Google Git
Sign in
aomedia/avm/f11364023ffeb5ebffd3bf84d63ebb5ce7b9cb88/./av2/common/av2_loopfilter.c
blob: 092c4f6300f32b51ffd4a995f0881f8ca4859698 [file] [log] [blame]
/*
* Copyright (c) 2021, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 3-Clause Clear License
* and the Alliance for Open Media Patent License 1.0. If the BSD 3-Clause Clear
* License was not distributed with this source code in the LICENSE file, you
* can obtain it at aomedia.org/license/software-license/bsd-3-c-c/. If the
* Alliance for Open Media Patent License 1.0 was not distributed with this
* source code in the PATENTS file, you can obtain it at
* aomedia.org/license/patent-license/.
*/
#include <math.h>
#include "config/avm_config.h"
#include "config/avm_dsp_rtcd.h"
#include "avm_dsp/avm_dsp_common.h"
#include "avm_mem/avm_mem.h"
#include "avm_ports/mem.h"
#include "av2/common/av2_common_int.h"
#include "av2/common/av2_loopfilter.h"
#include "av2/common/bru.h"
#include "av2/common/reconinter.h"
#include "av2/common/seg_common.h"
#define MAX_SIDE_TABLE 296
// based on int side_threshold = (int)(32 * AVMMAX(0.0444 * q_ind - 2.9936, 0.31
// * q_ind - 39) );
static const int16_t side_thresholds[MAX_SIDE_TABLE] = {
-16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16,
-16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16,
-16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16,
-16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16, -16,
-16, -16, -16, -16, -16, -14, -13, -11, -10, -9, -7, -6, -4,
-3, -2, 0, 0, 2, 3, 5, 6, 7, 9, 10, 12, 13,
15, 16, 17, 19, 20, 22, 23, 24, 26, 27, 29, 30, 32,
33, 34, 36, 37, 39, 40, 42, 43, 44, 46, 47, 49, 50,
51, 53, 54, 56, 57, 59, 60, 61, 63, 64, 66, 67, 69,
70, 71, 73, 74, 76, 77, 78, 80, 81, 83, 84, 86, 87,
88, 90, 91, 93, 94, 96, 101, 111, 120, 130, 140, 150, 160,
170, 180, 190, 200, 210, 220, 230, 240, 249, 259, 269, 279, 289,
299, 309, 319, 329, 339, 349, 359, 368, 378, 388, 398, 408, 418,
428, 438, 448, 458, 468, 478, 488, 497, 507, 517, 527, 537, 547,
557, 567, 577, 587, 597, 607, 616, 626, 636, 646, 656, 666, 676,
686, 696, 706, 716, 726, 736, 745, 755, 765, 775, 785, 795, 805,
815, 825, 835, 845, 855, 864, 874, 884, 894, 904, 914, 924, 934,
944, 954, 964, 974, 984, 993, 1003, 1013, 1023, 1033, 1043, 1053, 1063,
1073, 1083, 1093, 1103, 1112, 1122, 1132, 1142, 1152, 1162, 1172, 1182, 1192,
1202, 1212, 1222, 1232, 1241, 1251, 1261, 1271, 1281, 1291, 1301, 1311, 1321,
1331, 1341, 1351, 1360, 1370, 1380, 1390, 1400, 1410, 1420, 1430, 1440, 1450,
1460, 1470, 1480, 1489, 1499, 1509, 1519, 1529, 1539, 1549, 1559, 1569, 1579,
1589, 1599, 1608, 1618, 1628, 1638, 1648, 1658, 1668, 1678
};
// Function obtains q_threshold from the quantization index.
int df_quant_from_qindex(int q_index, int bit_depth) {
int qstep = ROUND_POWER_OF_TWO(av2_ac_quant_QTX(q_index, 0, 0, bit_depth),
QUANT_TABLE_BITS);
int q_threshold = qstep >> 6;
return q_threshold;
}
// Function obtains side_threshold from the quantization index.
int df_side_from_qindex(int q_index, int bit_depth) {
assert(bit_depth <= 12);
int q_ind = clamp(q_index - 24 * (bit_depth - 8), 0, MAX_SIDE_TABLE - 1);
int side_threshold = side_thresholds[q_ind];
side_threshold =
AVMMAX(side_threshold + (1 << (12 - bit_depth)), 0) >>
(13 - bit_depth); // to avoid rounding down for higher bit depths
return side_threshold;
}
uint16_t av2_get_filter_q(const loop_filter_info_n *lfi_n, const int dir_idx,
int plane, const MB_MODE_INFO *mbmi, int bit_depth) {
const int current_q_index = mbmi->final_qindex_ac[plane];
return df_quant_from_qindex(
current_q_index + lfi_n->q_thr_q_offset[plane][dir_idx], bit_depth);
}
uint16_t av2_get_filter_side(const loop_filter_info_n *lfi_n, const int dir_idx,
int plane, const MB_MODE_INFO *mbmi,
int bit_depth) {
const int current_q_index = mbmi->final_qindex_ac[plane];
return df_side_from_qindex(
current_q_index + lfi_n->side_thr_q_offset[plane][dir_idx], bit_depth);
}
void av2_loop_filter_init(AV2_COMMON *cm) {
struct loopfilter *lf = &cm->lf;
lf->combine_vert_horz_lf = 1;
}
// Update the loop filter for the current frame.
// This should be called before loop_filter_rows(),
// av2_loop_filter_frame() calls this function directly.
void av2_loop_filter_frame_init(AV2_COMMON *cm, int plane_start,
int plane_end) {
int q_ind[MAX_MB_PLANE][NUM_EDGE_DIRS], side_ind[MAX_MB_PLANE][NUM_EDGE_DIRS];
int plane;
// n_shift is the multiplier for lf_deltas
// the multiplier is 1 for when filter_lvl is between 0 and 31;
// 2 when filter_lvl is between 32 and 63
loop_filter_info_n *const lfi = &cm->lf_info;
for (int dir = 0; dir < NUM_EDGE_DIRS; ++dir) {
q_ind[0][dir] = cm->lf.delta_q_luma[dir] * DF_DELTA_SCALE;
side_ind[0][dir] = cm->lf.delta_side_luma[dir] * DF_DELTA_SCALE;
}
q_ind[1][0] = q_ind[1][1] = cm->lf.delta_q_u * DF_DELTA_SCALE;
side_ind[1][0] = side_ind[1][1] = cm->lf.delta_side_u * DF_DELTA_SCALE;
q_ind[2][0] = q_ind[2][1] = cm->lf.delta_q_v * DF_DELTA_SCALE;
side_ind[2][0] = side_ind[2][1] = cm->lf.delta_side_v * DF_DELTA_SCALE;
assert(plane_start >= AVM_PLANE_Y);
assert(plane_end <= MAX_MB_PLANE);
for (plane = plane_start; plane < plane_end; plane++) {
if (plane == 0 && !cm->lf.apply_deblocking_filter[0] &&
!cm->lf.apply_deblocking_filter[1])
break;
else if (plane == 1 && !cm->lf.apply_deblocking_filter_u)
continue;
else if (plane == 2 && !cm->lf.apply_deblocking_filter_v)
continue;
for (int dir = 0; dir < 2; ++dir) {
lfi->q_thr_q_offset[plane][dir] = q_ind[plane][dir];
lfi->side_thr_q_offset[plane][dir] = side_ind[plane][dir];
}
}
}
// Returns the starting mi location of chroma reference block for the current
// mbmi, by setting `chroma_mi_row_start` and `chroma_mi_col_start`.
