2010-05-18 17:58:33 +02:00
|
|
|
/*
|
2010-09-09 14:16:39 +02:00
|
|
|
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
|
2010-05-18 17:58:33 +02:00
|
|
|
*
|
2010-06-09 17:29:20 +02:00
|
|
|
* Use of this source code is governed by a BSD-style license
|
2010-06-04 22:19:40 +02:00
|
|
|
* that can be found in the LICENSE file in the root of the source
|
|
|
|
* tree. An additional intellectual property rights grant can be found
|
2010-06-09 17:29:20 +02:00
|
|
|
* in the file PATENTS. All contributing project authors may
|
2010-06-04 22:19:40 +02:00
|
|
|
* be found in the AUTHORS file in the root of the source tree.
|
2010-05-18 17:58:33 +02:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
2012-11-28 19:41:40 +01:00
|
|
|
#include "vp9/decoder/vp9_onyxd_int.h"
|
2013-01-06 03:20:25 +01:00
|
|
|
#include "vp9/common/vp9_common.h"
|
2012-11-27 22:59:17 +01:00
|
|
|
#include "vp9/common/vp9_header.h"
|
|
|
|
#include "vp9/common/vp9_reconintra.h"
|
|
|
|
#include "vp9/common/vp9_reconintra4x4.h"
|
|
|
|
#include "vp9/common/vp9_reconinter.h"
|
2012-11-29 00:15:51 +01:00
|
|
|
#include "vp9/common/vp9_entropy.h"
|
2012-11-27 22:59:17 +01:00
|
|
|
#include "vp9/decoder/vp9_decodframe.h"
|
2012-11-28 19:41:40 +01:00
|
|
|
#include "vp9/decoder/vp9_detokenize.h"
|
2012-11-27 22:59:17 +01:00
|
|
|
#include "vp9/common/vp9_invtrans.h"
|
|
|
|
#include "vp9/common/vp9_alloccommon.h"
|
|
|
|
#include "vp9/common/vp9_entropymode.h"
|
|
|
|
#include "vp9/common/vp9_quant_common.h"
|
2012-12-03 23:19:49 +01:00
|
|
|
#include "vpx_scale/vpx_scale.h"
|
2012-11-27 22:59:17 +01:00
|
|
|
#include "vp9/common/vp9_setupintrarecon.h"
|
2010-09-09 20:42:48 +02:00
|
|
|
|
2012-11-28 19:41:40 +01:00
|
|
|
#include "vp9/decoder/vp9_decodemv.h"
|
2012-11-27 22:59:17 +01:00
|
|
|
#include "vp9/common/vp9_extend.h"
|
|
|
|
#include "vp9/common/vp9_modecont.h"
|
2010-05-18 17:58:33 +02:00
|
|
|
#include "vpx_mem/vpx_mem.h"
|
2012-11-28 19:41:40 +01:00
|
|
|
#include "vp9/decoder/vp9_dboolhuff.h"
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-11-27 22:59:17 +01:00
|
|
|
#include "vp9/common/vp9_seg_common.h"
|
2012-11-09 02:09:30 +01:00
|
|
|
#include "vp9_rtcd.h"
|
2011-10-05 12:26:00 +02:00
|
|
|
|
2010-05-18 17:58:33 +02:00
|
|
|
#include <assert.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
|
2012-06-06 00:25:07 +02:00
|
|
|
#define COEFCOUNT_TESTING
|
|
|
|
|
2012-12-14 08:53:11 +01:00
|
|
|
//#define DEC_DEBUG
|
2012-11-16 00:14:38 +01:00
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
int dec_debug = 0;
|
|
|
|
#endif
|
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
static int merge_index(int v, int n, int modulus) {
|
|
|
|
int max1 = (n - 1 - modulus / 2) / modulus + 1;
|
|
|
|
if (v < max1) v = v * modulus + modulus / 2;
|
|
|
|
else {
|
|
|
|
int w;
|
|
|
|
v -= max1;
|
|
|
|
w = v;
|
|
|
|
v += (v + modulus - modulus / 2) / modulus;
|
|
|
|
while (v % modulus == modulus / 2 ||
|
|
|
|
w != v - (v + modulus - modulus / 2) / modulus) v++;
|
|
|
|
}
|
|
|
|
return v;
|
2012-05-03 11:22:26 +02:00
|
|
|
}
|
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
static int inv_remap_prob(int v, int m) {
|
|
|
|
const int n = 256;
|
|
|
|
const int modulus = MODULUS_PARAM;
|
2012-08-07 00:05:24 +02:00
|
|
|
int i;
|
2012-07-14 00:21:29 +02:00
|
|
|
v = merge_index(v, n - 1, modulus);
|
|
|
|
if ((m << 1) <= n) {
|
2012-10-30 05:42:10 +01:00
|
|
|
i = vp9_inv_recenter_nonneg(v + 1, m);
|
2012-07-14 00:21:29 +02:00
|
|
|
} else {
|
2012-10-30 05:42:10 +01:00
|
|
|
i = n - 1 - vp9_inv_recenter_nonneg(v + 1, n - 1 - m);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
|
|
|
return i;
|
2012-04-12 18:24:03 +02:00
|
|
|
}
|
2012-05-03 11:22:26 +02:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
static vp9_prob read_prob_diff_update(vp9_reader *const bc, int oldp) {
|
2012-10-30 22:51:31 +01:00
|
|
|
int delp = vp9_decode_term_subexp(bc, SUBEXP_PARAM, 255);
|
2012-10-31 22:40:53 +01:00
|
|
|
return (vp9_prob)inv_remap_prob(delp, oldp);
|
2012-05-03 11:22:26 +02:00
|
|
|
}
|
2012-04-12 18:24:03 +02:00
|
|
|
|
2012-10-31 01:53:32 +01:00
|
|
|
void vp9_init_de_quantizer(VP9D_COMP *pbi) {
|
2012-07-14 00:21:29 +02:00
|
|
|
int i;
|
|
|
|
int Q;
|
2012-10-31 01:53:32 +01:00
|
|
|
VP9_COMMON *const pc = &pbi->common;
|
2012-07-14 00:21:29 +02:00
|
|
|
|
|
|
|
for (Q = 0; Q < QINDEX_RANGE; Q++) {
|
2012-12-19 00:31:19 +01:00
|
|
|
pc->Y1dequant[Q][0] = (int16_t)vp9_dc_quant(Q, pc->y1dc_delta_q);
|
|
|
|
pc->Y2dequant[Q][0] = (int16_t)vp9_dc2quant(Q, pc->y2dc_delta_q);
|
|
|
|
pc->UVdequant[Q][0] = (int16_t)vp9_dc_uv_quant(Q, pc->uvdc_delta_q);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
|
|
|
/* all the ac values =; */
|
|
|
|
for (i = 1; i < 16; i++) {
|
2012-12-18 19:49:10 +01:00
|
|
|
int rc = vp9_default_zig_zag1d_4x4[i];
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2012-12-19 00:31:19 +01:00
|
|
|
pc->Y1dequant[Q][rc] = (int16_t)vp9_ac_yquant(Q);
|
|
|
|
pc->Y2dequant[Q][rc] = (int16_t)vp9_ac2quant(Q, pc->y2ac_delta_q);
|
|
|
|
pc->UVdequant[Q][rc] = (int16_t)vp9_ac_uv_quant(Q, pc->uvac_delta_q);
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
|
|
|
|
2012-10-31 01:53:32 +01:00
|
|
|
static void mb_init_dequantizer(VP9D_COMP *pbi, MACROBLOCKD *xd) {
|
2012-07-14 00:21:29 +02:00
|
|
|
int i;
|
|
|
|
int QIndex;
|
2012-10-31 01:53:32 +01:00
|
|
|
VP9_COMMON *const pc = &pbi->common;
|
2012-07-14 00:21:29 +02:00
|
|
|
int segment_id = xd->mode_info_context->mbmi.segment_id;
|
|
|
|
|
|
|
|
// Set the Q baseline allowing for any segment level adjustment
|
2012-10-30 06:15:27 +01:00
|
|
|
if (vp9_segfeature_active(xd, segment_id, SEG_LVL_ALT_Q)) {
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Abs Value */
|
|
|
|
if (xd->mb_segment_abs_delta == SEGMENT_ABSDATA)
|
2012-10-30 06:15:27 +01:00
|
|
|
QIndex = vp9_get_segdata(xd, segment_id, SEG_LVL_ALT_Q);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
|
|
|
/* Delta Value */
|
|
|
|
else {
|
|
|
|
QIndex = pc->base_qindex +
|
2012-10-30 06:15:27 +01:00
|
|
|
vp9_get_segdata(xd, segment_id, SEG_LVL_ALT_Q);
|
2012-07-14 00:21:29 +02:00
|
|
|
QIndex = (QIndex >= 0) ? ((QIndex <= MAXQ) ? QIndex : MAXQ) : 0; /* Clamp to valid range */
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
} else
|
|
|
|
QIndex = pc->base_qindex;
|
2012-10-16 01:41:41 +02:00
|
|
|
xd->q_index = QIndex;
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Set up the block level dequant pointers */
|
|
|
|
for (i = 0; i < 16; i++) {
|
|
|
|
xd->block[i].dequant = pc->Y1dequant[QIndex];
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
Add lossless compression mode.
This commit adds lossless compression capability to the experimental
branch. The lossless experiment can be enabled using --enable-lossless
in configure. When the experiment is enabled, the encoder will use
lossless compression mode by command line option --lossless, and the
decoder automatically recognizes a losslessly encoded clip and decodes
accordingly.
To achieve the lossless coding, this commit has changed the following:
1. To encode at lossless mode, encoder forces the use of unit
quantizer, i.e, Q 0, where effective quantization is 1. Encoder also
disables the usage of 8x8 transform and allows only 4x4 transform;
2. At Q 0, the first order 4x4 DCT/IDCT have been switched over
to a pair of forward and inverse Walsh-Hadamard Transform
(http://goo.gl/EIsfy), with proper scaling applied to match the range
of the original 4x4 DCT/IDCT pair;
3. At Q 0, the second order remains to use the previous
walsh-hadamard transform pair. However, to maintain the reversibility
in second order transform at Q 0, scaling down is applied to first
order DC coefficients prior to forward transform, and scaling up is
applied to the second order output prior to quantization. Symmetric
upscaling and downscaling are added around inverse second order
transform;
4. At lossless mode, encoder also disables a number of minor
features to ensure no loss is introduced, these features includes:
a. Trellis quantization optimization
b. Loop filtering
c. Aggressive zero-binning, rounding and zero-bin boosting
d. Mode based zero-bin boosting
Lossless coding test was performed on all clips within the derf set,
to verify that the commit has achieved lossless compression for all
clips. The average compression ratio is around 2.57 to 1.
(http://goo.gl/dEShs)
Change-Id: Ia3aba7dd09df40dd590f93b9aba134defbc64e34
2012-06-14 04:03:31 +02:00
|
|
|
#if CONFIG_LOSSLESS
|
2012-07-14 00:21:29 +02:00
|
|
|
if (!QIndex) {
|
2012-11-25 04:33:58 +01:00
|
|
|
pbi->mb.inv_xform4x4_1_x8 = vp9_short_inv_walsh4x4_1_x8;
|
|
|
|
pbi->mb.inv_xform4x4_x8 = vp9_short_inv_walsh4x4_x8;
|
|
|
|
pbi->mb.inv_walsh4x4_1 = vp9_short_inv_walsh4x4_1_lossless;
|
|
|
|
pbi->mb.inv_walsh4x4_lossless = vp9_short_inv_walsh4x4_lossless;
|
2012-10-30 22:51:31 +01:00
|
|
|
pbi->idct_add = vp9_dequant_idct_add_lossless_c;
|
|
|
|
pbi->dc_idct_add = vp9_dequant_dc_idct_add_lossless_c;
|
|
|
|
pbi->dc_idct_add_y_block = vp9_dequant_dc_idct_add_y_block_lossless_c;
|
|
|
|
pbi->idct_add_y_block = vp9_dequant_idct_add_y_block_lossless_c;
|
|
|
|
pbi->idct_add_uv_block = vp9_dequant_idct_add_uv_block_lossless_c;
|
2012-07-14 00:21:29 +02:00
|
|
|
} else {
|
2012-11-25 04:33:58 +01:00
|
|
|
pbi->mb.inv_xform4x4_1_x8 = vp9_short_idct4x4llm_1;
|
|
|
|
pbi->mb.inv_xform4x4_x8 = vp9_short_idct4x4llm;
|
|
|
|
pbi->mb.inv_walsh4x4_1 = vp9_short_inv_walsh4x4_1;
|
|
|
|
pbi->mb.inv_walsh4x4_lossless = vp9_short_inv_walsh4x4;
|
2012-10-30 22:51:31 +01:00
|
|
|
pbi->idct_add = vp9_dequant_idct_add;
|
|
|
|
pbi->dc_idct_add = vp9_dequant_dc_idct_add;
|
|
|
|
pbi->dc_idct_add_y_block = vp9_dequant_dc_idct_add_y_block;
|
|
|
|
pbi->idct_add_y_block = vp9_dequant_idct_add_y_block;
|
|
|
|
pbi->idct_add_uv_block = vp9_dequant_idct_add_uv_block;
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2012-10-30 20:09:49 +01:00
|
|
|
#else
|
2012-11-25 04:33:58 +01:00
|
|
|
pbi->mb.inv_xform4x4_1_x8 = vp9_short_idct4x4llm_1;
|
|
|
|
pbi->mb.inv_xform4x4_x8 = vp9_short_idct4x4llm;
|
|
|
|
pbi->mb.inv_walsh4x4_1 = vp9_short_inv_walsh4x4_1;
|
|
|
|
pbi->mb.inv_walsh4x4_lossless = vp9_short_inv_walsh4x4;
|
2012-10-30 22:51:31 +01:00
|
|
|
pbi->idct_add = vp9_dequant_idct_add;
|
|
|
|
pbi->dc_idct_add = vp9_dequant_dc_idct_add;
|
|
|
|
pbi->dc_idct_add_y_block = vp9_dequant_dc_idct_add_y_block;
|
|
|
|
pbi->idct_add_y_block = vp9_dequant_idct_add_y_block;
|
|
|
|
pbi->idct_add_uv_block = vp9_dequant_idct_add_uv_block;
|
Add lossless compression mode.
This commit adds lossless compression capability to the experimental
branch. The lossless experiment can be enabled using --enable-lossless
in configure. When the experiment is enabled, the encoder will use
lossless compression mode by command line option --lossless, and the
decoder automatically recognizes a losslessly encoded clip and decodes
accordingly.
