generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

hybrid transform of 16x16 dimension

Browse Source 
Enable ADST/DCT of dimension 16x16 for I16X16 modes. This change provides
benefits mostly for hd sequences.

Set up the framework for selectable transform dimension.

Also allowing quantization parameter threshold to control the use
of hybrid transform (This is currently disabled by setting threshold
always above the quantization parameter. Adaptive thresholding can
be built upon this, which will further improve the coding performance.)

The coding performance gains (with respect to the codec that has all
other configuration settings turned on) are

derf:   0.013
yt:     0.086
hd:     0.198
std-hd: 0.501

Change-Id: Ibb4263a61fc74e0b3c345f54d73e8c73552bf926
This commit is contained in:
Jingning Han
2012-08-29 11:25:38 -07:00
parent c59e36fc76
commit de6dfa6bb0
38 changed files with 600 additions and 255 deletions
Download Patch File Download Diff File Expand all files Collapse all files

47
vp8/decoder/decodframe.c
View File

@@ -125,7 +125,7 @@ void mb_init_dequantizer(VP8D_COMP *pbi, MACROBLOCKD *xd) {
xd->block[i].dequant = pc->Y1dequant[QIndex];
}
#if CONFIG_HYBRIDTRANSFORM
#if CONFIG_HYBRIDTRANSFORM || CONFIG_HYBRIDTRANSFORM8X8 || CONFIG_HYBRIDTRANSFORM16X16
xd->q_index = QIndex;
#endif
@@ -234,17 +234,21 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
int orig_skip_flag = xd->mode_info_context->mbmi.mb_skip_coeff;
#endif
#if CONFIG_HYBRIDTRANSFORM
#if CONFIG_HYBRIDTRANSFORM || CONFIG_HYBRIDTRANSFORM16X16
int QIndex;
int active_ht;
#endif
#if CONFIG_HYBRIDTRANSFORM16X16
int active_ht16;
#endif
// re-initialize macroblock dequantizer before detokenization
if (xd->segmentation_enabled)
mb_init_dequantizer(pbi, xd);
if (pbi->common.frame_type == KEY_FRAME) {
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
if (xd->mode_info_context->mbmi.mode <= TM_PRED ||
xd->mode_info_context->mbmi.mode == NEWMV ||
xd->mode_info_context->mbmi.mode == ZEROMV ||
@@ -263,7 +267,7 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
else
xd->mode_info_context->mbmi.txfm_size = TX_4X4;
} else {
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
if (xd->mode_info_context->mbmi.mode <= TM_PRED ||
xd->mode_info_context->mbmi.mode == NEWMV ||
xd->mode_info_context->mbmi.mode == ZEROMV ||
@@ -316,7 +320,7 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
xd->block[i].eob = 0;
xd->eobs[i] = 0;
}
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
if (tx_type == TX_16X16)
eobtotal = vp8_decode_mb_tokens_16x16(pbi, xd);
else
@@ -377,7 +381,7 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
// if (xd->segmentation_enabled)
// mb_init_dequantizer(pbi, xd);
#if CONFIG_HYBRIDTRANSFORM
#if CONFIG_HYBRIDTRANSFORM || CONFIG_HYBRIDTRANSFORM16X16
// parse transform types for intra 4x4 mode
QIndex = xd->q_index;
active_ht = (QIndex < ACTIVE_HT);
@@ -391,6 +395,10 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
}
#endif
#if CONFIG_HYBRIDTRANSFORM16X16
active_ht16 = (QIndex < ACTIVE_HT16);
#endif
/* do prediction */
if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) {
#if CONFIG_SUPERBLOCKS
@@ -537,11 +545,28 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
} else {
BLOCKD *b = &xd->block[24];
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
if (tx_type == TX_16X16) {
#if CONFIG_HYBRIDTRANSFORM16X16
if (mode < I8X8_PRED && active_ht16) {
BLOCKD *bd = &xd->block[0];
TX_TYPE txfm;
txfm_map(bd, pred_mode_conv(mode));
txfm = bd->bmi.as_mode.tx_type;
vp8_ht_dequant_idct_add_16x16_c(txfm, xd->qcoeff,
xd->block[0].dequant, xd->predictor,
xd->dst.y_buffer, 16, xd->dst.y_stride);
} else {
vp8_dequant_idct_add_16x16_c(xd->qcoeff, xd->block[0].dequant,
xd->predictor, xd->dst.y_buffer,
16, xd->dst.y_stride);
}
#else
vp8_dequant_idct_add_16x16_c(xd->qcoeff, xd->block[0].dequant,
xd->predictor, xd->dst.y_buffer,
16, xd->dst.y_stride);
#endif
}
else
#endif
@@ -641,7 +666,7 @@ static void decode_macroblock(VP8D_COMP *pbi, MACROBLOCKD *xd,
if (!xd->mode_info_context->mbmi.encoded_as_sb) {
#endif
if (tx_type == TX_8X8
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
|| tx_type == TX_16X16
#endif
)
@@ -1047,7 +1072,7 @@ static void read_coef_probs(VP8D_COMP *pbi) {
}
}
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
// 16x16
if (vp8_read_bit(bc)) {
// read coef probability tree
@@ -1430,7 +1455,7 @@ int vp8_decode_frame(VP8D_COMP *pbi) {
vp8_copy(pbi->common.fc.pre_coef_probs, pbi->common.fc.coef_probs);
vp8_copy(pbi->common.fc.pre_coef_probs_8x8, pbi->common.fc.coef_probs_8x8);
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
vp8_copy(pbi->common.fc.pre_coef_probs_16x16, pbi->common.fc.coef_probs_16x16);
#endif
vp8_copy(pbi->common.fc.pre_ymode_prob, pbi->common.fc.ymode_prob);
@@ -1443,7 +1468,7 @@ int vp8_decode_frame(VP8D_COMP *pbi) {
vp8_copy(pbi->common.fc.pre_mvc_hp, pbi->common.fc.mvc_hp);
vp8_zero(pbi->common.fc.coef_counts);
vp8_zero(pbi->common.fc.coef_counts_8x8);
#if CONFIG_TX16X16
#if CONFIG_TX16X16 || CONFIG_HYBRIDTRANSFORM16X16
vp8_zero(pbi->common.fc.coef_counts_16x16);
#endif
vp8_zero(pbi->common.fc.ymode_counts);