static void get_chroma_start_location(const MB_MODE_INFO *mbmi,
TREE_TYPE tree_type,
int *chroma_mi_row_start,
int *chroma_mi_col_start) {
assert(tree_type == SHARED_PART || tree_type == CHROMA_PART);
if (tree_type == SHARED_PART) {
*chroma_mi_row_start = mbmi->chroma_ref_info.mi_row_chroma_base;
*chroma_mi_col_start = mbmi->chroma_ref_info.mi_col_chroma_base;
} else {
*chroma_mi_row_start = mbmi->chroma_mi_row_start;
*chroma_mi_col_start = mbmi->chroma_mi_col_start;
}
}
// Returns true if we are at the transform boundary.
static bool is_tu_edge_helper(TX_SIZE tx_size, EDGE_DIR edge_dir,
int relative_row, int relative_col) {
assert(tx_size != TX_INVALID);
assert(relative_row >= 0);
assert(relative_col >= 0);
const int relative_coord =
(edge_dir == VERT_EDGE) ? relative_col : relative_row;
const uint32_t tu_mask = edge_dir == VERT_EDGE
? tx_size_wide_unit[tx_size] - 1
: tx_size_high_unit[tx_size] - 1;
return !(relative_coord & tu_mask);
}
// Structure to store the transform block related information.
typedef struct {
int row_offset; // row starting offset
int col_offset; // column starting offset
TX_PARTITION_TYPE tx_partition_type;
TX_SIZE tx_size;
} tx_info_t;
static TX_SIZE get_transform_size(const MACROBLOCKD *const xd,
const MB_MODE_INFO *const mbmi,
const EDGE_DIR edge_dir, const int mi_row,
const int mi_col, const int plane,
const TREE_TYPE tree_type,
const struct macroblockd_plane *plane_ptr,
bool *tu_edge, tx_info_t *tx_info,
int *tx_m_partition_size) {
assert(mbmi != NULL);
const BLOCK_SIZE bsize_base =
get_bsize_base_from_tree_type(mbmi, tree_type, plane);
if (xd && xd->lossless[mbmi->segment_id]) {
TX_SIZE tx_size = get_lossless_tx_size(xd, mbmi, plane);
int mi_row_start = mbmi->mi_row_start;
int mi_col_start = mbmi->mi_col_start;
if (plane != AVM_PLANE_Y) {
get_chroma_start_location(mbmi, tree_type, &mi_row_start, &mi_col_start);
}
*tu_edge = is_tu_edge_helper(
tx_size, edge_dir, (mi_row - mi_row_start) >> plane_ptr->subsampling_y,
(mi_col - mi_col_start) >> plane_ptr->subsampling_x);
assert(IMPLIES((plane != AVM_PLANE_Y), (*tu_edge == 1)));
tx_size = (edge_dir == VERT_EDGE) ? txsize_horz_map[tx_size]
: txsize_vert_map[tx_size];
return tx_size;
}
const int plane_type = av2_get_sdp_idx(tree_type);
TX_SIZE tx_size = TX_INVALID;
if (plane != AVM_PLANE_Y) {
tx_size = av2_get_max_uv_txsize(bsize_base, plane_ptr->subsampling_x,
plane_ptr->subsampling_y);
int chroma_mi_row_start;
int chroma_mi_col_start;
get_chroma_start_location(mbmi, tree_type, &chroma_mi_row_start,
&chroma_mi_col_start);
*tu_edge = is_tu_edge_helper(
tx_size, edge_dir,
(mi_row - chroma_mi_row_start) >> plane_ptr->subsampling_y,
(mi_col - chroma_mi_col_start) >> plane_ptr->subsampling_x);
}
if (plane == AVM_PLANE_Y && !mbmi->skip_txfm[SHARED_PART]) {
const BLOCK_SIZE sb_type = mbmi->sb_type[plane_type];
const int blk_row = mi_row - mbmi->mi_row_start;
const int blk_col = mi_col - mbmi->mi_col_start;
assert(blk_row >= 0);
assert(blk_col >= 0);
int txp_index = is_inter_block(mbmi, SHARED_PART)
? av2_get_txb_size_index(sb_type, blk_row, blk_col)
: 0;
const TX_PARTITION_TYPE partition = mbmi->tx_partition_type[txp_index];
tx_info->tx_partition_type = partition;
const TX_SIZE max_tx_size = max_txsize_rect_lookup[sb_type];
if (partition == TX_PARTITION_HORZ5 || partition == TX_PARTITION_VERT5) {
if (tx_m_partition_size != NULL) {
if ((edge_dir == VERT_EDGE && mi_size_wide[mbmi->sb_type[0]] <= 8 &&
partition == TX_PARTITION_VERT5) ||
(edge_dir == HORZ_EDGE && mi_size_high[mbmi->sb_type[0]] <= 8 &&
partition == TX_PARTITION_HORZ5)) {
*tx_m_partition_size = (edge_dir == VERT_EDGE)
? block_size_wide[mbmi->sb_type[0]]
: block_size_high[mbmi->sb_type[0]];
}
}
TXB_POS_INFO txb_pos = { { 0 }, { 0 }, 0 };
TX_SIZE sub_txs[MAX_TX_PARTITIONS] = { 0 };
assert(xd != NULL);
get_tx_partition_sizes(partition, max_tx_size, &txb_pos, sub_txs,
xd->error_info);
int mi_blk_row = blk_row & 0xf;
int mi_blk_col = blk_col & 0xf;
*tu_edge = false; // Default. May be updated below.
int txb_idx;
for (txb_idx = 0; txb_idx < txb_pos.n_partitions; ++txb_idx) {
TX_SIZE sub_tx = sub_txs[txb_idx];
int txh = tx_size_high_unit[sub_tx];
int txw = tx_size_wide_unit[sub_tx];
if (mi_blk_row >= txb_pos.row_offset[txb_idx] &&
mi_blk_row < txb_pos.row_offset[txb_idx] + txh &&
mi_blk_col >= txb_pos.col_offset[txb_idx] &&
mi_blk_col < txb_pos.col_offset[txb_idx] + txw) {
if ((edge_dir == VERT_EDGE &&
mi_blk_col == txb_pos.col_offset[txb_idx]) ||
(edge_dir == HORZ_EDGE &&
mi_blk_row == txb_pos.row_offset[txb_idx])) {
*tu_edge = true;
}
break;
}
}
assert(txb_pos.n_partitions > 1);
assert(txb_idx < txb_pos.n_partitions);
TX_SIZE tmp_tx_size = sub_txs[txb_idx];
tx_info->row_offset = txb_pos.row_offset[txb_idx];
tx_info->col_offset = txb_pos.col_offset[txb_idx];
tx_info->tx_size = tmp_tx_size;
assert(tmp_tx_size < TX_SIZES_ALL);
tx_size = tmp_tx_size;
} else {
tx_size = get_tx_partition_one_size(partition, max_tx_size);
*tu_edge = is_tu_edge_helper(tx_size, edge_dir, blk_row, blk_col);
}
}
if (plane == AVM_PLANE_Y && mbmi->skip_txfm[SHARED_PART]) {
const BLOCK_SIZE sb_type = mbmi->sb_type[plane_type];
tx_size = max_txsize_rect_lookup[sb_type];
*tu_edge = is_tu_edge_helper(tx_size, edge_dir, mi_row - mbmi->mi_row_start,
mi_col - mbmi->mi_col_start);
}
assert(tx_size < TX_SIZES_ALL);
// since in case of chrominance or non-square transform need to convert
// transform size into transform size in particular direction.