To achieve the lossless coding, this commit has changed the following:
1. To encode at lossless mode, encoder forces the use of unit
quantizer, i.e, Q 0, where effective quantization is 1. Encoder also
disables the usage of 8x8 transform and allows only 4x4 transform;
2. At Q 0, the first order 4x4 DCT/IDCT have been switched over
to a pair of forward and inverse Walsh-Hadamard Transform
(http://goo.gl/EIsfy), with proper scaling applied to match the range
of the original 4x4 DCT/IDCT pair;
3. At Q 0, the second order remains to use the previous
walsh-hadamard transform pair. However, to maintain the reversibility
in second order transform at Q 0, scaling down is applied to first
order DC coefficients prior to forward transform, and scaling up is
applied to the second order output prior to quantization. Symmetric
upscaling and downscaling are added around inverse second order
transform;
4. At lossless mode, encoder also disables a number of minor
features to ensure no loss is introduced, these features includes:
a. Trellis quantization optimization
b. Loop filtering
c. Aggressive zero-binning, rounding and zero-bin boosting
d. Mode based zero-bin boosting
Lossless coding test was performed on all clips within the derf set,
to verify that the commit has achieved lossless compression for all
clips. The average compression ratio is around 2.57 to 1.
(http://goo.gl/dEShs)
Change-Id: Ia3aba7dd09df40dd590f93b9aba134defbc64e34
2012-06-14 04:03:31 +02:00
|
|
|
#endif
|
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
for (i = 16; i < 24; i++) {
|
|
|
|
xd->block[i].dequant = pc->UVdequant[QIndex];
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
xd->block[24].dequant = pc->Y2dequant[QIndex];
|
2010-05-18 17:58:33 +02:00
|
|
|
|
|
|
|
}
|
|
|
|
|
2010-10-28 01:04:02 +02:00
|
|
|
/* skip_recon_mb() is Modified: Instead of writing the result to predictor buffer and then copying it
|
|
|
|
* to dst buffer, we can write the result directly to dst buffer. This eliminates unnecessary copy.
|
|
|
|
*/
|
2012-10-31 01:53:32 +01:00
|
|
|
static void skip_recon_mb(VP9D_COMP *pbi, MACROBLOCKD *xd) {
|
2012-07-14 00:21:29 +02:00
|
|
|
if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) {
|
2013-01-06 03:20:25 +01:00
|
|
|
if (xd->mode_info_context->mbmi.sb_type == BLOCK_SIZE_SB64X64) {
|
|
|
|
vp9_build_intra_predictors_sb64uv_s(xd);
|
|
|
|
vp9_build_intra_predictors_sb64y_s(xd);
|
2013-01-10 02:21:28 +01:00
|
|
|
} else if (xd->mode_info_context->mbmi.sb_type == BLOCK_SIZE_SB32X32) {
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_build_intra_predictors_sbuv_s(xd);
|
|
|
|
vp9_build_intra_predictors_sby_s(xd);
|
2013-01-08 19:29:22 +01:00
|
|
|
} else {
|
2013-01-06 03:20:25 +01:00
|
|
|
vp9_build_intra_predictors_mbuv_s(xd);
|
|
|
|
vp9_build_intra_predictors_mby_s(xd);
|
2012-08-20 23:43:34 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
} else {
|
2013-01-06 03:20:25 +01:00
|
|
|
if (xd->mode_info_context->mbmi.sb_type == BLOCK_SIZE_SB64X64) {
|
|
|
|
vp9_build_inter64x64_predictors_sb(xd,
|
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->dst.uv_stride);
|
2013-01-10 02:21:28 +01:00
|
|
|
} else if (xd->mode_info_context->mbmi.sb_type == BLOCK_SIZE_SB32X32) {
|
2012-11-07 15:50:25 +01:00
|
|
|
vp9_build_inter32x32_predictors_sb(xd,
|
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->dst.uv_stride);
|
2013-01-08 19:29:22 +01:00
|
|
|
} else {
|
2013-01-06 03:20:25 +01:00
|
|
|
vp9_build_1st_inter16x16_predictors_mb(xd,
|
2012-11-07 15:50:25 +01:00
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->dst.uv_stride);
|
2013-01-06 03:20:25 +01:00
|
|
|
|
|
|
|
if (xd->mode_info_context->mbmi.second_ref_frame > 0) {
|
|
|
|
vp9_build_2nd_inter16x16_predictors_mb(xd,
|
2012-11-07 15:50:25 +01:00
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->dst.uv_stride);
|
2013-01-06 03:20:25 +01:00
|
|
|
}
|
|
|
|
#if CONFIG_COMP_INTERINTRA_PRED
|
|
|
|
else if (xd->mode_info_context->mbmi.second_ref_frame == INTRA_FRAME) {
|
|
|
|
vp9_build_interintra_16x16_predictors_mb(xd,
|
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->dst.uv_stride);
|
|
|
|
}
|
2012-11-07 15:50:25 +01:00
|
|
|
#endif
|
2012-08-20 23:43:34 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
|
|
|
|
2012-11-16 00:14:38 +01:00
|
|
|
static void decode_16x16(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
|
|
|
BOOL_DECODER* const bc) {
|
|
|
|
BLOCKD *bd = &xd->block[0];
|
|
|
|
TX_TYPE tx_type = get_tx_type_16x16(xd, bd);
|
|
|
|
assert(get_2nd_order_usage(xd) == 0);
|
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
if (dec_debug) {
|
|
|
|
int i;
|
|
|
|
printf("\n");
|
|
|
|
printf("qcoeff 16x16\n");
|
|
|
|
for (i = 0; i < 400; i++) {
|
|
|
|
printf("%3d ", xd->qcoeff[i]);
|
|
|
|
if (i % 16 == 15) printf("\n");
|
|
|
|
}
|
|
|
|
printf("\n");
|
|
|
|
printf("predictor\n");
|
|
|
|
for (i = 0; i < 400; i++) {
|
|
|
|
printf("%3d ", xd->predictor[i]);
|
|
|
|
if (i % 16 == 15) printf("\n");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
vp9_ht_dequant_idct_add_16x16_c(tx_type, xd->qcoeff,
|
|
|
|
xd->block[0].dequant, xd->predictor,
|
2012-12-13 00:49:39 +01:00
|
|
|
xd->dst.y_buffer, 16, xd->dst.y_stride,
|
|
|
|
xd->eobs[0]);
|
2012-11-16 00:14:38 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_16x16(xd->qcoeff, xd->block[0].dequant,
|
|
|
|
xd->predictor, xd->dst.y_buffer,
|
|
|
|
16, xd->dst.y_stride, xd->eobs[0]);
|
|
|
|
}
|
|
|
|
vp9_dequant_idct_add_uv_block_8x8(
|
|
|
|
xd->qcoeff + 16 * 16, xd->block[16].dequant,
|
|
|
|
xd->predictor + 16 * 16, xd->dst.u_buffer, xd->dst.v_buffer,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16, xd);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void decode_8x8(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
|
|
|
BOOL_DECODER* const bc) {
|
|
|
|
// First do Y
|
|
|
|
// if the first one is DCT_DCT assume all the rest are as well
|
|
|
|
TX_TYPE tx_type = get_tx_type_8x8(xd, &xd->block[0]);
|
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
if (dec_debug) {
|
|
|
|
int i;
|
|
|
|
printf("\n");
|
|
|
|
printf("qcoeff 8x8\n");
|
|
|
|
for (i = 0; i < 400; i++) {
|
|
|
|
printf("%3d ", xd->qcoeff[i]);
|
|
|
|
if (i % 16 == 15) printf("\n");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
if (tx_type != DCT_DCT || xd->mode_info_context->mbmi.mode == I8X8_PRED) {
|
|
|
|
int i;
|
|
|
|
assert(get_2nd_order_usage(xd) == 0);
|
|
|
|
for (i = 0; i < 4; i++) {
|
|
|
|
int ib = vp9_i8x8_block[i];
|
|
|
|
int idx = (ib & 0x02) ? (ib + 2) : ib;
|
2012-12-19 00:31:19 +01:00
|
|
|
int16_t *q = xd->block[idx].qcoeff;
|
|
|
|
int16_t *dq = xd->block[0].dequant;
|
|
|
|
uint8_t *pre = xd->block[ib].predictor;
|
|
|
|
uint8_t *dst = *(xd->block[ib].base_dst) + xd->block[ib].dst;
|
2012-11-16 00:14:38 +01:00
|
|
|
int stride = xd->dst.y_stride;
|
|
|
|
BLOCKD *b = &xd->block[ib];
|
|
|
|
if (xd->mode_info_context->mbmi.mode == I8X8_PRED) {
|
|
|
|
int i8x8mode = b->bmi.as_mode.first;
|
|
|
|
vp9_intra8x8_predict(b, i8x8mode, b->predictor);
|
|
|
|
}
|
|
|
|
tx_type = get_tx_type_8x8(xd, &xd->block[ib]);
|
|
|
|
if (tx_type != DCT_DCT) {
|
2012-12-13 00:49:39 +01:00
|
|
|
vp9_ht_dequant_idct_add_8x8_c(tx_type, q, dq, pre, dst, 16, stride,
|
|
|
|
xd->eobs[idx]);
|
2012-11-16 00:14:38 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_8x8_c(q, dq, pre, dst, 16, stride,
|
|
|
|
0, xd->eobs[idx]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else if (xd->mode_info_context->mbmi.mode == SPLITMV) {
|
|
|
|
assert(get_2nd_order_usage(xd) == 0);
|
|
|
|
vp9_dequant_idct_add_y_block_8x8(xd->qcoeff,
|
|
|
|
xd->block[0].dequant,
|
|
|
|
xd->predictor,
|
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->eobs, xd);
|
|
|
|
} else {
|
|
|
|
BLOCKD *b = &xd->block[24];
|
2012-11-29 18:07:12 +01:00
|
|
|
assert(get_2nd_order_usage(xd) == 1);
|
2012-11-16 00:14:38 +01:00
|
|
|
vp9_dequantize_b_2x2(b);
|
|
|
|
vp9_short_ihaar2x2(&b->dqcoeff[0], b->diff, 8);
|
|
|
|
((int *)b->qcoeff)[0] = 0; // 2nd order block are set to 0 after idct
|
|
|
|
((int *)b->qcoeff)[1] = 0;
|
|
|
|
((int *)b->qcoeff)[2] = 0;
|
|
|
|
((int *)b->qcoeff)[3] = 0;
|
|
|
|
((int *)b->qcoeff)[4] = 0;
|
|
|
|
((int *)b->qcoeff)[5] = 0;
|
|
|
|
((int *)b->qcoeff)[6] = 0;
|
|
|
|
((int *)b->qcoeff)[7] = 0;
|
|
|
|
vp9_dequant_dc_idct_add_y_block_8x8(xd->qcoeff,
|
|
|
|
xd->block[0].dequant,
|
|
|
|
xd->predictor,
|
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->eobs,
|
|
|
|
xd->block[24].diff,
|
|
|
|
xd);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Now do UV
|
|
|
|
if (xd->mode_info_context->mbmi.mode == I8X8_PRED) {
|
|
|
|
int i;
|
|
|
|
for (i = 0; i < 4; i++) {
|
|
|
|
int ib = vp9_i8x8_block[i];
|
|
|
|
BLOCKD *b = &xd->block[ib];
|
|
|
|
int i8x8mode = b->bmi.as_mode.first;
|
|
|
|
b = &xd->block[16 + i];
|
|
|
|
vp9_intra_uv4x4_predict(&xd->block[16 + i], i8x8mode, b->predictor);
|
|
|
|
pbi->idct_add(b->qcoeff, b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 8, b->dst_stride);
|
|
|
|
b = &xd->block[20 + i];
|
|
|
|
vp9_intra_uv4x4_predict(&xd->block[20 + i], i8x8mode, b->predictor);
|
|
|
|
pbi->idct_add(b->qcoeff, b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 8, b->dst_stride);
|
|
|
|
}
|
|
|
|
} else if (xd->mode_info_context->mbmi.mode == SPLITMV) {
|
|
|
|
pbi->idct_add_uv_block(xd->qcoeff + 16 * 16, xd->block[16].dequant,
|
|
|
|
xd->predictor + 16 * 16, xd->dst.u_buffer, xd->dst.v_buffer,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16);
|
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_uv_block_8x8
|
|
|
|
(xd->qcoeff + 16 * 16, xd->block[16].dequant,
|
|
|
|
xd->predictor + 16 * 16, xd->dst.u_buffer, xd->dst.v_buffer,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16, xd);
|
|
|
|
}
|
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
if (dec_debug) {
|
|
|
|
int i;
|
|
|
|
printf("\n");
|
|
|
|
printf("predictor\n");
|
|
|
|
for (i = 0; i < 384; i++) {
|
|
|
|
printf("%3d ", xd->predictor[i]);
|
|
|
|
if (i % 16 == 15) printf("\n");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
static void decode_4x4(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
|
|
|
BOOL_DECODER* const bc) {
|
|
|
|
TX_TYPE tx_type;
|
|
|
|
int i, eobtotal = 0;
|
|
|
|
MB_PREDICTION_MODE mode = xd->mode_info_context->mbmi.mode;
|
|
|
|
if (mode == I8X8_PRED) {
|
|
|
|
assert(get_2nd_order_usage(xd) == 0);
|
|
|
|
for (i = 0; i < 4; i++) {
|
|
|
|
int ib = vp9_i8x8_block[i];
|
|
|
|
const int iblock[4] = {0, 1, 4, 5};
|
|
|
|
int j;
|
|
|
|
int i8x8mode;
|
|
|
|
BLOCKD *b;
|
|
|
|
b = &xd->block[ib];
|
|
|
|
i8x8mode = b->bmi.as_mode.first;
|
|
|
|
vp9_intra8x8_predict(b, i8x8mode, b->predictor);
|
|
|
|
for (j = 0; j < 4; j++) {
|
|
|
|
b = &xd->block[ib + iblock[j]];
|
|
|
|
tx_type = get_tx_type_4x4(xd, b);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
vp9_ht_dequant_idct_add_c(tx_type, b->qcoeff,
|
|
|
|
b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 16,
|
2012-12-13 00:49:39 +01:00
|
|
|
b->dst_stride, b->eob);
|
2012-11-16 00:14:38 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add(b->qcoeff, b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 16, b->dst_stride);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
b = &xd->block[16 + i];
|
|
|
|
vp9_intra_uv4x4_predict(b, i8x8mode, b->predictor);
|
|
|
|
pbi->idct_add(b->qcoeff, b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 8, b->dst_stride);
|
|
|
|
b = &xd->block[20 + i];
|
|
|
|
vp9_intra_uv4x4_predict(b, i8x8mode, b->predictor);
|
|
|
|
pbi->idct_add(b->qcoeff, b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 8, b->dst_stride);
|
|
|
|
}
|
|
|
|
} else if (mode == B_PRED) {
|
|
|
|
assert(get_2nd_order_usage(xd) == 0);
|
|
|
|
for (i = 0; i < 16; i++) {
|
|
|
|
int b_mode;
|
|
|
|
BLOCKD *b = &xd->block[i];
|
|
|
|
b_mode = xd->mode_info_context->bmi[i].as_mode.first;
|
|
|
|
#if CONFIG_NEWBINTRAMODES
|
|
|
|
xd->mode_info_context->bmi[i].as_mode.context = b->bmi.as_mode.context =
|
|
|
|
vp9_find_bpred_context(b);
|
|
|
|
#endif
|
|
|
|
if (!xd->mode_info_context->mbmi.mb_skip_coeff)
|
|
|
|
