mode = B_PRED; |
- rate += rate4x4; |
+ int i; |
+ for (i = 0; i < 16; i++) |
+ { |
+ best_bmodes[i].mv.as_int = 0; |
+ } |
} |
- else |
+ |
+ // macroblock modes |
+ vpx_memcpy(&x->e_mbd.mode_info_context->mbmi, &best_mbmode, sizeof(MB_MODE_INFO)); |
+ vpx_memcpy(x->partition_info, &best_partition, sizeof(PARTITION_INFO)); |
+ |
+ for (i = 0; i < 16; i++) |
{ |
- rate += rate16x16; |
+ vpx_memcpy(&x->e_mbd.block[i].bmi, &best_bmodes[i], sizeof(B_MODE_INFO)); |
} |
- *rate_ = rate; |
+ x->e_mbd.mode_info_context->mbmi.mv.as_mv = x->e_mbd.block[15].bmi.mv.as_mv; |
+ |
+ return best_rd; |
} |
+#endif |