// for vertical edge, filter direction is horizontal, for horizontal
// edge, filter direction is vertical.
tx_size = (edge_dir == VERT_EDGE) ? txsize_horz_map[tx_size]
: txsize_vert_map[tx_size];
return tx_size;
}
typedef struct AV2_DEBLOCKING_PARAMETERS {
// Length of the filter applied to the outer edge.
uint32_t filter_length_neg;
uint32_t filter_length_pos;
// Thresholds.
uint16_t q_threshold;
uint16_t side_threshold;
} AV2_DEBLOCKING_PARAMETERS;
static uint32_t get_pu_starting_cooord(const MB_MODE_INFO *const mbmi,
int plane, TREE_TYPE tree_type,
int scale_horz, int scale_vert,
EDGE_DIR edge_dir) {
uint32_t pu_starting_mi;
const bool vert_edge = (edge_dir == VERT_EDGE);
if (plane == AVM_PLANE_Y) {
pu_starting_mi = vert_edge ? mbmi->mi_col_start : mbmi->mi_row_start;
} else {
int chroma_mi_row_start;
int chroma_mi_col_start;
get_chroma_start_location(mbmi, tree_type, &chroma_mi_row_start,
&chroma_mi_col_start);
pu_starting_mi = vert_edge ? chroma_mi_col_start : chroma_mi_row_start;
}
const uint32_t pu_stating_coord_luma = pu_starting_mi * MI_SIZE;
const int scale = vert_edge ? scale_horz : scale_vert;
return pu_stating_coord_luma >> scale;
}
// Check whether current block is TIP mode
static AVM_INLINE void check_tip_edge(const MB_MODE_INFO *const mbmi,
const int scale, TX_SIZE *ts,
int32_t *tip_edge,
int enable_tip_refinemv) {
const bool is_tip_mode = is_tip_ref_frame(mbmi->ref_frame[0]);
if (is_tip_mode) {
*tip_edge = 1;
const BLOCK_SIZE bsize = mbmi->sb_type[AVM_PLANE_Y];
const int bw = block_size_wide[bsize];
const int bh = block_size_high[bsize];
const int sub_pu_size_y =
get_unit_bsize_for_tip_ref(TIP_FRAME_AS_REF, bw, bh,
enable_tip_refinemv) == BLOCK_16X16
? 16
: 8;
const int sub_pu_size = scale ? sub_pu_size_y >> 1 : sub_pu_size_y;
const int tip_ts = sub_pu_size == 16 ? TX_16X16
: sub_pu_size == 8 ? TX_8X8
: TX_4X4;
*ts = tip_ts;
}
}
// Check whether current block is OPFL mode
static AVM_INLINE void check_opfl_edge(const AV2_COMMON *const cm,
const int plane, const MACROBLOCKD *xd,
const MB_MODE_INFO *const mbmi,
const int scale, TX_SIZE *ts,
int32_t *opfl_edge) {
const bool is_opfl_mode = opfl_allowed_cur_pred_mode(cm, xd, mbmi);
(void)plane;
if (is_opfl_mode) {
*opfl_edge = 1;
const BLOCK_SIZE bsize = mbmi->sb_type[AVM_PLANE_Y];
const int bw = block_size_wide[bsize];
const int bh = block_size_high[bsize];
const int use_4x4 = is_tip_ref_frame(mbmi->ref_frame[0]) ? 0 : 1;
const int sub_pu_size_y =
opfl_get_subblock_size(bw, bh, AVM_PLANE_Y, use_4x4);
if (sub_pu_size_y == OF_MIN_BSIZE) {
*ts = TX_4X4;
} else {
const int sub_pu_size = scale ? sub_pu_size_y >> 1 : sub_pu_size_y;
*ts = (sub_pu_size == OF_MIN_BSIZE) ? TX_4X4 : TX_8X8;
}
}
}
// Check whether current block is RFMV mode
static AVM_INLINE void check_rfmv_edge(const AV2_COMMON *const cm,
const MB_MODE_INFO *const mbmi,
const EDGE_DIR edge_dir, const int scale,
TX_SIZE *ts, int32_t *rfmv_edge) {
const int tip_ref_frame = is_tip_ref_frame(mbmi->ref_frame[0]);
int is_rfmv_mode = mbmi->refinemv_flag && !tip_ref_frame;
(void)cm;
if (is_rfmv_mode && default_refinemv_modes(mbmi))
is_rfmv_mode &= (mbmi->comp_group_idx == 0 &&
mbmi->interinter_comp.type == COMPOUND_AVERAGE);
if (is_rfmv_mode) {
*rfmv_edge = 1;
const BLOCK_SIZE bsize = mbmi->sb_type[AVM_PLANE_Y];
const int bw = block_size_wide[bsize];
const int bh = block_size_high[bsize];
if (bw >= REFINEMV_SUBBLOCK_WIDTH && bh >= REFINEMV_SUBBLOCK_HEIGHT) {
*ts = scale ? TX_8X8 : TX_16X16;
} else if (bw >= REFINEMV_SUBBLOCK_WIDTH &&
bh == (REFINEMV_SUBBLOCK_HEIGHT >> 1)) {
if (edge_dir == VERT_EDGE) {
*ts = scale ? TX_8X8 : TX_16X16;
} else {
*ts = scale ? TX_4X4 : TX_8X8;
}
} else if (bw == (REFINEMV_SUBBLOCK_WIDTH >> 1) &&
bh >= REFINEMV_SUBBLOCK_HEIGHT) {
if (edge_dir == VERT_EDGE) {
*ts = scale ? TX_4X4 : TX_8X8;
} else {
*ts = scale ? TX_8X8 : TX_16X16;
}
}
}
}
// Check whether current block is sub-prediction mode
static AVM_INLINE void check_sub_pu_edge(
const AV2_COMMON *const cm, const MACROBLOCKD *const xd,
const MB_MODE_INFO *const mbmi, const int plane, TREE_TYPE tree_type,
const int scale_horz, const int scale_vert, const EDGE_DIR edge_dir,
const uint32_t coord, TX_SIZE *ts, int32_t *sub_pu_edge,
int *tx_m_partition_size) {
if (!cm->features.allow_lf_sub_pu) return;
const int is_inter = is_inter_block(mbmi, tree_type);
if (!is_inter) return;
int temp_edge = 0;
TX_SIZE temp_ts = 0;
int scale = edge_dir == VERT_EDGE ? scale_horz : scale_vert;
check_tip_edge(mbmi, scale, &temp_ts, &temp_edge,
cm->seq_params.enable_tip_refinemv);
if (!temp_edge)
check_opfl_edge(cm, plane, xd, mbmi, scale, &temp_ts, &temp_edge);
if (!temp_edge)
check_rfmv_edge(cm, mbmi, edge_dir, scale, &temp_ts, &temp_edge);
if (temp_edge) {
const int sub_pu_size =
edge_dir == VERT_EDGE ? tx_size_wide[temp_ts] : tx_size_high[temp_ts];
const int tx_size =
edge_dir == VERT_EDGE ? tx_size_wide[*ts] : tx_size_high[*ts];
if (sub_pu_size <= tx_size) {
const uint32_t sub_pu_masks = edge_dir == VERT_EDGE
? tx_size_wide[temp_ts] - 1
: tx_size_high[temp_ts] - 1;
const uint32_t pu_starting_coord = get_pu_starting_cooord(
mbmi, plane, tree_type, scale_horz, scale_vert, edge_dir);
if (coord) {
assert(coord >= pu_starting_coord);
const uint32_t relative_coord = coord - pu_starting_coord;
*sub_pu_edge = (relative_coord & sub_pu_masks) ? (0) : (1);
} else {
*sub_pu_edge = 1;
}
if (*tx_m_partition_size == 16 && sub_pu_size == 8) {
*ts = TX_4X4;
} else if (*tx_m_partition_size == 32 && sub_pu_size == 16) {
*ts = TX_8X8;
} else if (*sub_pu_edge) {
*ts = temp_ts;
}
}
}
}
// Returns pointer to appropriate 'mi' with 'mi_grid_base', which contains
// information about current coding block and given current 'x'/'y' location,
// 'plane', 'region_type' etc. The `mi_row` and `mi_col` corresponding to
// 'x'/'y' location are also set.