eobtotal += vp9_decode_coefs_4x4(pbi, xd, bc, PLANE_TYPE_Y_WITH_DC, i);
|
|
|
|
|
2013-01-14 23:37:53 +01:00
|
|
|
vp9_intra4x4_predict(b, b_mode, b->predictor);
|
2012-11-16 00:14:38 +01:00
|
|
|
tx_type = get_tx_type_4x4(xd, b);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
vp9_ht_dequant_idct_add_c(tx_type, b->qcoeff,
|
|
|
|
b->dequant, b->predictor,
|
2012-12-13 00:49:39 +01:00
|
|
|
*(b->base_dst) + b->dst, 16, b->dst_stride,
|
|
|
|
b->eob);
|
2012-11-16 00:14:38 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add(b->qcoeff, b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 16, b->dst_stride);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (!xd->mode_info_context->mbmi.mb_skip_coeff) {
|
|
|
|
vp9_decode_mb_tokens_4x4_uv(pbi, xd, bc);
|
|
|
|
}
|
2012-11-29 00:15:51 +01:00
|
|
|
xd->above_context->y2 = 0;
|
|
|
|
xd->left_context->y2 = 0;
|
2012-11-16 00:14:38 +01:00
|
|
|
vp9_build_intra_predictors_mbuv(xd);
|
|
|
|
pbi->idct_add_uv_block(xd->qcoeff + 16 * 16,
|
|
|
|
xd->block[16].dequant,
|
|
|
|
xd->predictor + 16 * 16,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.uv_stride,
|
|
|
|
xd->eobs + 16);
|
|
|
|
} else if (mode == SPLITMV) {
|
|
|
|
assert(get_2nd_order_usage(xd) == 0);
|
|
|
|
pbi->idct_add_y_block(xd->qcoeff,
|
|
|
|
xd->block[0].dequant,
|
|
|
|
xd->predictor,
|
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->eobs);
|
|
|
|
pbi->idct_add_uv_block(xd->qcoeff + 16 * 16,
|
|
|
|
xd->block[16].dequant,
|
|
|
|
xd->predictor + 16 * 16,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.uv_stride,
|
|
|
|
xd->eobs + 16);
|
|
|
|
} else {
|
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
if (dec_debug) {
|
|
|
|
int i;
|
|
|
|
printf("\n");
|
|
|
|
printf("qcoeff 4x4\n");
|
|
|
|
for (i = 0; i < 400; i++) {
|
|
|
|
printf("%3d ", xd->qcoeff[i]);
|
|
|
|
if (i % 16 == 15) printf("\n");
|
|
|
|
}
|
|
|
|
printf("\n");
|
|
|
|
printf("predictor\n");
|
|
|
|
for (i = 0; i < 400; i++) {
|
|
|
|
printf("%3d ", xd->predictor[i]);
|
|
|
|
if (i % 16 == 15) printf("\n");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
tx_type = get_tx_type_4x4(xd, &xd->block[0]);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
assert(get_2nd_order_usage(xd) == 0);
|
|
|
|
for (i = 0; i < 16; i++) {
|
|
|
|
BLOCKD *b = &xd->block[i];
|
|
|
|
tx_type = get_tx_type_4x4(xd, b);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
vp9_ht_dequant_idct_add_c(tx_type, b->qcoeff,
|
|
|
|
b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 16,
|
2012-12-13 00:49:39 +01:00
|
|
|
b->dst_stride, b->eob);
|
2012-11-16 00:14:38 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add(b->qcoeff, b->dequant, b->predictor,
|
|
|
|
*(b->base_dst) + b->dst, 16, b->dst_stride);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
BLOCKD *b = &xd->block[24];
|
2012-11-29 18:07:12 +01:00
|
|
|
assert(get_2nd_order_usage(xd) == 1);
|
2012-11-16 00:14:38 +01:00
|
|
|
vp9_dequantize_b(b);
|
|
|
|
if (xd->eobs[24] > 1) {
|
|
|
|
vp9_short_inv_walsh4x4(&b->dqcoeff[0], b->diff);
|
|
|
|
((int *)b->qcoeff)[0] = 0;
|
|
|
|
((int *)b->qcoeff)[1] = 0;
|
|
|
|
((int *)b->qcoeff)[2] = 0;
|
|
|
|
((int *)b->qcoeff)[3] = 0;
|
|
|
|
((int *)b->qcoeff)[4] = 0;
|
|
|
|
((int *)b->qcoeff)[5] = 0;
|
|
|
|
((int *)b->qcoeff)[6] = 0;
|
|
|
|
((int *)b->qcoeff)[7] = 0;
|
|
|
|
} else {
|
|
|
|
xd->inv_walsh4x4_1(&b->dqcoeff[0], b->diff);
|
|
|
|
((int *)b->qcoeff)[0] = 0;
|
|
|
|
}
|
|
|
|
vp9_dequantize_b(b);
|
|
|
|
pbi->dc_idct_add_y_block(xd->qcoeff,
|
|
|
|
xd->block[0].dequant,
|
|
|
|
xd->predictor,
|
|
|
|
xd->dst.y_buffer,
|
|
|
|
xd->dst.y_stride,
|
|
|
|
xd->eobs,
|
|
|
|
xd->block[24].diff);
|
|
|
|
}
|
|
|
|
pbi->idct_add_uv_block(xd->qcoeff + 16 * 16,
|
|
|
|
xd->block[16].dequant,
|
|
|
|
xd->predictor + 16 * 16,
|
|
|
|
xd->dst.u_buffer,
|
|
|
|
xd->dst.v_buffer,
|
|
|
|
xd->dst.uv_stride,
|
|
|
|
xd->eobs + 16);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-11-30 00:26:56 +01:00
|
|
|
static void decode_16x16_sb(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
2013-01-06 03:20:25 +01:00
|
|
|
BOOL_DECODER* const bc, int n,
|
|
|
|
int maska, int shiftb) {
|
|
|
|
int x_idx = n & maska, y_idx = n >> shiftb;
|
2012-11-30 00:26:56 +01:00
|
|
|
TX_TYPE tx_type = get_tx_type_16x16(xd, &xd->block[0]);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
vp9_ht_dequant_idct_add_16x16_c(
|
|
|
|
tx_type, xd->qcoeff, xd->block[0].dequant,
|
|
|
|
xd->dst.y_buffer + y_idx * 16 * xd->dst.y_stride + x_idx * 16,
|
|
|
|
xd->dst.y_buffer + y_idx * 16 * xd->dst.y_stride + x_idx * 16,
|
2012-12-13 00:49:39 +01:00
|
|
|
xd->dst.y_stride, xd->dst.y_stride, xd->block[0].eob);
|
2012-11-30 00:26:56 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_16x16(
|
|
|
|
xd->qcoeff, xd->block[0].dequant,
|
|
|
|
xd->dst.y_buffer + y_idx * 16 * xd->dst.y_stride + x_idx * 16,
|
|
|
|
xd->dst.y_buffer + y_idx * 16 * xd->dst.y_stride + x_idx * 16,
|
|
|
|
xd->dst.y_stride, xd->dst.y_stride, xd->eobs[0]);
|
|
|
|
}
|
|
|
|
vp9_dequant_idct_add_uv_block_8x8_inplace_c(
|
|
|
|
xd->qcoeff + 16 * 16,
|
|
|
|
xd->block[16].dequant,
|
|
|
|
xd->dst.u_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.v_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16, xd);
|
|
|
|
};
|
|
|
|
|
|
|
|
static void decode_8x8_sb(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
2013-01-06 03:20:25 +01:00
|
|
|
BOOL_DECODER* const bc, int n,
|
|
|
|
int maska, int shiftb) {
|
|
|
|
int x_idx = n & maska, y_idx = n >> shiftb;
|
2012-11-30 00:26:56 +01:00
|
|
|
BLOCKD *b = &xd->block[24];
|
|
|
|
TX_TYPE tx_type = get_tx_type_8x8(xd, &xd->block[0]);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
int i;
|
|
|
|
for (i = 0; i < 4; i++) {
|
|
|
|
int ib = vp9_i8x8_block[i];
|
|
|
|
int idx = (ib & 0x02) ? (ib + 2) : ib;
|
2012-12-19 00:31:19 +01:00
|
|
|
int16_t *q = xd->block[idx].qcoeff;
|
|
|
|
int16_t *dq = xd->block[0].dequant;
|
2012-11-30 00:26:56 +01:00
|
|
|
int stride = xd->dst.y_stride;
|
|
|
|
BLOCKD *b = &xd->block[ib];
|
|
|
|
tx_type = get_tx_type_8x8(xd, &xd->block[ib]);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
vp9_ht_dequant_idct_add_8x8_c(
|
|
|
|
tx_type, q, dq,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 2) * 8) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 1) * 8,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 2) * 8) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 1) * 8,
|
2012-12-13 00:49:39 +01:00
|
|
|
stride, stride, b->eob);
|
2012-11-30 00:26:56 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_8x8_c(
|
|
|
|
q, dq,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 2) * 8) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 1) * 8,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 2) * 8) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 1) * 8,
|
|
|
|
stride, stride, 0, b->eob);
|
|
|
|
}
|
|
|
|
vp9_dequant_idct_add_uv_block_8x8_inplace_c(
|
|
|
|
xd->qcoeff + 16 * 16, xd->block[16].dequant,
|
|
|
|
xd->dst.u_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.v_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16, xd);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
vp9_dequantize_b_2x2(b);
|
|
|
|
vp9_short_ihaar2x2(&b->dqcoeff[0], b->diff, 8);
|
|
|
|
((int *)b->qcoeff)[0] = 0; // 2nd order block are set to 0 after idct
|
|
|
|
((int *)b->qcoeff)[1] = 0;
|
|
|
|
((int *)b->qcoeff)[2] = 0;
|
|
|
|
((int *)b->qcoeff)[3] = 0;
|
|
|
|
((int *)b->qcoeff)[4] = 0;
|
|
|
|
((int *)b->qcoeff)[5] = 0;
|
|
|
|
((int *)b->qcoeff)[6] = 0;
|
|
|
|
((int *)b->qcoeff)[7] = 0;
|
|
|
|
vp9_dequant_dc_idct_add_y_block_8x8_inplace_c(
|
|
|
|
xd->qcoeff, xd->block[0].dequant,
|
|
|
|
xd->dst.y_buffer + y_idx * 16 * xd->dst.y_stride + x_idx * 16,
|
|
|
|
xd->dst.y_stride, xd->eobs, xd->block[24].diff, xd);
|
|
|
|
vp9_dequant_idct_add_uv_block_8x8_inplace_c(
|
|
|
|
xd->qcoeff + 16 * 16, xd->block[16].dequant,
|
|
|
|
xd->dst.u_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.v_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16, xd);
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
static void decode_4x4_sb(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
2013-01-06 03:20:25 +01:00
|
|
|
BOOL_DECODER* const bc, int n,
|
|
|
|
int maska, int shiftb) {
|
|
|
|
int x_idx = n & maska, y_idx = n >> shiftb;
|
2012-11-30 00:26:56 +01:00
|
|
|
BLOCKD *b = &xd->block[24];
|
|
|
|
TX_TYPE tx_type = get_tx_type_4x4(xd, &xd->block[0]);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
int i;
|
|
|
|
for (i = 0; i < 16; i++) {
|
|
|
|
BLOCKD *b = &xd->block[i];
|
|
|
|
tx_type = get_tx_type_4x4(xd, b);
|
|
|
|
if (tx_type != DCT_DCT) {
|
|
|
|
vp9_ht_dequant_idct_add_c(
|
|
|
|
tx_type, b->qcoeff, b->dequant,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 4) * 4) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 3) * 4,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 4) * 4) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 3) * 4,
|
2012-12-13 00:49:39 +01:00
|
|
|
xd->dst.y_stride, xd->dst.y_stride, b->eob);
|
2012-11-30 00:26:56 +01:00
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_c(
|
|
|
|
b->qcoeff, b->dequant,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 4) * 4) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 3) * 4,
|
|
|
|
xd->dst.y_buffer + (y_idx * 16 + (i / 4) * 4) * xd->dst.y_stride
|
|
|
|
+ x_idx * 16 + (i & 3) * 4,
|
|
|
|
xd->dst.y_stride, xd->dst.y_stride);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
vp9_dequantize_b(b);
|
|
|
|
if (xd->eobs[24] > 1) {
|
|
|
|
vp9_short_inv_walsh4x4(&b->dqcoeff[0], b->diff);
|
|
|
|
((int *)b->qcoeff)[0] = 0;
|
|
|
|
((int *)b->qcoeff)[1] = 0;
|
|
|
|
((int *)b->qcoeff)[2] = 0;
|
|
|
|
((int *)b->qcoeff)[3] = 0;
|
|
|
|
((int *)b->qcoeff)[4] = 0;
|
|
|
|
((int *)b->qcoeff)[5] = 0;
|
|
|
|
((int *)b->qcoeff)[6] = 0;
|
|
|
|
((int *)b->qcoeff)[7] = 0;
|
|
|
|
} else {
|
|
|
|
xd->inv_walsh4x4_1(&b->dqcoeff[0], b->diff);
|
|
|
|
((int *)b->qcoeff)[0] = 0;
|
|
|
|
}
|
|
|
|
vp9_dequant_dc_idct_add_y_block_4x4_inplace_c(
|
|
|
|
xd->qcoeff, xd->block[0].dequant,
|
|
|
|
xd->dst.y_buffer + y_idx * 16 * xd->dst.y_stride + x_idx * 16,
|
|
|
|
xd->dst.y_stride, xd->eobs, xd->block[24].diff, xd);
|
|
|
|
}
|
|
|
|
vp9_dequant_idct_add_uv_block_4x4_inplace_c(
|
|
|
|
xd->qcoeff + 16 * 16, xd->block[16].dequant,
|
|
|
|
xd->dst.u_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.v_buffer + y_idx * 8 * xd->dst.uv_stride + x_idx * 8,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16, xd);
|
|
|
|
};
|
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
static void decode_superblock64(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
|
|
|
int mb_row, unsigned int mb_col,
|
|
|
|
BOOL_DECODER* const bc) {
|
2012-11-08 20:03:00 +01:00
|
|
|
int i, n, eobtotal;
|
|
|
|
TX_SIZE tx_size = xd->mode_info_context->mbmi.txfm_size;
|
|
|
|
VP9_COMMON *const pc = &pbi->common;
|
|
|
|
MODE_INFO *orig_mi = xd->mode_info_context;
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
const int mis = pc->mode_info_stride;
|
2012-11-08 20:03:00 +01:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
assert(xd->mode_info_context->mbmi.sb_type == BLOCK_SIZE_SB64X64);
|
|
|
|
|
|
|
|
if (pbi->common.frame_type != KEY_FRAME)
|
|
|
|
vp9_setup_interp_filters(xd, xd->mode_info_context->mbmi.interp_filter, pc);
|
|
|
|
|
|
|
|
// re-initialize macroblock dequantizer before detokenization
|
|
|
|
if (xd->segmentation_enabled)
|
|
|
|
mb_init_dequantizer(pbi, xd);
|
|
|
|
|
|
|
|
if (xd->mode_info_context->mbmi.mb_skip_coeff) {
|
|
|
|
int n;
|
|
|
|
|
|
|
|
vp9_reset_mb_tokens_context(xd);
|
|
|
|
for (n = 1; n <= 3; n++) {
|
|
|
|
if (mb_col < pc->mb_cols - n)
|
|
|
|
xd->above_context += n;
|
|
|
|
if (mb_row < pc->mb_rows - n)
|
|
|
|
xd->left_context += n;
|
|
|
|
vp9_reset_mb_tokens_context(xd);
|
|
|
|
if (mb_col < pc->mb_cols - n)
|
|
|
|
xd->above_context -= n;
|
|
|
|
if (mb_row < pc->mb_rows - n)
|
|
|
|
xd->left_context -= n;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Special case: Force the loopfilter to skip when eobtotal and
|
|
|
|
* mb_skip_coeff are zero.