// Note that, this function is required because, for chroma plane in particular,
// the actual 'mi' location maybe at an offset from the mi_row/mi_col.
MB_MODE_INFO **get_mi_location(const AV2_COMMON *const cm, int scale_horz,
int scale_vert, uint32_t x, uint32_t y,
int plane, int *mi_row, int *mi_col) {
const int this_mi_row = (y << scale_vert) >> MI_SIZE_LOG2;
const int this_mi_col = (x << scale_horz) >> MI_SIZE_LOG2;
MB_MODE_INFO **this_mi = cm->mi_params.mi_grid_base +
this_mi_row * cm->mi_params.mi_stride + this_mi_col;
*mi_row = this_mi_row;
*mi_col = this_mi_col;
return get_mi_location_from_collocated_mi(cm, this_mi, plane);
}
// Returns the remaining mi units for which the same loop-filter parameters as
// the current unit hold. This is used to skip the redundant setting of
// loop-filter parameters.
static int get_remaining_mi_size(const MB_MODE_INFO *mbmi,
const tx_info_t *tx_info, EDGE_DIR edge_dir,
uint32_t x, uint32_t y, int plane,
TREE_TYPE tree_type, int ss_x, int ss_y) {
const bool vert_edge = (edge_dir == VERT_EDGE);
const int mi_cur_coord =
vert_edge ? (y >> MI_SIZE_LOG2) : (x >> MI_SIZE_LOG2);
int mi_pu_start;
if (plane == PLANE_TYPE_Y) {
mi_pu_start = vert_edge ? mbmi->mi_row_start : mbmi->mi_col_start;
} else {
int mi_row_start_uv;
int mi_col_start_uv;
get_chroma_start_location(mbmi, tree_type, &mi_row_start_uv,
&mi_col_start_uv);
const int mi_pu_start_y = vert_edge ? mi_row_start_uv : mi_col_start_uv;
const int scale = vert_edge ? ss_y : ss_x;
mi_pu_start = mi_pu_start_y >> scale;
}
// All the transform block sizes with a given prediction block are of the same
// size for transform partition types other than TX_PARTITION_HORZ5,
// TX_PARTITION_VERT5. Thus, when the previous and current transform partition
// types are neither HORZ5 / VERT5, the filter length (which is determined
// based on the minimum transform block size from both sides of the edge) does
// not change for different transform blocks of a given prediction block and
// the same loop filter parameters hold for the complete prediction block.
// However, for transform partition types HORZ5 / VERT5, same loop filter
// parameters hold for the transform block.
int mi_offset;
if (tx_info->tx_partition_type == TX_PARTITION_HORZ5 && vert_edge) {
mi_offset = tx_info->row_offset + tx_size_high_unit[tx_info->tx_size];
} else if (tx_info->tx_partition_type == TX_PARTITION_VERT5 && !vert_edge) {
mi_offset = tx_info->col_offset + tx_size_wide_unit[tx_info->tx_size];
} else {
const BLOCK_SIZE plane_bsize = get_mb_plane_block_size_from_tree_type(
mbmi, tree_type, plane, ss_x, ss_y);
assert(plane_bsize < BLOCK_SIZES_ALL);
mi_offset =
vert_edge ? mi_size_high[plane_bsize] : mi_size_wide[plane_bsize];
}
// The remaining mi units is the number of mi units to the right/bottom of the
// current unit within a prediction/transform block.
const int mi_remain_size = mi_pu_start + mi_offset - mi_cur_coord;
return mi_remain_size;
}
// Return TX_SIZE from get_transform_size(), so it is plane and direction
// aware
static TX_SIZE set_lpf_parameters(
AV2_DEBLOCKING_PARAMETERS *const params, uint32_t prev_x, uint32_t prev_y,
AV2_COMMON *const cm, const MACROBLOCKD *const xd, const EDGE_DIR edge_dir,
const uint32_t x, const uint32_t y, const int plane,
const struct macroblockd_plane *const plane_ptr, TX_SIZE *tx_size,
int *mi_size_min) {
if (*mi_size_min) {
*mi_size_min -= 1;
return *tx_size;
}
// reset to initial values
params->filter_length_neg = 0;
params->filter_length_pos = 0;
TREE_TYPE tree_type = SHARED_PART;
// no deblocking is required
const uint32_t width = plane_ptr->dst.width;
const uint32_t height = plane_ptr->dst.height;
if ((width <= x) || (height <= y)) {
// just return the smallest transform unit size
return TX_4X4;
}
const int scale_horz = plane_ptr->subsampling_x;
const int scale_vert = plane_ptr->subsampling_y;
int mi_row;
int mi_col;
MB_MODE_INFO **mi = get_mi_location(cm, scale_horz, scale_vert, x, y, plane,
&mi_row, &mi_col);
const MB_MODE_INFO *mbmi = mi[0];
// If current mbmi is not correctly setup, return an invalid value to stop
// filtering. One example is that if this tile is not coded, then its mbmi
// it not set up.
if (mbmi == NULL) return TX_INVALID;
// SDP is always disabled for monochrome.
assert(IMPLIES(cm->seq_params.enable_sdp, !cm->seq_params.monochrome));
const bool is_sdp_eligible =
cm->seq_params.enable_sdp && mbmi->region_type == INTRA_REGION;
if (is_sdp_eligible) {
tree_type = (plane == AVM_PLANE_Y) ? LUMA_PART : CHROMA_PART;
}
const int plane_type = is_sdp_eligible && plane > 0;
bool tu_edge;
tx_info_t tx_info = { 0, 0, TX_PARTITION_NONE, TX_4X4 };
int tx_m_partition_size = 0;
TX_SIZE ts =
get_transform_size(xd, mi[0], edge_dir, mi_row, mi_col, plane, tree_type,
plane_ptr, &tu_edge, &tx_info, &tx_m_partition_size);
const BLOCK_SIZE superblock_size = get_plane_block_size(
cm->sb_size, plane_ptr->subsampling_x, plane_ptr->subsampling_y);
assert(superblock_size < BLOCK_SIZES_ALL);
int mi_size_prev = mi_size_high[superblock_size];
{
const uint32_t coord = (VERT_EDGE == edge_dir) ? (x) : (y);
int32_t sub_pu_edge = 0;
check_sub_pu_edge(cm, xd, mbmi, plane, tree_type, scale_horz, scale_vert,
edge_dir, coord, &ts, &sub_pu_edge, &tx_m_partition_size);
if (!tu_edge && !sub_pu_edge) return ts;
if (cm->seq_params.disable_loopfilters_across_tiles) {
if (edge_dir == VERT_EDGE)
if (is_vert_tile_boundary(&cm->tiles, mi_col)) return ts;
if (edge_dir == HORZ_EDGE)
if (is_horz_tile_boundary(&cm->tiles, mi_row)) return ts;
}
if (cm->bru.enabled) {
if (mbmi->sb_active_mode != BRU_ACTIVE_SB) {
avm_internal_error(&cm->error, AVM_CODEC_ERROR,
"Invalid BRU activity in deblocking: only active SB "
"can be filtered");
}
}
// prepare outer edge parameters. deblock the edge if it's an edge of a TU
{
const uint32_t curr_q = av2_get_filter_q(&cm->lf_info, edge_dir, plane,
mbmi, cm->seq_params.bit_depth);
const uint32_t curr_side = av2_get_filter_side(
&cm->lf_info, edge_dir, plane, mbmi, cm->seq_params.bit_depth);
const int curr_skipped =
mbmi->skip_txfm[plane_type] && is_inter_block(mbmi, tree_type);
if (coord) {
{
assert(prev_x <= x && prev_y <= y);
int pv_row;
int pv_col;
MB_MODE_INFO **mi_prev_ptr =
get_mi_location(cm, scale_horz, scale_vert, prev_x, prev_y, plane,
&pv_row, &pv_col);
const MB_MODE_INFO *const mi_prev = mi_prev_ptr[0];
TREE_TYPE prev_tree_type = SHARED_PART;
const bool is_prev_sdp_eligible =
cm->seq_params.enable_sdp && mi_prev->region_type == INTRA_REGION;
// With SDP in inter frames, the tree type of current block can be
// different with previous block, so we can't copy the tree type of
// current block to previous block, and we need to fetch the tree type
// of a previous block.
if (is_prev_sdp_eligible) {
prev_tree_type = (plane == AVM_PLANE_Y) ? LUMA_PART : CHROMA_PART;
}
bool prev_tu_edge;
tx_info_t prev_tx_info = { 0, 0, TX_PARTITION_NONE, TX_4X4 };
int pv_tx_m_partition_size = 0;
TX_SIZE pv_ts = get_transform_size(
xd, mi_prev, edge_dir, pv_row, pv_col, plane, prev_tree_type,
plane_ptr, &prev_tu_edge, &prev_tx_info, &pv_tx_m_partition_size);
mi_size_prev = get_remaining_mi_size(mi_prev, &prev_tx_info, edge_dir,
x, y, plane, prev_tree_type,
scale_horz, scale_vert);
int32_t pv_sub_pu_edge = 0;
check_sub_pu_edge(cm, xd, mi_prev, plane, prev_tree_type, scale_horz,
scale_vert, edge_dir, 0, &pv_ts, &pv_sub_pu_edge,
&pv_tx_m_partition_size);
const uint32_t pv_q = av2_get_filter_q(
&cm->lf_info, edge_dir, plane, mi_prev, cm->seq_params.bit_depth);
const uint32_t pv_side = av2_get_filter_side(
&cm->lf_info, edge_dir, plane, mi_prev, cm->seq_params.bit_depth);
const uint32_t pu_starting_coord = get_pu_starting_cooord(
mbmi, plane, tree_type, scale_horz, scale_vert, edge_dir);
const bool pu_edge = (coord == pu_starting_coord);
// if the current and the previous blocks are skipped,
// deblock the edge if the edge belongs to a PU's edge only.
const BLOCK_SIZE block64_size = get_plane_block_size(
BLOCK_64X64, plane_ptr->subsampling_x, plane_ptr->subsampling_y);
const int vert_sb_mask = block_size_high[block64_size] - 1;
int horz_superblock_edge =
(HORZ_EDGE == edge_dir) && !(coord & vert_sb_mask);
const unsigned int hor_sb_size = block_size_wide[superblock_size];
int vert_tile_edge = 0;
for (int i = 1; i < cm->tiles.cols; ++i) {
if ((cm->tiles.col_start_sb[i] * hor_sb_size == coord) &&
(VERT_EDGE == edge_dir)) {
vert_tile_edge = 1;
}
}
if (((curr_q && curr_side) || (pv_q && pv_side)) &&
(!curr_skipped || sub_pu_edge || pu_edge)) {
TX_SIZE clipped_ts = ts;
if (!plane) {
if (((VERT_EDGE == edge_dir) && (width < x + 16)) ||
((HORZ_EDGE == edge_dir) && (height < y + 16))) {
// make sure filtering does not get outside the frame size
clipped_ts = AVMMIN(clipped_ts, TX_16X16);
}
} else {
if (((VERT_EDGE == edge_dir) && (width < x + 8)) ||
((HORZ_EDGE == edge_dir) && (height < y + 8))) {
// make sure filtering does not get outside the frame size
clipped_ts = AVMMIN(clipped_ts, TX_8X8);
}
}
const TX_SIZE min_ts = AVMMIN(clipped_ts, pv_ts);
if (TX_4X4 >= min_ts) {
params->filter_length_neg = 4;
params->filter_length_pos = 4;
} else if (TX_8X8 == min_ts) {
if (plane != 0) {
if (horz_superblock_edge || vert_tile_edge) {
params->filter_length_neg = 6;
params->filter_length_pos = 8;
} else {
params->filter_length_neg = 8;
params->filter_length_pos = 8;
}
} else {
params->filter_length_neg = 8;
params->filter_length_pos = 8;
}
} else if (TX_16X16 == min_ts) {
params->filter_length_neg = 14;
params->filter_length_pos = 14;
// No wide filtering for chroma plane
if (plane != 0) {
if (horz_superblock_edge || vert_tile_edge) {
params->filter_length_neg = 6;
params->filter_length_pos = 10;
} else {
params->filter_length_neg = 10;
params->filter_length_pos = 10;
}
}
} else {
if (horz_superblock_edge || vert_tile_edge) {
if (plane != 0) {
params->filter_length_neg = 6;
params->filter_length_pos = 10;