|
|
|
|
*/
|
|
|
|
skip_recon_mb(pbi, xd);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* do prediction */
|
|
|
|
if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) {
|
|
|
|
vp9_build_intra_predictors_sb64y_s(xd);
|
|
|
|
vp9_build_intra_predictors_sb64uv_s(xd);
|
|
|
|
} else {
|
|
|
|
vp9_build_inter64x64_predictors_sb(xd, xd->dst.y_buffer,
|
|
|
|
xd->dst.u_buffer, xd->dst.v_buffer,
|
|
|
|
xd->dst.y_stride, xd->dst.uv_stride);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* dequantization and idct */
|
|
|
|
if (xd->mode_info_context->mbmi.txfm_size == TX_32X32) {
|
|
|
|
for (n = 0; n < 4; n++) {
|
|
|
|
const int x_idx = n & 1, y_idx = n >> 1;
|
|
|
|
|
|
|
|
if (mb_col + x_idx * 2 >= pc->mb_cols ||
|
|
|
|
mb_row + y_idx * 2 >= pc->mb_rows)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
xd->left_context = pc->left_context + (y_idx << 1);
|
|
|
|
xd->above_context = pc->above_context + mb_col + (x_idx << 1);
|
|
|
|
xd->mode_info_context = orig_mi + x_idx * 2 + y_idx * 2 * mis;
|
|
|
|
eobtotal = vp9_decode_sb_tokens(pbi, xd, bc);
|
|
|
|
if (eobtotal == 0) { // skip loopfilter
|
|
|
|
xd->mode_info_context->mbmi.mb_skip_coeff = 1;
|
|
|
|
if (mb_col + 1 < pc->mb_cols)
|
|
|
|
xd->mode_info_context[1].mbmi.mb_skip_coeff = 1;
|
|
|
|
if (mb_row + 1 < pc->mb_rows) {
|
|
|
|
xd->mode_info_context[mis].mbmi.mb_skip_coeff = 1;
|
|
|
|
if (mb_col + 1 < pc->mb_cols)
|
|
|
|
xd->mode_info_context[mis + 1].mbmi.mb_skip_coeff = 1;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_32x32(xd->sb_coeff_data.qcoeff, xd->block[0].dequant,
|
|
|
|
xd->dst.y_buffer + x_idx * 32 +
|
|
|
|
xd->dst.y_stride * y_idx * 32,
|
|
|
|
xd->dst.y_buffer + x_idx * 32 +
|
|
|
|
xd->dst.y_stride * y_idx * 32,
|
|
|
|
xd->dst.y_stride, xd->dst.y_stride,
|
|
|
|
xd->eobs[0]);
|
|
|
|
vp9_dequant_idct_add_uv_block_16x16_c(xd->sb_coeff_data.qcoeff + 1024,
|
|
|
|
xd->block[16].dequant,
|
|
|
|
xd->dst.u_buffer + x_idx * 16 +
|
|
|
|
xd->dst.uv_stride * y_idx * 16,
|
|
|
|
xd->dst.v_buffer + x_idx * 16 +
|
|
|
|
xd->dst.uv_stride * y_idx * 16,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
for (n = 0; n < 16; n++) {
|
|
|
|
int x_idx = n & 3, y_idx = n >> 2;
|
|
|
|
|
|
|
|
if (mb_col + x_idx >= pc->mb_cols || mb_row + y_idx >= pc->mb_rows)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
xd->above_context = pc->above_context + mb_col + x_idx;
|
|
|
|
xd->left_context = pc->left_context + y_idx;
|
|
|
|
xd->mode_info_context = orig_mi + x_idx + y_idx * mis;
|
|
|
|
for (i = 0; i < 25; i++) {
|
|
|
|
xd->block[i].eob = 0;
|
|
|
|
xd->eobs[i] = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
eobtotal = vp9_decode_mb_tokens(pbi, xd, bc);
|
|
|
|
if (eobtotal == 0) { // skip loopfilter
|
|
|
|
xd->mode_info_context->mbmi.mb_skip_coeff = 1;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (tx_size == TX_16X16) {
|
|
|
|
decode_16x16_sb(pbi, xd, bc, n, 3, 2);
|
|
|
|
} else if (tx_size == TX_8X8) {
|
|
|
|
decode_8x8_sb(pbi, xd, bc, n, 3, 2);
|
|
|
|
} else {
|
|
|
|
decode_4x4_sb(pbi, xd, bc, n, 3, 2);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
xd->above_context = pc->above_context + mb_col;
|
|
|
|
xd->left_context = pc->left_context;
|
|
|
|
xd->mode_info_context = orig_mi;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void decode_superblock32(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
|
|
|
int mb_row, unsigned int mb_col,
|
|
|
|
BOOL_DECODER* const bc) {
|
|
|
|
int i, n, eobtotal;
|
|
|
|
TX_SIZE tx_size = xd->mode_info_context->mbmi.txfm_size;
|
|
|
|
VP9_COMMON *const pc = &pbi->common;
|
|
|
|
MODE_INFO *orig_mi = xd->mode_info_context;
|
|
|
|
const int mis = pc->mode_info_stride;
|
|
|
|
|
|
|
|
assert(xd->mode_info_context->mbmi.sb_type == BLOCK_SIZE_SB32X32);
|
2012-11-08 20:03:00 +01:00
|
|
|
|
2012-11-30 01:39:15 +01:00
|
|
|
if (pbi->common.frame_type != KEY_FRAME)
|
|
|
|
vp9_setup_interp_filters(xd, xd->mode_info_context->mbmi.interp_filter, pc);
|
|
|
|
|
2012-11-08 20:03:00 +01:00
|
|
|
// re-initialize macroblock dequantizer before detokenization
|
|
|
|
if (xd->segmentation_enabled)
|
|
|
|
mb_init_dequantizer(pbi, xd);
|
|
|
|
|
|
|
|
if (xd->mode_info_context->mbmi.mb_skip_coeff) {
|
|
|
|
vp9_reset_mb_tokens_context(xd);
|
|
|
|
if (mb_col < pc->mb_cols - 1)
|
|
|
|
xd->above_context++;
|
|
|
|
if (mb_row < pc->mb_rows - 1)
|
|
|
|
xd->left_context++;
|
|
|
|
vp9_reset_mb_tokens_context(xd);
|
|
|
|
if (mb_col < pc->mb_cols - 1)
|
|
|
|
xd->above_context--;
|
|
|
|
if (mb_row < pc->mb_rows - 1)
|
|
|
|
xd->left_context--;
|
|
|
|
|
|
|
|
/* Special case: Force the loopfilter to skip when eobtotal and
|
|
|
|
* mb_skip_coeff are zero.
|
|
|
|
*/
|
|
|
|
skip_recon_mb(pbi, xd);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* do prediction */
|
|
|
|
if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) {
|
|
|
|
vp9_build_intra_predictors_sby_s(xd);
|
|
|
|
vp9_build_intra_predictors_sbuv_s(xd);
|
|
|
|
} else {
|
|
|
|
vp9_build_inter32x32_predictors_sb(xd, xd->dst.y_buffer,
|
|
|
|
xd->dst.u_buffer, xd->dst.v_buffer,
|
|
|
|
xd->dst.y_stride, xd->dst.uv_stride);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* dequantization and idct */
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
if (xd->mode_info_context->mbmi.txfm_size == TX_32X32) {
|
|
|
|
eobtotal = vp9_decode_sb_tokens(pbi, xd, bc);
|
|
|
|
if (eobtotal == 0) { // skip loopfilter
|
|
|
|
xd->mode_info_context->mbmi.mb_skip_coeff = 1;
|
|
|
|
if (mb_col + 1 < pc->mb_cols)
|
|
|
|
xd->mode_info_context[1].mbmi.mb_skip_coeff = 1;
|
|
|
|
if (mb_row + 1 < pc->mb_rows) {
|
|
|
|
xd->mode_info_context[mis].mbmi.mb_skip_coeff = 1;
|
|
|
|
if (mb_col + 1 < pc->mb_cols)
|
|
|
|
xd->mode_info_context[mis + 1].mbmi.mb_skip_coeff = 1;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
vp9_dequant_idct_add_32x32(xd->sb_coeff_data.qcoeff, xd->block[0].dequant,
|
|
|
|
xd->dst.y_buffer, xd->dst.y_buffer,
|
|
|
|
xd->dst.y_stride, xd->dst.y_stride,
|
|
|
|
xd->eobs[0]);
|
|
|
|
vp9_dequant_idct_add_uv_block_16x16_c(xd->sb_coeff_data.qcoeff + 1024,
|
|
|
|
xd->block[16].dequant,
|
|
|
|
xd->dst.u_buffer, xd->dst.v_buffer,
|
|
|
|
xd->dst.uv_stride, xd->eobs + 16);
|
|
|
|
}
|
2013-01-10 17:23:59 +01:00
|
|
|
} else {
|
2013-01-08 19:29:22 +01:00
|
|
|
for (n = 0; n < 4; n++) {
|
|
|
|
int x_idx = n & 1, y_idx = n >> 1;
|
2012-11-08 20:03:00 +01:00
|
|
|
|
2013-01-08 19:29:22 +01:00
|
|
|
if (mb_col + x_idx >= pc->mb_cols || mb_row + y_idx >= pc->mb_rows)
|
|
|
|
continue;
|
2012-11-30 01:39:15 +01:00
|
|
|
|
2013-01-08 19:29:22 +01:00
|
|
|
xd->above_context = pc->above_context + mb_col + x_idx;
|
|
|
|
xd->left_context = pc->left_context + y_idx + (mb_row & 2);
|
|
|
|
xd->mode_info_context = orig_mi + x_idx + y_idx * mis;
|
|
|
|
for (i = 0; i < 25; i++) {
|
|
|
|
xd->block[i].eob = 0;
|
|
|
|
xd->eobs[i] = 0;
|
|
|
|
}
|
2012-11-08 20:03:00 +01:00
|
|
|
|
2013-01-08 19:29:22 +01:00
|
|
|
eobtotal = vp9_decode_mb_tokens(pbi, xd, bc);
|
|
|
|
if (eobtotal == 0) { // skip loopfilter
|
|
|
|
xd->mode_info_context->mbmi.mb_skip_coeff = 1;
|
|
|
|
continue;
|
|
|
|
}
|
2012-11-08 20:03:00 +01:00
|
|
|
|
2013-01-08 19:29:22 +01:00
|
|
|
if (tx_size == TX_16X16) {
|
|
|
|
decode_16x16_sb(pbi, xd, bc, n, 1, 1);
|
|
|
|
} else if (tx_size == TX_8X8) {
|
|
|
|
decode_8x8_sb(pbi, xd, bc, n, 1, 1);
|
|
|
|
} else {
|
|
|
|
decode_4x4_sb(pbi, xd, bc, n, 1, 1);
|
|
|
|
}
|
2012-11-08 20:03:00 +01:00
|
|
|
}
|
|
|
|
|
2013-01-08 19:29:22 +01:00
|
|
|
xd->above_context = pc->above_context + mb_col;
|
|
|
|
xd->left_context = pc->left_context + (mb_row & 2);
|
|
|
|
xd->mode_info_context = orig_mi;
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
}
|
2012-11-08 20:03:00 +01:00
|
|
|
}
|
|
|
|
|
2012-10-31 01:53:32 +01:00
|
|
|
static void decode_macroblock(VP9D_COMP *pbi, MACROBLOCKD *xd,
|
2012-10-30 01:58:18 +01:00
|
|
|
int mb_row, unsigned int mb_col,
|
2012-10-18 01:47:38 +02:00
|
|
|
BOOL_DECODER* const bc) {
|
2012-07-14 00:21:29 +02:00
|
|
|
int eobtotal = 0;
|
|
|
|
MB_PREDICTION_MODE mode;
|
|
|
|
int i;
|
2012-10-16 01:41:41 +02:00
|
|
|
int tx_size;
|
2012-11-08 20:03:00 +01:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
assert(!xd->mode_info_context->mbmi.sb_type);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2012-08-16 22:25:49 +02:00
|
|
|
// re-initialize macroblock dequantizer before detokenization
|
|
|
|
if (xd->segmentation_enabled)
|
|
|
|
mb_init_dequantizer(pbi, xd);
|
|
|
|
|
2012-10-16 01:41:41 +02:00
|
|
|
tx_size = xd->mode_info_context->mbmi.txfm_size;
|
2012-09-12 04:36:28 +02:00
|
|
|
mode = xd->mode_info_context->mbmi.mode;
|
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (xd->mode_info_context->mbmi.mb_skip_coeff) {
|
2012-10-30 22:51:31 +01:00
|
|
|
vp9_reset_mb_tokens_context(xd);
|
2012-10-31 22:40:53 +01:00
|
|
|
} else if (!bool_error(bc)) {
|
2012-07-14 00:21:29 +02:00
|
|
|
for (i = 0; i < 25; i++) {
|
|
|
|
xd->block[i].eob = 0;
|
|
|
|
xd->eobs[i] = 0;
|
2010-05-28 20:28:12 +02:00
|
|
|
}
|
2012-11-23 20:23:50 +01:00
|
|
|
if (mode != B_PRED) {
|
|
|
|
eobtotal = vp9_decode_mb_tokens(pbi, xd, bc);
|
2012-10-16 01:41:41 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-28 20:28:12 +02:00
|
|
|
|
2012-09-12 04:36:28 +02:00
|
|
|
//mode = xd->mode_info_context->mbmi.mode;
|
2012-07-18 22:43:01 +02:00
|
|
|
if (pbi->common.frame_type != KEY_FRAME)
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_setup_interp_filters(xd, xd->mode_info_context->mbmi.interp_filter,
|
2012-07-18 22:43:01 +02:00
|
|
|
&pbi->common);
|
2011-06-28 16:22:13 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (eobtotal == 0 && mode != B_PRED && mode != SPLITMV
|
|
|
|
&& mode != I8X8_PRED
|
2012-10-31 22:40:53 +01:00
|
|
|
&& !bool_error(bc)) {
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Special case: Force the loopfilter to skip when eobtotal and
|
|
|
|
* mb_skip_coeff are zero.