} else {
params->filter_length_neg = 14;
params->filter_length_pos = 18;
}
} else {
params->filter_length_neg = 18;
params->filter_length_pos = 18;
// No wide filtering for chroma plane
if (plane != 0) {
params->filter_length_neg = 10;
params->filter_length_pos = 10;
}
}
}
// update the level if the current block is skipped,
// but the previous one is not
params->q_threshold = (curr_q && pv_q) ? (curr_q + pv_q + 1) >> 1
: (curr_q) ? (curr_q)
: (pv_q);
params->side_threshold = curr_side && pv_side
? (curr_side + pv_side + 1) >> 1
: (curr_side) ? (curr_side)
: (pv_side);
if (sub_pu_edge && !tu_edge) {
params->q_threshold >>= SUB_PU_THR_SHIFT;
params->side_threshold >>= SUB_PU_THR_SHIFT;
}
}
}
}
}
}
const int mi_size_cur = get_remaining_mi_size(
mbmi, &tx_info, edge_dir, x, y, plane, tree_type, scale_horz, scale_vert);
*mi_size_min = AVMMIN(mi_size_prev, mi_size_cur) - 1;
return ts;
}
// Extract if the block is lossless or not based on the mbmi map of the frame
static uint8_t get_lossless_flag(
AV2_COMMON *const cm, uint32_t x, uint32_t y, uint32_t scale_horz,
uint32_t scale_vert, int plane,
const struct macroblockd_plane *const plane_ptr) {
if (!cm->features.has_lossless_segment) return 0;
const uint32_t width = plane_ptr->dst.width;
const uint32_t height = plane_ptr->dst.height;
if ((width <= x) || (height <= y)) {
return 0;
}
int mi_row;
int mi_col;
MB_MODE_INFO **this_mi = get_mi_location(cm, scale_horz, scale_vert, x, y,
plane, &mi_row, &mi_col);
return this_mi[0] ? cm->features.lossless_segment[this_mi[0]->segment_id] : 0;
}
void av2_filter_block_plane_vert(AV2_COMMON *const cm,
const MACROBLOCKD *const xd, const int plane,
const MACROBLOCKD_PLANE *const plane_ptr,
const uint32_t mi_row, const uint32_t mi_col) {
if (!plane && !cm->lf.apply_deblocking_filter[0]) return;
const int mib_size = cm->mib_size;
const uint32_t scale_horz = plane_ptr->subsampling_x;
const uint32_t scale_vert = plane_ptr->subsampling_y;
if (cm->bru.enabled) {
MB_MODE_INFO **mi =
cm->mi_params.mi_grid_base + mi_row * cm->mi_params.mi_stride + mi_col;
if (mi[0]->sb_active_mode != BRU_ACTIVE_SB) {
return;
}
}
uint16_t *const dst_ptr = plane_ptr->dst.buf;
const int dst_stride = plane_ptr->dst.stride;
const int y_range = (mib_size >> scale_vert);
const int x_range = (mib_size >> scale_horz);
AV2_DEBLOCKING_PARAMETERS params_buf[MAX_MIB_SIZE];
TX_SIZE tx_size_buf[MAX_MIB_SIZE] = { 0 };
int mi_size_min_height_buf[MAX_MIB_SIZE] = { 0 };
for (int y = 0; y < y_range; y++) {
uint16_t *p = dst_ptr + y * MI_SIZE * dst_stride;
uint32_t prev_x =
(mi_col == 0) ? 0 : ((mi_col - 1) * MI_SIZE) >> scale_horz;
AV2_DEBLOCKING_PARAMETERS *params = params_buf;
TX_SIZE *tx_size = tx_size_buf;
int *mi_size_min_height = mi_size_min_height_buf;
for (int x = 0; x < x_range;) {
// inner loop always filter vertical edges in a MI block. If MI size
// is 8x8, it will filter the vertical edge aligned with a 8x8 block.
// If 4x4 transform is used, it will then filter the internal edge
// aligned with a 4x4 block
const uint32_t curr_x = ((mi_col * MI_SIZE) >> scale_horz) + x * MI_SIZE;
const uint32_t curr_y = ((mi_row * MI_SIZE) >> scale_vert) + y * MI_SIZE;
uint32_t advance_units;
TX_SIZE cur_tx_size = TX_4X4;
cur_tx_size = set_lpf_parameters(params, prev_x, curr_y, cm, xd,
VERT_EDGE, curr_x, curr_y, plane,
plane_ptr, tx_size, mi_size_min_height);
if (cur_tx_size == TX_INVALID) {
params->filter_length_neg = 0;
params->filter_length_pos = 0;
cur_tx_size = TX_4X4;
}
const avm_bit_depth_t bit_depth = cm->seq_params.bit_depth;
bool is_lossless_current_block = get_lossless_flag(
cm, curr_x, curr_y, scale_horz, scale_vert, plane, plane_ptr);
bool is_lossless_prev_block = get_lossless_flag(
cm, prev_x, curr_y, scale_horz, scale_vert, plane, plane_ptr);
bool skip_deblock_lossless =
is_lossless_current_block && is_lossless_prev_block;
int do_filter = 1;
if (cm->seq_params.disable_loopfilters_across_tiles) {
if (is_vert_tile_boundary(&cm->tiles, mi_col + (x << scale_horz)))
do_filter = 0;
}
if (do_filter) {
if (!skip_deblock_lossless &&
(params->filter_length_neg || params->filter_length_pos)) {
if (!(is_lossless_prev_block || is_lossless_current_block)) {
avm_highbd_lpf_vertical_generic(
p, dst_stride, params->filter_length_neg,
params->filter_length_pos, &params->q_threshold,
&params->side_threshold, bit_depth, is_lossless_prev_block,
is_lossless_current_block);
} else {
avm_highbd_lpf_vertical_generic_c(
p, dst_stride, params->filter_length_neg,
params->filter_length_pos, &params->q_threshold,
&params->side_threshold, bit_depth, is_lossless_prev_block,
is_lossless_current_block);
}
}
}
// advance the destination pointer
prev_x = curr_x;
*tx_size = cur_tx_size;
advance_units = tx_size_wide_unit[cur_tx_size];
x += advance_units;
p += advance_units * MI_SIZE;
params += advance_units;
tx_size += advance_units;
mi_size_min_height += advance_units;
}
}
}
void av2_filter_block_plane_horz(AV2_COMMON *const cm,
const MACROBLOCKD *const xd, const int plane,
const MACROBLOCKD_PLANE *const plane_ptr,
const uint32_t mi_row, const uint32_t mi_col) {
if (!plane && !cm->lf.apply_deblocking_filter[1]) return;
const int mib_size = cm->mib_size;
const uint32_t scale_horz = plane_ptr->subsampling_x;
const uint32_t scale_vert = plane_ptr->subsampling_y;
if (cm->bru.enabled) {
MB_MODE_INFO **mi =
cm->mi_params.mi_grid_base + mi_row * cm->mi_params.mi_stride + mi_col;
if (mi[0]->sb_active_mode != BRU_ACTIVE_SB) {
return;
}
}
uint16_t *const dst_ptr = plane_ptr->dst.buf;
const int dst_stride = plane_ptr->dst.stride;
const int y_range = (mib_size >> scale_vert);
const int x_range = (mib_size >> scale_horz);
AV2_DEBLOCKING_PARAMETERS params_buf[MAX_MIB_SIZE];
TX_SIZE tx_size_buf[MAX_MIB_SIZE] = { 0 };
int mi_size_min_width_buf[MAX_MIB_SIZE] = { 0 };
for (int x = 0; x < x_range; x++) {
uint16_t *p = dst_ptr + x * MI_SIZE;
uint32_t prev_y =
(mi_row == 0) ? 0 : ((mi_row - 1) * MI_SIZE) >> scale_vert;
AV2_DEBLOCKING_PARAMETERS *params = params_buf;
TX_SIZE *tx_size = tx_size_buf;
int *mi_size_min_width = mi_size_min_width_buf;
for (int y = 0; y < y_range;) {
// inner loop always filter vertical edges in a MI block. If MI size
// is 8x8, it will first filter the vertical edge aligned with a 8x8
// block. If 4x4 transform is used, it will then filter the internal
// edge aligned with a 4x4 block
const uint32_t curr_x = ((mi_col * MI_SIZE) >> scale_horz) + x * MI_SIZE;
const uint32_t curr_y = ((mi_row * MI_SIZE) >> scale_vert) + y * MI_SIZE;
uint32_t advance_units;
TX_SIZE cur_tx_size = TX_4X4;
cur_tx_size = set_lpf_parameters(params, curr_x, prev_y, cm, xd,
HORZ_EDGE, curr_x, curr_y, plane,
plane_ptr, tx_size, mi_size_min_width);
if (cur_tx_size == TX_INVALID) {
params->filter_length_neg = 0;
params->filter_length_pos = 0;
cur_tx_size = TX_4X4;
}
bool is_lossless_current_block = get_lossless_flag(
cm, curr_x, curr_y, scale_horz, scale_vert, plane, plane_ptr);
bool is_lossless_prev_block = get_lossless_flag(
cm, curr_x, prev_y, scale_horz, scale_vert, plane, plane_ptr);
bool skip_deblock_lossless =
is_lossless_current_block && is_lossless_prev_block;
int do_filter = 1;
if (cm->seq_params.disable_loopfilters_across_tiles) {
if (is_horz_tile_boundary(&cm->tiles, mi_row + (y << scale_vert)))
do_filter = 0;
}
if (do_filter) {
const avm_bit_depth_t bit_depth = cm->seq_params.bit_depth;
if (!skip_deblock_lossless &&
(params->filter_length_neg || params->filter_length_pos)) {
if (!(is_lossless_current_block || is_lossless_prev_block)) {
avm_highbd_lpf_horizontal_generic(
p, dst_stride, params->filter_length_neg,
params->filter_length_pos, &params->q_threshold,
&params->side_threshold, bit_depth, is_lossless_prev_block,
is_lossless_current_block);
} else {
avm_highbd_lpf_horizontal_generic_c(
p, dst_stride, params->filter_length_neg,
params->filter_length_pos, &params->q_threshold,
&params->side_threshold, bit_depth, is_lossless_prev_block,
is_lossless_current_block);
}
}
}
// advance the destination pointer
prev_y = curr_y;
*tx_size = cur_tx_size;
advance_units = tx_size_high_unit[cur_tx_size];
y += advance_units;
p += advance_units * dst_stride * MI_SIZE;
params += advance_units;
tx_size += advance_units;
mi_size_min_width += advance_units;
}
}
}
static void loop_filter_rows(YV12_BUFFER_CONFIG *frame_buffer, AV2_COMMON *cm,
MACROBLOCKD *xd, int start, int stop,
int plane_start, int plane_end) {
struct macroblockd_plane *pd = xd->plane;
const int col_start = 0;
const int col_end = cm->mi_params.mi_cols;
int mi_row, mi_col;
int plane;
const int mib_size = cm->mib_size;
for (plane = plane_start; plane < plane_end; plane++) {
if (plane == 0 && !(cm->lf.apply_deblocking_filter[0]) &&
!(cm->lf.apply_deblocking_filter[1]))
break;
else if (plane == 1 && !(cm->lf.apply_deblocking_filter_u))
continue;
else if (plane == 2 && !(cm->lf.apply_deblocking_filter_v))
continue;
if (cm->lf.combine_vert_horz_lf) {
// filter all vertical and horizontal edges in every super block
for (mi_row = start; mi_row < stop; mi_row += mib_size) {
for (mi_col = col_start; mi_col < col_end; mi_col += mib_size) {
// filter vertical edges
av2_setup_dst_planes(pd, frame_buffer, mi_row, mi_col, plane,
plane + 1, NULL);
av2_filter_block_plane_vert(cm, xd, plane, &pd[plane], mi_row,
mi_col);
// filter horizontal edges
if (mi_col - mib_size >= 0) {
av2_setup_dst_planes(pd, frame_buffer, mi_row, mi_col - mib_size,
plane, plane + 1, NULL);
av2_filter_block_plane_horz(cm, xd, plane, &pd[plane], mi_row,
mi_col - mib_size);
}
}
// filter horizontal edges
av2_setup_dst_planes(pd, frame_buffer, mi_row, mi_col - mib_size, plane,
plane + 1, NULL);
av2_filter_block_plane_horz(cm, xd, plane, &pd[plane], mi_row,
mi_col - mib_size);
}
} else {
// filter all vertical edges in every 128x128 super block
for (mi_row = start; mi_row < stop; mi_row += mib_size) {
for (mi_col = col_start; mi_col < col_end; mi_col += mib_size) {
av2_setup_dst_planes(pd, frame_buffer, mi_row, mi_col, plane,
plane + 1, NULL);
av2_filter_block_plane_vert(cm, xd, plane, &pd[plane], mi_row,
mi_col);
}
}
// filter all horizontal edges in every 128x128 super block
for (mi_row = start; mi_row < stop; mi_row += mib_size) {
for (mi_col = col_start; mi_col < col_end; mi_col += mib_size) {
av2_setup_dst_planes(pd, frame_buffer, mi_row, mi_col, plane,
plane + 1, NULL);
av2_filter_block_plane_horz(cm, xd, plane, &pd[plane], mi_row,
mi_col);
}
}
}
}
}
void av2_loop_filter_frame(YV12_BUFFER_CONFIG *frame, AV2_COMMON *cm,
MACROBLOCKD *xd, int plane_start, int plane_end,
int partial_frame) {
if (cm->bridge_frame_info.is_bridge_frame) {
return;
}
int start_mi_row, end_mi_row, mi_rows_to_filter;
start_mi_row = 0;
mi_rows_to_filter = cm->mi_params.mi_rows;
if (partial_frame && cm->mi_params.mi_rows > 8) {
start_mi_row = cm->mi_params.mi_rows >> 1;
start_mi_row &= 0xfffffff8;
mi_rows_to_filter = AVMMAX(cm->mi_params.mi_rows / 8, 8);
}
end_mi_row = start_mi_row + mi_rows_to_filter;
av2_loop_filter_frame_init(cm, plane_start, plane_end);
loop_filter_rows(frame, cm, xd, start_mi_row, end_mi_row, plane_start,
plane_end);
}
// Set TIP filter length
static AVM_INLINE void set_tip_filter_length(
AV2_COMMON *cm, const int plane, const int subsampling_x,