|
|
|
|
* */
|
|
|
|
xd->mode_info_context->mbmi.mb_skip_coeff = 1;
|
2012-11-08 20:03:00 +01:00
|
|
|
skip_recon_mb(pbi, xd);
|
|
|
|
return;
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2012-11-16 00:14:38 +01:00
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
if (dec_debug)
|
|
|
|
printf("Decoding mb: %d %d\n", xd->mode_info_context->mbmi.mode, tx_size);
|
|
|
|
#endif
|
2011-07-20 23:21:24 +02:00
|
|
|
|
2012-08-16 22:25:49 +02:00
|
|
|
// moved to be performed before detokenization
|
|
|
|
// if (xd->segmentation_enabled)
|
|
|
|
// mb_init_dequantizer(pbi, xd);
|
2010-05-28 20:28:12 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* do prediction */
|
|
|
|
if (xd->mode_info_context->mbmi.ref_frame == INTRA_FRAME) {
|
|
|
|
if (mode != I8X8_PRED) {
|
2012-11-17 07:26:12 +01:00
|
|
|
vp9_build_intra_predictors_mbuv(xd);
|
2012-07-14 00:21:29 +02:00
|
|
|
if (mode != B_PRED) {
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_build_intra_predictors_mby(xd);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
} else {
|
2012-11-30 01:39:15 +01:00
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
if (dec_debug)
|
|
|
|
printf("Decoding mb: %d %d interp %d\n",
|
|
|
|
xd->mode_info_context->mbmi.mode, tx_size,
|
|
|
|
xd->mode_info_context->mbmi.interp_filter);
|
|
|
|
#endif
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_build_inter_predictors_mb(xd);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
|
|
|
|
2012-11-16 00:14:38 +01:00
|
|
|
if (tx_size == TX_16X16) {
|
|
|
|
decode_16x16(pbi, xd, bc);
|
|
|
|
} else if (tx_size == TX_8X8) {
|
|
|
|
decode_8x8(pbi, xd, bc);
|
|
|
|
} else {
|
|
|
|
decode_4x4(pbi, xd, bc);
|
|
|
|
}
|
|
|
|
#ifdef DEC_DEBUG
|
|
|
|
if (dec_debug) {
|
|
|
|
int i, j;
|
|
|
|
printf("\n");
|
|
|
|
printf("final y\n");
|
2012-07-14 00:21:29 +02:00
|
|
|
for (i = 0; i < 16; i++) {
|
2012-11-16 00:14:38 +01:00
|
|
|
for (j = 0; j < 16; j++)
|
|
|
|
printf("%3d ", xd->dst.y_buffer[i * xd->dst.y_stride + j]);
|
|
|
|
printf("\n");
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-11-16 00:14:38 +01:00
|
|
|
printf("\n");
|
|
|
|
printf("final u\n");
|
|
|
|
for (i = 0; i < 8; i++) {
|
|
|
|
for (j = 0; j < 8; j++)
|
|
|
|
printf("%3d ", xd->dst.u_buffer[i * xd->dst.uv_stride + j]);
|
|
|
|
printf("\n");
|
2012-10-22 20:49:00 +02:00
|
|
|
}
|
2012-11-16 00:14:38 +01:00
|
|
|
printf("\n");
|
|
|
|
printf("final v\n");
|
|
|
|
for (i = 0; i < 8; i++) {
|
|
|
|
for (j = 0; j < 8; j++)
|
|
|
|
printf("%3d ", xd->dst.v_buffer[i * xd->dst.uv_stride + j]);
|
|
|
|
printf("\n");
|
2011-02-14 23:18:18 +01:00
|
|
|
}
|
2012-11-16 00:14:38 +01:00
|
|
|
fflush(stdout);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2012-11-16 00:14:38 +01:00
|
|
|
#endif
|
2011-07-20 23:21:24 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2010-09-16 20:08:52 +02:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
static int get_delta_q(vp9_reader *bc, int prev, int *q_update) {
|
2012-07-14 00:21:29 +02:00
|
|
|
int ret_val = 0;
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(bc)) {
|
|
|
|
ret_val = vp9_read_literal(bc, 4);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(bc))
|
2012-07-14 00:21:29 +02:00
|
|
|
ret_val = -ret_val;
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Trigger a quantizer update if the delta-q value has changed */
|
|
|
|
if (ret_val != prev)
|
|
|
|
*q_update = 1;
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
return ret_val;
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef PACKET_TESTING
|
|
|
|
#include <stdio.h>
|
|
|
|
FILE *vpxlog = 0;
|
|
|
|
#endif
|
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
static void set_offsets(VP9D_COMP *pbi, int block_size,
|
|
|
|
int mb_row, int mb_col) {
|
|
|
|
VP9_COMMON *const cm = &pbi->common;
|
|
|
|
MACROBLOCKD *const xd = &pbi->mb;
|
|
|
|
const int mis = cm->mode_info_stride;
|
|
|
|
const int idx = mis * mb_row + mb_col;
|
|
|
|
const int dst_fb_idx = cm->new_fb_idx;
|
|
|
|
const int recon_y_stride = cm->yv12_fb[dst_fb_idx].y_stride;
|
|
|
|
const int recon_uv_stride = cm->yv12_fb[dst_fb_idx].uv_stride;
|
|
|
|
const int recon_yoffset = mb_row * 16 * recon_y_stride + 16 * mb_col;
|
|
|
|
const int recon_uvoffset = mb_row * 8 * recon_uv_stride + 8 * mb_col;
|
|
|
|
|
|
|
|
xd->mode_info_context = cm->mi + idx;
|
|
|
|
xd->mode_info_context->mbmi.sb_type = block_size >> 5;
|
|
|
|
xd->prev_mode_info_context = cm->prev_mi + idx;
|
|
|
|
xd->above_context = cm->above_context + mb_col;
|
|
|
|
xd->left_context = cm->left_context + (mb_row & 3);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
/* Distance of Mb to the various image edges.
|
|
|
|
* These are specified to 8th pel as they are always compared to
|
|
|
|
* values that are in 1/8th pel units
|
|
|
|
*/
|
|
|
|
block_size >>= 4; // in mb units
|
|
|
|
xd->mb_to_top_edge = -((mb_row * 16)) << 3;
|
|
|
|
xd->mb_to_left_edge = -((mb_col * 16) << 3);
|
|
|
|
xd->mb_to_bottom_edge = ((cm->mb_rows - block_size - mb_row) * 16) << 3;
|
|
|
|
xd->mb_to_right_edge = ((cm->mb_cols - block_size - mb_col) * 16) << 3;
|
|
|
|
|
|
|
|
xd->up_available = (mb_row != 0);
|
|
|
|
xd->left_available = (mb_col != 0);
|
|
|
|
|
|
|
|
xd->dst.y_buffer = cm->yv12_fb[dst_fb_idx].y_buffer + recon_yoffset;
|
|
|
|
xd->dst.u_buffer = cm->yv12_fb[dst_fb_idx].u_buffer + recon_uvoffset;
|
|
|
|
xd->dst.v_buffer = cm->yv12_fb[dst_fb_idx].v_buffer + recon_uvoffset;
|
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
static void set_refs(VP9D_COMP *pbi, int block_size,
|
|
|
|
int mb_row, int mb_col) {
|
|
|
|
VP9_COMMON *const cm = &pbi->common;
|
|
|
|
MACROBLOCKD *const xd = &pbi->mb;
|
|
|
|
MODE_INFO *mi = xd->mode_info_context;
|
|
|
|
MB_MODE_INFO *const mbmi = &mi->mbmi;
|
|
|
|
|
|
|
|
if (mbmi->ref_frame > INTRA_FRAME) {
|
|
|
|
int ref_fb_idx, ref_yoffset, ref_uvoffset, ref_y_stride, ref_uv_stride;
|
|
|
|
|
|
|
|
/* Select the appropriate reference frame for this MB */
|
2013-01-15 22:49:44 +01:00
|
|
|
ref_fb_idx = cm->active_ref_idx[mbmi->ref_frame - 1];
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
ref_y_stride = cm->yv12_fb[ref_fb_idx].y_stride;
|
|
|
|
ref_yoffset = mb_row * 16 * ref_y_stride + 16 * mb_col;
|
|
|
|
xd->pre.y_buffer = cm->yv12_fb[ref_fb_idx].y_buffer + ref_yoffset;
|
|
|
|
ref_uv_stride = cm->yv12_fb[ref_fb_idx].uv_stride;
|
|
|
|
ref_uvoffset = mb_row * 8 * ref_uv_stride + 8 * mb_col;
|
|
|
|
xd->pre.u_buffer = cm->yv12_fb[ref_fb_idx].u_buffer + ref_uvoffset;
|
|
|
|
xd->pre.v_buffer = cm->yv12_fb[ref_fb_idx].v_buffer + ref_uvoffset;
|
2012-08-20 23:43:34 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
/* propagate errors from reference frames */
|
|
|
|
xd->corrupted |= cm->yv12_fb[ref_fb_idx].corrupted;
|
2012-08-20 23:43:34 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
if (mbmi->second_ref_frame > INTRA_FRAME) {
|
|
|
|
int second_ref_fb_idx;
|
2011-12-07 22:03:57 +01:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
/* Select the appropriate reference frame for this MB */
|
2013-01-15 22:49:44 +01:00
|
|
|
second_ref_fb_idx = cm->active_ref_idx[mbmi->second_ref_frame - 1];
|
2013-01-06 03:20:25 +01:00
|
|
|
|
|
|
|
xd->second_pre.y_buffer =
|
|
|
|
cm->yv12_fb[second_ref_fb_idx].y_buffer + ref_yoffset;
|
|
|
|
xd->second_pre.u_buffer =
|
|
|
|
cm->yv12_fb[second_ref_fb_idx].u_buffer + ref_uvoffset;
|
|
|
|
xd->second_pre.v_buffer =
|
|
|
|
cm->yv12_fb[second_ref_fb_idx].v_buffer + ref_uvoffset;
|
|
|
|
|
|
|
|
/* propagate errors from reference frames */
|
|
|
|
xd->corrupted |= cm->yv12_fb[second_ref_fb_idx].corrupted;
|
|
|
|
}
|
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
if (mbmi->sb_type) {
|
|
|
|
const int n_mbs = 1 << mbmi->sb_type;
|
|
|
|
const int y_mbs = MIN(n_mbs, cm->mb_rows - mb_row);
|
|
|
|
const int x_mbs = MIN(n_mbs, cm->mb_cols - mb_col);
|
|
|
|
const int mis = cm->mode_info_stride;
|
|
|
|
int x, y;
|
|
|
|
|
|
|
|
for (y = 0; y < y_mbs; y++) {
|
|
|
|
for (x = !y; x < x_mbs; x++) {
|
|
|
|
mi[y * mis + x] = *mi;
|
2012-11-13 00:43:11 +01:00
|
|
|
}
|
2013-01-06 03:20:25 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
/* Decode a row of Superblocks (2x2 region of MBs) */
|
|
|
|
static void decode_sb_row(VP9D_COMP *pbi, VP9_COMMON *pc,
|
|
|
|
int mb_row, MACROBLOCKD *xd,
|
|
|
|
BOOL_DECODER* const bc) {
|
|
|
|
int mb_col;
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
// For a SB there are 2 left contexts, each pertaining to a MB row within
|
|
|
|
vpx_memset(pc->left_context, 0, sizeof(pc->left_context));
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
for (mb_col = 0; mb_col < pc->mb_cols; mb_col += 4) {
|
|
|
|
if (vp9_read(bc, pc->sb64_coded)) {
|
|
|
|
set_offsets(pbi, 64, mb_row, mb_col);
|
2012-10-30 22:51:31 +01:00
|
|
|
vp9_decode_mb_mode_mv(pbi, xd, mb_row, mb_col, bc);
|
2013-01-06 03:20:25 +01:00
|
|
|
set_refs(pbi, 64, mb_row, mb_col);
|
|
|
|
decode_superblock64(pbi, xd, mb_row, mb_col, bc);
|
|
|
|
xd->corrupted |= bool_error(bc);
|
2013-01-10 02:21:28 +01:00
|
|
|
} else {
|
2013-01-06 03:20:25 +01:00
|
|
|
int j;
|
2012-08-06 19:51:20 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
for (j = 0; j < 4; j++) {
|
|
|
|
const int x_idx_sb = (j & 1) << 1, y_idx_sb = j & 2;
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
if (mb_row + y_idx_sb >= pc->mb_rows ||
|
|
|
|
mb_col + x_idx_sb >= pc->mb_cols) {
|
|
|
|
// MB lies outside frame, skip on to next
|
|
|
|
continue;
|
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
xd->sb_index = j;
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
if (vp9_read(bc, pc->sb32_coded)) {
|
|
|
|
set_offsets(pbi, 32, mb_row + y_idx_sb, mb_col + x_idx_sb);
|
|
|
|
vp9_decode_mb_mode_mv(pbi,
|
|
|
|
xd, mb_row + y_idx_sb, mb_col + x_idx_sb, bc);
|
|
|
|
set_refs(pbi, 32, mb_row + y_idx_sb, mb_col + x_idx_sb);
|
|
|
|
decode_superblock32(pbi,
|
|
|
|
xd, mb_row + y_idx_sb, mb_col + x_idx_sb, bc);
|
|
|
|
xd->corrupted |= bool_error(bc);
|
2013-01-08 19:29:22 +01:00
|
|
|
} else {
|
2013-01-06 03:20:25 +01:00
|
|
|
int i;
|
|
|
|
|
|
|
|
// Process the 4 MBs within the SB in the order:
|
|
|
|
// top-left, top-right, bottom-left, bottom-right
|
|
|
|
for (i = 0; i < 4; i++) {
|
|
|
|
const int x_idx = x_idx_sb + (i & 1), y_idx = y_idx_sb + (i >> 1);
|
|
|
|
|
|
|
|
if (mb_row + y_idx >= pc->mb_rows ||
|
|
|
|
mb_col + x_idx >= pc->mb_cols) {
|
|
|
|
// MB lies outside frame, skip on to next
|
|
|
|
continue;
|
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
set_offsets(pbi, 16, mb_row + y_idx, mb_col + x_idx);
|
|
|
|
xd->mb_index = i;
|
|
|
|
vp9_decode_mb_mode_mv(pbi, xd, mb_row + y_idx, mb_col + x_idx, bc);
|
|
|
|
update_blockd_bmi(xd);
|
|
|
|
set_refs(pbi, 16, mb_row + y_idx, mb_col + x_idx);
|
|
|
|
vp9_intra_prediction_down_copy(xd);
|
|
|
|
decode_macroblock(pbi, xd, mb_row, mb_col, bc);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
/* check if the boolean decoder has suffered an error */
|
|
|
|
xd->corrupted |= bool_error(bc);
|
|
|
|
}
|
|
|
|
}
|
2012-08-20 23:43:34 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
static unsigned int read_partition_size(const unsigned char *cx_size) {
|
|
|
|
const unsigned int size =
|
|
|
|
cx_size[0] + (cx_size[1] << 8) + (cx_size[2] << 16);
|
|
|
|
return size;
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
|
|
|
|
2011-08-05 18:27:25 +02:00
|
|
|
static int read_is_valid(const unsigned char *start,
|
|
|
|
size_t len,
|
2012-07-14 00:21:29 +02:00
|
|
|
const unsigned char *end) {
|
|
|
|
return (start + len > start && start + len <= end);
|
2011-08-05 18:27:25 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2012-10-31 01:53:32 +01:00
|
|
|
static void setup_token_decoder(VP9D_COMP *pbi,
|
2012-10-18 01:47:38 +02:00
|
|
|
const unsigned char *cx_data,
|
|
|
|
BOOL_DECODER* const bool_decoder) {
|
2012-10-31 01:53:32 +01:00
|
|
|
VP9_COMMON *pc = &pbi->common;
|
2012-07-14 00:21:29 +02:00
|
|
|
const unsigned char *user_data_end = pbi->Source + pbi->source_sz;
|
|
|
|
const unsigned char *partition;
|
|
|
|
|
|
|
|
ptrdiff_t partition_size;
|
|
|
|
ptrdiff_t bytes_left;
|
|
|
|
|
|
|
|
// Set up pointers to token partition
|
|
|
|
partition = cx_data;
|
|
|
|
bytes_left = user_data_end - partition;
|
|
|
|
partition_size = bytes_left;
|
|
|
|
|
|
|
|
/* Validate the calculated partition length. If the buffer
|
|
|
|
* described by the partition can't be fully read, then restrict
|
|
|
|
* it to the portion that can be (for EC mode) or throw an error.
|
|
|
|
*/
|
|
|
|
if (!read_is_valid(partition, partition_size, user_data_end)) {
|
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
|
|
|
|
"Truncated packet or corrupt partition "
|
|
|
|
"%d length", 1);
|
|
|
|
}
|
|
|
|
|
2012-11-05 23:22:59 +01:00
|
|
|
if (vp9_start_decode(bool_decoder,
|
|
|
|
partition, (unsigned int)partition_size))
|
2012-07-14 00:21:29 +02:00
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
|
|
|
|
"Failed to allocate bool decoder %d", 1);
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
|
|
|
|
2012-10-31 01:53:32 +01:00
|
|
|
static void init_frame(VP9D_COMP *pbi) {
|
|
|
|
VP9_COMMON *const pc = &pbi->common;
|
2012-10-17 23:51:27 +02:00
|
|
|
MACROBLOCKD *const xd = &pbi->mb;
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (pc->frame_type == KEY_FRAME) {
|
2013-01-15 15:43:35 +01:00
|
|
|
vp9_setup_past_independence(pc, xd);
|
2012-07-14 00:21:29 +02:00
|
|
|
/* All buffers are implicitly updated on key frames. */
|
2013-01-15 22:49:44 +01:00
|
|
|
pbi->refresh_frame_flags = (1 << NUM_REF_FRAMES) - 1;
|
2013-01-15 15:43:35 +01:00
|
|
|
} else if (pc->error_resilient_mode) {
|
|
|
|
vp9_setup_past_independence(pc, xd);
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2013-01-15 15:43:35 +01:00
|
|
|
if (pc->frame_type != KEY_FRAME) {
|
2012-07-14 00:21:29 +02:00
|
|
|
if (!pc->use_bilinear_mc_filter)
|
|
|
|
pc->mcomp_filter_type = EIGHTTAP;
|
|
|
|
else
|
|
|
|
pc->mcomp_filter_type = BILINEAR;
|
|
|
|
|
2012-07-28 02:46:33 +02:00
|
|
|
/* To enable choice of different interpolation filters */
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_setup_interp_filters(xd, pc->mcomp_filter_type, pc);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
xd->mode_info_context = pc->mi;
|
2012-08-03 21:31:38 +02:00
|
|
|
xd->prev_mode_info_context = pc->prev_mi;
|
2012-07-14 00:21:29 +02:00
|
|
|
xd->frame_type = pc->frame_type;
|
|
|
|
xd->mode_info_context->mbmi.mode = DC_PRED;
|
|
|
|
xd->mode_info_stride = pc->mode_info_stride;
|
|
|
|
xd->corrupted = 0; /* init without corruption */
|
2011-08-24 20:42:26 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
xd->fullpixel_mask = 0xffffffff;
|
|
|
|
if (pc->full_pixel)
|
|
|
|
xd->fullpixel_mask = 0xfffffff8;
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
|
|
|
|
2012-12-08 01:09:59 +01:00
|
|
|
static void read_coef_probs_common(BOOL_DECODER* const bc,
|
|
|
|
vp9_coeff_probs *coef_probs,
|
|
|
|
int block_types) {
|
2012-07-14 00:21:29 +02:00
|
|
|
int i, j, k, l;
|
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(bc)) {
|
2012-12-08 01:09:59 +01:00
|
|
|
for (i = 0; i < block_types; i++) {
|
2012-10-20 00:35:36 +02:00
|
|
|
for (j = !i; j < COEF_BANDS; j++) {
|
|
|
|
/* NB: This j loop starts from 1 on block type i == 0 */
|
2012-09-10 07:42:35 +02:00
|
|
|
for (k = 0; k < PREV_COEF_CONTEXTS; k++) {
|
|
|
|
if (k >= 3 && ((i == 0 && j == 1) ||
|
|
|
|
(i > 0 && j == 0)))
|
|
|
|
continue;
|
|
|
|
for (l = 0; l < ENTROPY_NODES; l++) {
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_prob *const p = coef_probs[i][j][k] + l;
|
2012-08-03 02:03:14 +02:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read(bc, COEF_UPDATE_PROB)) {
|
2012-09-10 07:42:35 +02:00
|
|
|
*p = read_prob_diff_update(bc, *p);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2012-10-20 00:35:36 +02:00
|
|
|
}
|
|
|
|
}
|
2012-09-10 07:42:35 +02:00
|
|
|
}
|
2012-10-20 00:35:36 +02:00
|
|
|
}
|
2012-09-10 07:42:35 +02:00
|
|
|
|
2012-10-31 01:53:32 +01:00
|
|
|
static void read_coef_probs(VP9D_COMP *pbi, BOOL_DECODER* const bc) {
|
|
|
|
VP9_COMMON *const pc = &pbi->common;
|
2012-08-03 02:03:14 +02:00
|
|
|
|
2012-12-08 01:09:59 +01:00
|
|
|
read_coef_probs_common(bc, pc->fc.coef_probs_4x4, BLOCK_TYPES_4X4);
|
|
|
|
read_coef_probs_common(bc, pc->fc.hybrid_coef_probs_4x4, BLOCK_TYPES_4X4);
|
2012-10-24 21:59:22 +02:00
|
|
|
|
2012-10-20 00:35:36 +02:00
|
|
|
if (pbi->common.txfm_mode != ONLY_4X4) {
|
2012-12-08 01:09:59 +01:00
|
|
|
read_coef_probs_common(bc, pc->fc.coef_probs_8x8, BLOCK_TYPES_8X8);
|
|
|
|
read_coef_probs_common(bc, pc->fc.hybrid_coef_probs_8x8, BLOCK_TYPES_8X8);
|
2012-08-03 02:03:14 +02:00
|
|
|
}
|
2012-10-20 00:35:36 +02:00
|
|
|
if (pbi->common.txfm_mode > ALLOW_8X8) {
|
2012-12-08 01:09:59 +01:00
|
|
|
read_coef_probs_common(bc, pc->fc.coef_probs_16x16, BLOCK_TYPES_16X16);
|
|
|
|
read_coef_probs_common(bc, pc->fc.hybrid_coef_probs_16x16,
|
|
|
|
BLOCK_TYPES_16X16);
|
2012-09-10 07:42:35 +02:00
|
|
|
}
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
if (pbi->common.txfm_mode > ALLOW_16X16) {
|
2012-12-08 01:09:59 +01:00
|
|
|
read_coef_probs_common(bc, pc->fc.coef_probs_32x32, BLOCK_TYPES_32X32);
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
}
|
2012-04-12 18:24:03 +02:00
|
|
|
}
|
|
|
|
|
2012-11-15 21:19:07 +01:00
|
|
|
int vp9_decode_frame(VP9D_COMP *pbi, const unsigned char **p_data_end) {
|
2012-10-18 01:47:38 +02:00
|
|
|
BOOL_DECODER header_bc, residual_bc;
|
2012-10-31 01:53:32 +01:00
|
|
|
VP9_COMMON *const pc = &pbi->common;
|
2012-10-17 23:51:27 +02:00
|
|
|
MACROBLOCKD *const xd = &pbi->mb;
|
2012-07-14 00:21:29 +02:00
|
|
|
const unsigned char *data = (const unsigned char *)pbi->Source;
|
|
|
|
const unsigned char *data_end = data + pbi->source_sz;
|
|
|
|
ptrdiff_t first_partition_length_in_bytes = 0;
|
|
|
|
|
|
|
|
int mb_row;
|
|
|
|
int i, j;
|
|
|
|
int corrupt_tokens = 0;
|
|
|
|
|
2013-01-15 15:43:35 +01:00
|
|
|
// printf("Decoding frame %d\n", pc->current_video_frame);
|
2012-07-14 00:21:29 +02:00
|
|
|
/* start with no corruption of current frame */
|
|
|
|
xd->corrupted = 0;
|
|
|
|
pc->yv12_fb[pc->new_fb_idx].corrupted = 0;
|
|
|
|
|
|
|
|
if (data_end - data < 3) {
|
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
|
|
|
|
"Truncated packet");
|
|
|
|
} else {
|
|
|
|
pc->last_frame_type = pc->frame_type;
|
|
|
|
pc->frame_type = (FRAME_TYPE)(data[0] & 1);
|
|
|
|
pc->version = (data[0] >> 1) & 7;
|
|
|
|
pc->show_frame = (data[0] >> 4) & 1;
|
|
|
|
first_partition_length_in_bytes =
|
|
|
|
(data[0] | (data[1] << 8) | (data[2] << 16)) >> 5;
|
|
|
|
|
|
|
|
if ((data + first_partition_length_in_bytes > data_end
|
|
|
|
|| data + first_partition_length_in_bytes < data))
|
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
|
|
|
|
"Truncated packet or corrupt partition 0 length");
|
|
|
|
|
|
|
|
data += 3;
|
|
|
|
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_setup_version(pc);
|
2011-07-20 23:21:24 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (pc->frame_type == KEY_FRAME) {
|
|
|
|
const int Width = pc->Width;
|
|
|
|
const int Height = pc->Height;
|
|
|
|
|
|
|
|
/* vet via sync code */
|
|
|
|
/* When error concealment is enabled we should only check the sync
|
|
|
|
* code if we have enough bits available
|
|
|
|
*/
|
|
|
|
if (data + 3 < data_end) {
|
|
|
|
if (data[0] != 0x9d || data[1] != 0x01 || data[2] != 0x2a)
|
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_UNSUP_BITSTREAM,
|
|
|
|
"Invalid frame sync code");
|
|
|
|
}
|
|
|
|
|
|
|
|
/* If error concealment is enabled we should only parse the new size
|
|
|
|
* if we have enough data. Otherwise we will end up with the wrong
|
|
|
|
* size.
|
|
|
|
*/
|
|
|
|
if (data + 6 < data_end) {
|
|
|
|
pc->Width = (data[3] | (data[4] << 8)) & 0x3fff;
|
|
|
|
pc->horiz_scale = data[4] >> 6;
|
|
|
|
pc->Height = (data[5] | (data[6] << 8)) & 0x3fff;
|
|
|
|
pc->vert_scale = data[6] >> 6;
|
|
|
|
}
|
|
|
|
data += 7;
|
|
|
|
|
|
|
|
if (Width != pc->Width || Height != pc->Height) {
|
|
|
|
if (pc->Width <= 0) {
|
|
|
|
pc->Width = Width;
|
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
|
|
|
|
"Invalid frame width");
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (pc->Height <= 0) {
|
|
|
|
pc->Height = Height;
|
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_CORRUPT_FRAME,
|
|
|
|
"Invalid frame height");
|
|
|
|
}
|
2010-12-16 16:46:31 +01:00
|
|
|
|
2012-10-31 00:25:53 +01:00
|
|
|
if (vp9_alloc_frame_buffers(pc, pc->Width, pc->Height))
|
2012-07-14 00:21:29 +02:00
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
|
|
|
|
"Failed to allocate frame buffers");
|
|
|
|
}
|
2011-05-02 15:30:51 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if ((!pbi->decoded_key_frame && pc->frame_type != KEY_FRAME) ||
|
|
|
|
pc->Width == 0 || pc->Height == 0) {
|
|
|
|
return -1;
|
|
|
|
}
|
2011-05-02 15:30:51 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
init_frame(pbi);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-11-05 23:22:59 +01:00
|
|
|
if (vp9_start_decode(&header_bc, data,
|
|
|
|
(unsigned int)first_partition_length_in_bytes))
|
2012-07-14 00:21:29 +02:00
|
|
|
vpx_internal_error(&pc->error, VPX_CODEC_MEM_ERROR,
|
|
|
|
"Failed to allocate bool decoder 0");
|
|
|
|
if (pc->frame_type == KEY_FRAME) {
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->clr_type = (YUV_TYPE)vp9_read_bit(&header_bc);
|
|
|
|
pc->clamp_type = (CLAMP_TYPE)vp9_read_bit(&header_bc);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2013-01-15 15:43:35 +01:00
|
|
|
pc->error_resilient_mode = vp9_read_bit(&header_bc);
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Is segmentation enabled */
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->segmentation_enabled = (unsigned char)vp9_read_bit(&header_bc);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (xd->segmentation_enabled) {
|
|
|
|
// Read whether or not the segmentation map is being explicitly
|
|
|
|
// updated this frame.
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->update_mb_segmentation_map = (unsigned char)vp9_read_bit(&header_bc);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
// If so what method will be used.
|
2012-10-12 07:15:33 +02:00
|
|
|
if (xd->update_mb_segmentation_map) {
|
|
|
|
// Which macro block level features are enabled
|
|
|
|
|
|
|
|
// Read the probs used to decode the segment id for each macro
|
|
|
|
// block.
|
|
|
|
for (i = 0; i < MB_FEATURE_TREE_PROBS; i++) {
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->mb_segment_tree_probs[i] = vp9_read_bit(&header_bc) ?
|
|
|
|
(vp9_prob)vp9_read_literal(&header_bc, 8) : 255;
|
2012-10-12 07:15:33 +02:00
|
|
|
}
|
2012-02-16 18:29:54 +01:00
|
|
|
|
2012-10-12 07:15:33 +02:00
|
|
|
// Read the prediction probs needed to decode the segment id
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->temporal_update = (unsigned char)vp9_read_bit(&header_bc);
|
2012-10-12 07:15:33 +02:00
|
|
|
for (i = 0; i < PREDICTION_PROBS; i++) {
|
|
|
|
if (pc->temporal_update) {
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->segment_pred_probs[i] = vp9_read_bit(&header_bc) ?
|
|
|
|
(vp9_prob)vp9_read_literal(&header_bc, 8) : 255;
|
2012-10-12 07:15:33 +02:00
|
|
|
} else {
|
|
|
|
pc->segment_pred_probs[i] = 255;
|
|
|
|
}
|
|
|
|
}
|
2013-01-30 18:30:46 +01:00
|
|
|
|
|
|
|
if (pc->temporal_update) {
|
|
|
|
int count[4];
|
|
|
|
const vp9_prob *p = xd->mb_segment_tree_probs;
|
|
|
|
vp9_prob *p_mod = xd->mb_segment_mispred_tree_probs;
|
|
|
|
|
|
|
|
count[0] = p[0] * p[1];
|
|
|
|
count[1] = p[0] * (256 - p[1]);
|
|
|
|
count[2] = (256 - p[0]) * p[2];
|
|
|
|
count[3] = (256 - p[0]) * (256 - p[2]);
|
|
|
|
|
|
|
|
p_mod[0] = get_binary_prob(count[1], count[2] + count[3]);
|
|
|
|
p_mod[1] = get_binary_prob(count[0], count[2] + count[3]);
|
|
|
|
p_mod[2] = get_binary_prob(count[0] + count[1], count[3]);
|
|
|
|
p_mod[3] = get_binary_prob(count[0] + count[1], count[2]);
|
|
|
|
}
|
2012-10-12 07:15:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
// Is the segment data being updated
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->update_mb_segmentation_data = (unsigned char)vp9_read_bit(&header_bc);
|
2011-11-11 11:10:06 +01:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (xd->update_mb_segmentation_data) {
|
|
|
|
int data;
|
2011-11-11 11:10:06 +01:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->mb_segment_abs_delta = (unsigned char)vp9_read_bit(&header_bc);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-30 06:15:27 +01:00
|
|
|
vp9_clearall_segfeatures(xd);
|
2011-11-03 13:50:09 +01:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
// For each segmentation...
|
|
|
|
for (i = 0; i < MAX_MB_SEGMENTS; i++) {
|
|
|
|
// For each of the segments features...
|
|
|
|
for (j = 0; j < SEG_LVL_MAX; j++) {
|
|
|
|
// Is the feature enabled
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc)) {
|
2012-07-14 00:21:29 +02:00
|
|
|
// Update the feature data and mask
|
2012-10-30 06:15:27 +01:00
|
|
|
vp9_enable_segfeature(xd, i, j);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2012-11-27 20:16:15 +01:00
|
|
|
data = vp9_decode_unsigned_max(&header_bc,
|
|
|
|
vp9_seg_feature_data_max(j));
|
2012-07-14 00:21:29 +02:00
|
|
|
|
|
|
|
// Is the segment data signed..
|
2012-10-30 06:15:27 +01:00
|
|
|
if (vp9_is_segfeature_signed(j)) {
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc))
|
2012-11-27 20:16:15 +01:00
|
|
|
data = -data;
|
2011-11-15 12:13:33 +01:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
} else
|
|
|
|
data = 0;
|
|
|
|
|
2012-10-30 06:15:27 +01:00
|
|
|
vp9_set_segdata(xd, i, j, data);
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
// Read common prediction model status flag probability updates for the
|
|
|
|
// reference frame
|
|
|
|
if (pc->frame_type == KEY_FRAME) {
|
|
|
|
// Set the prediction probabilities to defaults
|
|
|
|
pc->ref_pred_probs[0] = 120;
|
|
|
|
pc->ref_pred_probs[1] = 80;
|
|
|
|
pc->ref_pred_probs[2] = 40;
|
2012-12-14 08:53:11 +01:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
} else {
|
|
|
|
for (i = 0; i < PREDICTION_PROBS; i++) {
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc))
|
|
|
|
pc->ref_pred_probs[i] = (vp9_prob)vp9_read_literal(&header_bc, 8);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
|
|
|
}
|
2012-01-28 13:20:14 +01:00
|
|
|
|
2013-01-06 03:20:25 +01:00
|
|
|
pc->sb64_coded = vp9_read_literal(&header_bc, 8);
|
|
|
|
pc->sb32_coded = vp9_read_literal(&header_bc, 8);
|
2012-08-20 23:43:34 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Read the loop filter level and type */
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->txfm_mode = vp9_read_literal(&header_bc, 2);
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
if (pc->txfm_mode == 3)
|
|
|
|
pc->txfm_mode += vp9_read_bit(&header_bc);
|
2012-10-09 18:18:21 +02:00
|
|
|
if (pc->txfm_mode == TX_MODE_SELECT) {
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->prob_tx[0] = vp9_read_literal(&header_bc, 8);
|
|
|
|
pc->prob_tx[1] = vp9_read_literal(&header_bc, 8);
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
pc->prob_tx[2] = vp9_read_literal(&header_bc, 8);
|
2012-10-09 18:18:21 +02:00
|
|
|
}
|
Improved coding using 8x8 transform
In summary, this commit encompasses a series of changes in attempt to
improve the 8x8 transform based coding to help overall compression
quality, please refer to the detailed commit history below for what
are the rationale underly the series of changes:
a. A frame level flag to indicate if 8x8 transform is used at all.
b. 8x8 transform is not used for key frames and small image size.
c. On inter coded frame, macroblocks using modes B_PRED, SPLIT_MV
and I8X8_PRED are forced to using 4x4 transform based coding, the
rest uses 8x8 transform based coding.
d. Encoder and decoder has the same assumption on the relationship
between prediction modes and transform size, therefore no signaling
is encoded in bitstream.
e. Mode decision process now calculate the rate and distortion scores
using their respective transforms.
Overall test results:
1. HD set
http://www.corp.google.com/~yaowu/no_crawl/t8x8/HD_t8x8_20120206.html
(avg psnr: 3.09% glb psnr: 3.22%, ssim: 3.90%)
2. Cif set:
http://www.corp.google.com/~yaowu/no_crawl/t8x8/cif_t8x8_20120206.html
(avg psnr: -0.03%, glb psnr: -0.02%, ssim: -0.04%)
It should be noted here, as 8x8 transform coding itself is disabled
for cif size clips, the 0.03% loss is purely from the 1 bit/frame
flag overhead on if 8x8 transform is used or not for the frame.
---patch history for future reference---
Patch 1:
this commit tries to select transform size based on macroblock
prediction mode. If the size of a prediction mode is 16x16, then
the macroblock is forced to use 8x8 transform. If the prediction
mode is B_PRED, SPLITMV or I8X8_PRED, then the macroblock is forced
to use 4x4 transform. Tests on the following HD clips showed mixed
results: (all hd clips only used first 100 frames in the test)
http://www.corp.google.com/~yaowu/no_crawl/t8x8/hdmodebased8x8.html
http://www.corp.google.com/~yaowu/no_crawl/t8x8/hdmodebased8x8_log.html
while the results are mixed and overall negative, it is interesting to
see 8x8 helped a few of the clips.
Patch 2:
this patch tries to hard-wire selection of transform size based on
prediction modes without using segmentation to signal the transform size.
encoder and decoder both takes the same assumption that all macroblocks
use 8x8 transform except when prediciton mode is B_PRED, I8X8_PRED or
SPLITMV. Test results are as follows:
http://www.corp.google.com/~yaowu/no_crawl/t8x8/cifmodebase8x8_0125.html
http://www.corp.google.com/~yaowu/no_crawl/t8x8/hdmodebased8x8_0125log.html
Interestingly, by removing the overhead or coding the segmentation, the
results on this limited HD set have turn positive on average.
Patch 3:
this patch disabled the usage of 8x8 transform on key frames, and kept the
logic from patch 2 for inter frames only. test results on HD set turned
decidedly positive with 8x8 transform enabled on inter frame with 16x16
prediction modes: (avg psnr: .81% glb psnr: .82 ssim: .55%)
http://www.corp.google.com/~yaowu/no_crawl/t8x8/hdintermode8x8_0125.html
results on cif set still negative overall
Patch 4:
continued from last patch, but now in mode decision process, the rate and
distortion estimates are computed based on 8x8 transform results for MBs
with modes associated with 8x8 transform. This patch also fixed a problem
related to segment based eob coding when 8x8 transform is used. The patch
significantly improved the results on HD clips:
http://www.corp.google.com/~yaowu/no_crawl/t8x8/hd8x8RDintermode.html
(avg psnr: 2.70% glb psnr: 2.76% ssim: 3.34%)
results on cif also improved, though they are still negative compared to
baseline that uses 4x4 transform only:
http://www.corp.google.com/~yaowu/no_crawl/t8x8/cif8x8RDintermode.html
(avg psnr: -.78% glb psnr: -.86% ssim: -.19%)
Patch 5:
This patch does 3 things:
a. a bunch of decoder bug fixes, encodings and decodings were verified
to have matched recon buffer on a number of encodes on cif size mobile and
hd version of _pedestrian.
b. the patch further improved the rate distortion calculation of MBS that
use 8x8 transform. This provided some further gain on compression.
c. the patch also got the experimental work SEG_LVL_EOB to work with 8x8
transformed macroblock, test results indicates it improves the cif set
but hurt the HD set slightly.
Tests results on HD clips:
http://www.corp.google.com/~yaowu/no_crawl/t8x8/HD_t8x8_20120201.html
(avg psnr: 3.19% glb psnr: 3.30% ssim: 3.93%)
Test results on cif clips:
http://www.corp.google.com/~yaowu/no_crawl/t8x8/cif_t8x8_20120201.html
(avg psnr: -.47% glb psnr: -.51% ssim: +.28%)
Patch 6:
Added a frame level flag to indicate if 8x8 transform is allowed at all.
temporarily the decision is based on frame size, can be optimized later
one. This get the cif results to basically unchanged, with one bit per
frame overhead on both cif and hd clips.
Patch 8:
Rebase and Merge to head by PGW.
Fixed some suspect 4s that look like hey should be 64s in regard
to segmented EOB. Perhaps #defines would be bette.
Bulit and tested without T8x8 enabled and produces unchanged
output.
Patch 9:
Corrected misalligned code/decode of "txfm_mode" bit.
Limited testing for correct encode and decode with
T8x8 configured on derf clips.
Change-Id: I156e1405d25f81579d579dff8ab9af53944ec49c
2012-02-10 01:12:23 +01:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->filter_type = (LOOPFILTERTYPE) vp9_read_bit(&header_bc);
|
|
|
|
pc->filter_level = vp9_read_literal(&header_bc, 6);
|
|
|
|
pc->sharpness_level = vp9_read_literal(&header_bc, 3);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Read in loop filter deltas applied at the MB level based on mode or ref frame. */
|
|
|
|
xd->mode_ref_lf_delta_update = 0;
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->mode_ref_lf_delta_enabled = (unsigned char)vp9_read_bit(&header_bc);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (xd->mode_ref_lf_delta_enabled) {
|
|
|
|
/* Do the deltas need to be updated */
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->mode_ref_lf_delta_update = (unsigned char)vp9_read_bit(&header_bc);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (xd->mode_ref_lf_delta_update) {
|
|
|
|
/* Send update */
|
|
|
|
for (i = 0; i < MAX_REF_LF_DELTAS; i++) {
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc)) {
|
|
|
|
/*sign = vp9_read_bit( &header_bc );*/
|
|
|
|
xd->ref_lf_deltas[i] = (signed char)vp9_read_literal(&header_bc, 6);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc)) /* Apply sign */
|
2012-07-14 00:21:29 +02:00
|
|
|
xd->ref_lf_deltas[i] = xd->ref_lf_deltas[i] * -1;
|
|
|
|
}
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Send update */
|
|
|
|
for (i = 0; i < MAX_MODE_LF_DELTAS; i++) {
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc)) {
|
|
|
|
/*sign = vp9_read_bit( &header_bc );*/
|
|
|
|
xd->mode_lf_deltas[i] = (signed char)vp9_read_literal(&header_bc, 6);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc)) /* Apply sign */
|
2012-07-14 00:21:29 +02:00
|
|
|
xd->mode_lf_deltas[i] = xd->mode_lf_deltas[i] * -1;
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
|
|
|
|
2012-10-18 01:47:38 +02:00
|
|
|
// Dummy read for now
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_read_literal(&header_bc, 2);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2012-10-18 01:47:38 +02:00
|
|
|
setup_token_decoder(pbi, data + first_partition_length_in_bytes,
|
|
|
|
&residual_bc);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
|
|
|
/* Read the default quantizers. */
|
|
|
|
{
|
|
|
|
int Q, q_update;
|
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
Q = vp9_read_literal(&header_bc, QINDEX_BITS);
|
2012-07-14 00:21:29 +02:00
|
|
|
pc->base_qindex = Q;
|
|
|
|
q_update = 0;
|
2012-10-18 01:47:38 +02:00
|
|
|
/* AC 1st order Q = default */
|
|
|
|
pc->y1dc_delta_q = get_delta_q(&header_bc, pc->y1dc_delta_q, &q_update);
|
|
|
|
pc->y2dc_delta_q = get_delta_q(&header_bc, pc->y2dc_delta_q, &q_update);
|
|
|
|
pc->y2ac_delta_q = get_delta_q(&header_bc, pc->y2ac_delta_q, &q_update);
|
|
|
|
pc->uvdc_delta_q = get_delta_q(&header_bc, pc->uvdc_delta_q, &q_update);
|
|
|
|
pc->uvac_delta_q = get_delta_q(&header_bc, pc->uvac_delta_q, &q_update);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
|
|
|
if (q_update)
|
2012-10-31 00:16:28 +01:00
|
|
|
vp9_init_de_quantizer(pbi);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
|
|
|
/* MB level dequantizer setup */
|
|
|
|
mb_init_dequantizer(pbi, &pbi->mb);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Determine if the golden frame or ARF buffer should be updated and how.
|
|
|
|
* For all non key frames the GF and ARF refresh flags and sign bias
|
|
|
|
* flags must be set explicitly.
|
|
|
|
*/
|
2013-01-16 21:19:42 +01:00
|
|
|
if (pc->frame_type == KEY_FRAME) {
|
|
|
|
pc->active_ref_idx[0] = pc->new_fb_idx;
|
|
|
|
pc->active_ref_idx[1] = pc->new_fb_idx;
|
|
|
|
pc->active_ref_idx[2] = pc->new_fb_idx;
|
|
|
|
} else {
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Should the GF or ARF be updated from the current frame */
|
2013-01-15 22:49:44 +01:00
|
|
|
pbi->refresh_frame_flags = vp9_read_literal(&header_bc, NUM_REF_FRAMES);
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2013-01-16 21:19:42 +01:00
|
|
|
/* Select active reference frames */
|
|
|
|
for (i = 0; i < 3; i++) {
|
|
|
|
int ref_frame_num = vp9_read_literal(&header_bc, NUM_REF_FRAMES_LG2);
|
|
|
|
|
|
|
|
pc->active_ref_idx[i] = pc->ref_frame_map[ref_frame_num];
|
|
|
|
}
|
|
|
|
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->ref_frame_sign_bias[GOLDEN_FRAME] = vp9_read_bit(&header_bc);
|
|
|
|
pc->ref_frame_sign_bias[ALTREF_FRAME] = vp9_read_bit(&header_bc);
|
2012-02-16 18:29:54 +01:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Is high precision mv allowed */
|
2012-10-31 22:40:53 +01:00
|
|
|
xd->allow_high_precision_mv = (unsigned char)vp9_read_bit(&header_bc);
|
2012-07-14 00:21:29 +02:00
|
|
|
// Read the type of subpel filter to use
|
2012-10-31 22:40:53 +01:00
|
|
|
if (vp9_read_bit(&header_bc)) {
|
2012-07-18 22:43:01 +02:00
|
|
|
pc->mcomp_filter_type = SWITCHABLE;
|
2012-10-22 23:43:01 +02:00
|
|
|
} else {
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->mcomp_filter_type = vp9_read_literal(&header_bc, 2);
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|
2012-11-07 15:50:25 +01:00
|
|
|
#if CONFIG_COMP_INTERINTRA_PRED
|
|
|
|
pc->use_interintra = vp9_read_bit(&header_bc);
|
|
|
|
#endif
|
2012-07-18 22:43:01 +02:00
|
|
|
/* To enable choice of different interploation filters */
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_setup_interp_filters(xd, pc->mcomp_filter_type, pc);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
|
|
|
|
2013-01-25 20:30:28 +01:00
|
|
|
if (!pc->error_resilient_mode) {
|
|
|
|
pc->refresh_entropy_probs = vp9_read_bit(&header_bc);
|
|
|
|
pc->frame_parallel_decoding_mode = vp9_read_bit(&header_bc);
|
|
|
|
} else {
|
|
|
|
pc->refresh_entropy_probs = 0;
|
|
|
|
pc->frame_parallel_decoding_mode = 1;
|
|
|
|
}
|
2013-01-16 00:57:11 +01:00
|
|
|
pc->frame_context_idx = vp9_read_literal(&header_bc, NUM_FRAME_CONTEXTS_LG2);
|
|
|
|
vpx_memcpy(&pc->fc, &pc->frame_contexts[pc->frame_context_idx],
|
|
|
|
sizeof(pc->fc));
|
2012-07-14 00:21:29 +02:00
|
|
|
|
2012-11-16 17:31:32 +01:00
|
|
|
// Read inter mode probability context updates
|
|
|
|
if (pc->frame_type != KEY_FRAME) {
|
|
|
|
int i, j;
|
|
|
|
for (i = 0; i < INTER_MODE_CONTEXTS; i++) {
|
|
|
|
for (j = 0; j < 4; j++) {
|
|
|
|
if (vp9_read(&header_bc, 252)) {
|
|
|
|
pc->fc.vp9_mode_contexts[i][j] =
|
|
|
|
(vp9_prob)vp9_read_literal(&header_bc, 8);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-12-10 13:38:48 +01:00
|
|
|
#if CONFIG_NEW_MVREF
|
|
|
|
// If Key frame reset mv ref id probabilities to defaults
|
2013-01-15 15:43:35 +01:00
|
|
|
if (pc->frame_type != KEY_FRAME) {
|
2012-12-10 13:38:48 +01:00
|
|
|
// Read any mv_ref index probability updates
|
|
|
|
int i, j;
|
|
|
|
|
|
|
|
for (i = 0; i < MAX_REF_FRAMES; ++i) {
|
|
|
|
// Skip the dummy entry for intra ref frame.
|
|
|
|
if (i == INTRA_FRAME) {
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Read any updates to probabilities
|
|
|
|
for (j = 0; j < MAX_MV_REF_CANDIDATES - 1; ++j) {
|
|
|
|
if (vp9_read(&header_bc, VP9_MVREF_UPDATE_PROB)) {
|
|
|
|
xd->mb_mv_ref_probs[i][j] =
|
|
|
|
(vp9_prob)vp9_read_literal(&header_bc, 8);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (0) {
|
|
|
|
FILE *z = fopen("decodestats.stt", "a");
|
2013-01-15 22:49:44 +01:00
|
|
|
fprintf(z, "%6d F:%d,R:%d,Q:%d\n",
|
2012-07-14 00:21:29 +02:00
|
|
|
pc->current_video_frame,
|
|
|
|
pc->frame_type,
|
2013-01-15 22:49:44 +01:00
|
|
|
pbi->refresh_frame_flags,
|
2012-07-14 00:21:29 +02:00
|
|
|
pc->base_qindex);
|
|
|
|
fclose(z);
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-12-08 01:09:59 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_coef_probs_4x4,
|
|
|
|
pbi->common.fc.coef_probs_4x4);
|
|
|
|
vp9_copy(pbi->common.fc.pre_hybrid_coef_probs_4x4,
|
|
|
|
pbi->common.fc.hybrid_coef_probs_4x4);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_coef_probs_8x8,
|
2012-09-10 07:42:35 +02:00
|
|
|
pbi->common.fc.coef_probs_8x8);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_hybrid_coef_probs_8x8,
|
2012-09-10 07:42:35 +02:00
|
|
|
pbi->common.fc.hybrid_coef_probs_8x8);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_coef_probs_16x16,
|
2012-09-10 07:42:35 +02:00
|
|
|
pbi->common.fc.coef_probs_16x16);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_hybrid_coef_probs_16x16,
|
2012-09-10 07:42:35 +02:00
|
|
|
pbi->common.fc.hybrid_coef_probs_16x16);
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_coef_probs_32x32,
|
|
|
|
pbi->common.fc.coef_probs_32x32);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_ymode_prob, pbi->common.fc.ymode_prob);
|
2012-11-16 00:50:07 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_sb_ymode_prob, pbi->common.fc.sb_ymode_prob);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_copy(pbi->common.fc.pre_uv_mode_prob, pbi->common.fc.uv_mode_prob);
|
|
|
|
vp9_copy(pbi->common.fc.pre_bmode_prob, pbi->common.fc.bmode_prob);
|
|
|
|
vp9_copy(pbi->common.fc.pre_i8x8_mode_prob, pbi->common.fc.i8x8_mode_prob);
|
|
|
|
vp9_copy(pbi->common.fc.pre_sub_mv_ref_prob, pbi->common.fc.sub_mv_ref_prob);
|
|
|
|
vp9_copy(pbi->common.fc.pre_mbsplit_prob, pbi->common.fc.mbsplit_prob);
|
2012-11-07 15:50:25 +01:00
|
|
|
#if CONFIG_COMP_INTERINTRA_PRED
|
|
|
|
pbi->common.fc.pre_interintra_prob = pbi->common.fc.interintra_prob;
|
|
|
|
#endif
|
2012-07-26 22:42:07 +02:00
|
|
|
pbi->common.fc.pre_nmvc = pbi->common.fc.nmvc;
|
2012-12-08 01:09:59 +01:00
|
|
|
vp9_zero(pbi->common.fc.coef_counts_4x4);
|
|
|
|
vp9_zero(pbi->common.fc.hybrid_coef_counts_4x4);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_zero(pbi->common.fc.coef_counts_8x8);
|
|
|
|
vp9_zero(pbi->common.fc.hybrid_coef_counts_8x8);
|
|
|
|
vp9_zero(pbi->common.fc.coef_counts_16x16);
|
|
|
|
vp9_zero(pbi->common.fc.hybrid_coef_counts_16x16);
|
32x32 transform for superblocks.
This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds
code all over the place to wrap that in the bitstream/encoder/decoder/RD.
Some implementation notes (these probably need careful review):
- token range is extended by 1 bit, since the value range out of this
transform is [-16384,16383].
- the coefficients coming out of the FDCT are manually scaled back by
1 bit, or else they won't fit in int16_t (they are 17 bits). Because
of this, the RD error scoring does not right-shift the MSE score by
two (unlike for 4x4/8x8/16x16).
- to compensate for this loss in precision, the quantizer is halved
also. This is currently a little hacky.
- FDCT and IDCT is double-only right now. Needs a fixed-point impl.
- There are no default probabilities for the 32x32 transform yet; I'm
simply using the 16x16 luma ones. A future commit will add newly
generated probabilities for all transforms.
- No ADST version. I don't think we'll add one for this level; if an
ADST is desired, transform-size selection can scale back to 16x16
or lower, and use an ADST at that level.
Additional notes specific to Debargha's DWT/DCT hybrid:
- coefficient scale is different for the top/left 16x16 (DCT-over-DWT)
block than for the rest (DWT pixel differences) of the block. Therefore,
RD error scoring isn't easily scalable between coefficient and pixel
domain. Thus, unfortunately, we need to compute the RD distortion in
the pixel domain until we figure out how to scale these appropriately.
Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
|
|
|
vp9_zero(pbi->common.fc.coef_counts_32x32);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_zero(pbi->common.fc.ymode_counts);
|
2012-11-16 00:50:07 +01:00
|
|
|
vp9_zero(pbi->common.fc.sb_ymode_counts);
|
2012-10-31 22:40:53 +01:00
|
|
|
vp9_zero(pbi->common.fc.uv_mode_counts);
|
|
|
|
vp9_zero(pbi->common.fc.bmode_counts);
|
|
|
|
vp9_zero(pbi->common.fc.i8x8_mode_counts);
|
|
|
|
vp9_zero(pbi->common.fc.sub_mv_ref_counts);
|
|
|
|
vp9_zero(pbi->common.fc.mbsplit_counts);
|
|
|
|
vp9_zero(pbi->common.fc.NMVcount);
|
|
|
|
vp9_zero(pbi->common.fc.mv_ref_ct);
|
2012-11-07 15:50:25 +01:00
|
|
|
#if CONFIG_COMP_INTERINTRA_PRED
|
|
|
|
vp9_zero(pbi->common.fc.interintra_counts);
|
|
|
|
#endif
|
2012-09-10 07:42:35 +02:00
|
|
|
|
2012-10-18 01:47:38 +02:00
|
|
|
read_coef_probs(pbi, &header_bc);
|
2012-02-29 02:11:12 +01:00
|
|
|
|
2013-01-15 22:49:44 +01:00
|
|
|
/* Initialize xd pointers. Any reference should do for xd->pre, so use 0. */
|
|
|
|
vpx_memcpy(&xd->pre, &pc->yv12_fb[pc->active_ref_idx[0]],
|
|
|
|
sizeof(YV12_BUFFER_CONFIG));
|
|
|
|
vpx_memcpy(&xd->dst, &pc->yv12_fb[pc->new_fb_idx],
|
|
|
|
sizeof(YV12_BUFFER_CONFIG));
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
// Create the segmentation map structure and set to 0
|
|
|
|
if (!pc->last_frame_seg_map)
|
|
|
|
CHECK_MEM_ERROR(pc->last_frame_seg_map,
|
|
|
|
vpx_calloc((pc->mb_rows * pc->mb_cols), 1));
|
2010-09-24 00:25:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* set up frame new frame for intra coded blocks */
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_setup_intra_recon(&pc->yv12_fb[pc->new_fb_idx]);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_setup_block_dptrs(xd);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-31 00:25:53 +01:00
|
|
|
vp9_build_block_doffsets(xd);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* clear out the coeff buffer */
|
|
|
|
vpx_memset(xd->qcoeff, 0, sizeof(xd->qcoeff));
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Read the mb_no_coeff_skip flag */
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->mb_no_coeff_skip = (int)vp9_read_bit(&header_bc);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-10-30 22:51:31 +01:00
|
|
|
vp9_decode_mode_mvs_init(pbi, &header_bc);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
vpx_memset(pc->above_context, 0, sizeof(ENTROPY_CONTEXT_PLANES) * pc->mb_cols);
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
/* Decode a row of superblocks */
|
2013-01-06 03:20:25 +01:00
|
|
|
for (mb_row = 0; mb_row < pc->mb_rows; mb_row += 4) {
|
2012-10-18 01:47:38 +02:00
|
|
|
decode_sb_row(pbi, pc, mb_row, xd, &residual_bc);
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
|
|
|
corrupt_tokens |= xd->corrupted;
|
|
|
|
|
|
|
|
/* Collect information about decoder corruption. */
|
|
|
|
/* 1. Check first boolean decoder for errors. */
|
2012-10-31 22:40:53 +01:00
|
|
|
pc->yv12_fb[pc->new_fb_idx].corrupted = bool_error(&header_bc);
|
2012-07-14 00:21:29 +02:00
|
|
|
/* 2. Check the macroblock information */
|
|
|
|
pc->yv12_fb[pc->new_fb_idx].corrupted |= corrupt_tokens;
|
|
|
|
|
|
|
|
if (!pbi->decoded_key_frame) {
|
|
|
|
if (pc->frame_type == KEY_FRAME &&
|
|
|
|
!pc->yv12_fb[pc->new_fb_idx].corrupted)
|
|
|
|
pbi->decoded_key_frame = 1;
|
|
|
|
else
|
|
|
|
vpx_internal_error(&pbi->common.error, VPX_CODEC_CORRUPT_FRAME,
|
|
|
|
"A stream must start with a complete key frame");
|
|
|
|
}
|
2010-12-16 16:46:31 +01:00
|
|
|
|
2013-01-25 20:30:28 +01:00
|
|
|
if (!pc->error_resilient_mode &&
|
|
|
|
!pc->frame_parallel_decoding_mode)
|
2013-01-15 15:43:35 +01:00
|
|
|
vp9_adapt_coef_probs(pc);
|
2012-07-14 00:21:29 +02:00
|
|
|
if (pc->frame_type != KEY_FRAME) {
|
2013-01-25 20:30:28 +01:00
|
|
|
if (!pc->error_resilient_mode &&
|
|
|
|
!pc->frame_parallel_decoding_mode) {
|
2013-01-15 15:43:35 +01:00
|
|
|
vp9_adapt_mode_probs(pc);
|
|
|
|
vp9_adapt_nmv_probs(pc, xd->allow_high_precision_mv);
|
|
|
|
vp9_adapt_mode_context(&pbi->common);
|
|
|
|
}
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
2012-07-14 00:21:29 +02:00
|
|
|
if (pc->refresh_entropy_probs) {
|
2013-01-16 00:57:11 +01:00
|
|
|
vpx_memcpy(&pc->frame_contexts[pc->frame_context_idx], &pc->fc,
|
|
|
|
sizeof(pc->fc));
|
2012-07-14 00:21:29 +02:00
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
|
|
|
|
#ifdef PACKET_TESTING
|
2012-07-14 00:21:29 +02:00
|
|
|
{
|
|
|
|
FILE *f = fopen("decompressor.VP8", "ab");
|
2012-10-18 01:47:38 +02:00
|
|
|
unsigned int size = residual_bc.pos + header_bc.pos + 8;
|
2012-07-14 00:21:29 +02:00
|
|
|
fwrite((void *) &size, 4, 1, f);
|
|
|
|
fwrite((void *) pbi->Source, size, 1, f);
|
|
|
|
fclose(f);
|
|
|
|
}
|
2010-05-18 17:58:33 +02:00
|
|
|
#endif
|
|
|
|
|
2012-11-15 21:19:07 +01:00
|
|
|
/* Find the end of the coded buffer */
|
|
|
|
while (residual_bc.count > CHAR_BIT
|
|
|
|
&& residual_bc.count < VP9_BD_VALUE_SIZE) {
|
|
|
|
residual_bc.count -= CHAR_BIT;
|
|
|
|
residual_bc.user_buffer--;
|
|
|
|
}
|
|
|
|
*p_data_end = residual_bc.user_buffer;
|
2012-07-14 00:21:29 +02:00
|
|
|
return 0;
|
2010-05-18 17:58:33 +02:00
|
|
|
}
|