const int subsampling_y, const int blk_sz, const int edge_dir,
const unsigned int coord, int *filter_length_neg, int *filter_length_pos) {
const BLOCK_SIZE superblock_size =
get_plane_block_size(cm->sb_size, subsampling_x, subsampling_y);
const BLOCK_SIZE block64_size =
get_plane_block_size(BLOCK_64X64, subsampling_x, subsampling_y);
const int vert_sb_mask = block_size_high[block64_size] - 1;
int horz_superblock_edge = (HORZ_EDGE == edge_dir) && !(coord & vert_sb_mask);
const unsigned int hor_sb_size = block_size_wide[superblock_size];
int vert_tile_edge = 0;
for (int i = 1; i < cm->tiles.cols; ++i) {
if ((cm->tiles.col_start_sb[i] * hor_sb_size == coord) &&
(VERT_EDGE == edge_dir)) {
vert_tile_edge = 1;
}
}
if (4 >= blk_sz) {
*filter_length_neg = 4;
*filter_length_pos = 4;
} else if (8 == blk_sz) {
if ((plane != 0) && (horz_superblock_edge || vert_tile_edge)) {
*filter_length_neg = 6;
*filter_length_pos = 8;
} else {
*filter_length_neg = 8;
*filter_length_pos = 8;
}
} else if (16 == blk_sz) {
*filter_length_neg = 14;
*filter_length_pos = 14;
if (plane != 0) {
if (horz_superblock_edge || vert_tile_edge) {
*filter_length_neg = 6;
*filter_length_pos = 10;
} else {
*filter_length_neg = 10;
*filter_length_pos = 10;
}
}
}
}
// Apply loop filtering on TIP plane
AVM_INLINE void loop_filter_tip_plane(AV2_COMMON *cm, const int plane,
uint16_t *dst, const int dst_stride,
const int plane_w, const int plane_h) {
// retrieve filter parameters
loop_filter_info_n *const lfi = &cm->lf_info;
const uint16_t q_horz = lfi->tip_q_thr[plane][HORZ_EDGE];
const uint16_t side_horz = lfi->tip_side_thr[plane][HORZ_EDGE];
const uint16_t q_vert = lfi->tip_q_thr[plane][VERT_EDGE];
const uint16_t side_vert = lfi->tip_side_thr[plane][VERT_EDGE];
const int bit_depth = cm->seq_params.bit_depth;
int sub_bw = get_unit_bsize_for_tip_frame(
cm->features.tip_frame_mode, cm->tip_interp_filter,
cm->seq_params.enable_tip_refinemv) == BLOCK_16X16
? 16
: 8;
int sub_bh = sub_bw;
int subsampling_x = 0;
int subsampling_y = 0;
if (plane > 0) {
subsampling_x = cm->seq_params.subsampling_x;
subsampling_y = cm->seq_params.subsampling_y;
sub_bw >>= subsampling_x;
sub_bh >>= subsampling_y;
}
// start filtering
const int h = plane_h;
const int w = plane_w;
for (int j = 0; j < h; j += 4) {
uint16_t *p = dst + j * dst_stride;
for (int i = 0; i < w; i += sub_bw) {
// filter vertical boundary
if (cm->seq_params.disable_loopfilters_across_tiles) {
if (is_vert_tile_boundary(&cm->tiles,
(i << subsampling_x) >> MI_SIZE_LOG2)) {
p += sub_bw;
continue;
}
}
if (i > 0) {
int filter_length_neg = 0;
int filter_length_pos = 0;
set_tip_filter_length(cm, plane, subsampling_x, subsampling_y, sub_bw,
VERT_EDGE, i, &filter_length_neg,
&filter_length_pos);
avm_highbd_lpf_vertical_generic(p, dst_stride, filter_length_neg,
filter_length_pos, &q_vert, &side_vert,
bit_depth, 0, 0);
}
p += sub_bw;
}
}
for (int i = 0; i < w; i += 4) {
uint16_t *p = dst + i;
for (int j = 0; j < h; j += sub_bh) {
// filter horizontal boundary
if (cm->seq_params.disable_loopfilters_across_tiles) {
if (is_horz_tile_boundary(&cm->tiles,
(j << subsampling_y) >> MI_SIZE_LOG2)) {
p += sub_bh * dst_stride;
continue;
}
}
if (j > 0) {
int filter_length_neg = 0;
int filter_length_pos = 0;
set_tip_filter_length(cm, plane, subsampling_x, subsampling_y, sub_bh,
HORZ_EDGE, j, &filter_length_neg,
&filter_length_pos);
avm_highbd_lpf_horizontal_generic(p, dst_stride, filter_length_neg,
filter_length_pos, &q_horz,
&side_horz, bit_depth, 0, 0);
}
p += sub_bh * dst_stride;
}
}
}
// setup dst buffer for each color component
static AVM_INLINE void setup_tip_dst_plane(struct buf_2d *dst, uint16_t *src,
int width, int height, int stride,
int tpl_row, int tpl_col,
const struct scale_factors *scale,
int subsampling_x,
int subsampling_y) {
const int x = tpl_col >> subsampling_x;
const int y = tpl_row >> subsampling_y;
dst->buf = src + scaled_buffer_offset(x, y, stride, scale);
dst->buf0 = src;
dst->width = width;
dst->height = height;
dst->stride = stride;
}
// setup dst buffer
AVM_INLINE void setup_tip_dst_planes(AV2_COMMON *const cm, MACROBLOCKD *xd,
const int plane, const int tpl_row,
const int tpl_col) {
const YV12_BUFFER_CONFIG *src = &cm->tip_ref.tip_frame->buf;
TIP_PLANE *const pd = &xd->tip_plane[plane];
int is_uv = 0;
int subsampling_x = 0;
int subsampling_y = 0;
if (plane > 0) {
is_uv = 1;
subsampling_x = cm->seq_params.subsampling_x;
subsampling_y = cm->seq_params.subsampling_y;
}
setup_tip_dst_plane(&pd->dst, src->buffers[plane], src->widths[is_uv],
src->heights[is_uv], src->strides[is_uv], tpl_row,
tpl_col, NULL, subsampling_x, subsampling_y);
}
// Initialize TIP lf parameters
void init_tip_lf_parameter(struct AV2Common *cm, int plane_start,
int plane_end) {
if (!cm->lf.apply_deblocking_filter_tip) return;
int q_ind[MAX_MB_PLANE], side_ind[MAX_MB_PLANE];
loop_filter_info_n *const lfi = &cm->lf_info;
const int tip_delta_scale = DF_DELTA_SCALE;
const int tip_delta_luma = cm->lf.tip_delta;
const int tip_delta_chroma = cm->lf.tip_delta;
const int base_qindex = cm->quant_params.base_qindex;
const int u_ac_delta_q = cm->quant_params.u_ac_delta_q;
const int v_ac_delta_q = cm->quant_params.v_ac_delta_q;
q_ind[0] = side_ind[0] = base_qindex + tip_delta_luma * tip_delta_scale;
q_ind[1] = side_ind[1] = base_qindex + u_ac_delta_q +
cm->seq_params.base_uv_ac_delta_q +
tip_delta_chroma * tip_delta_scale;
q_ind[2] = side_ind[2] = base_qindex + v_ac_delta_q +
cm->seq_params.base_uv_ac_delta_q +
tip_delta_chroma * tip_delta_scale;
assert(plane_start >= AVM_PLANE_Y);
assert(plane_end <= MAX_MB_PLANE);
int plane;
for (plane = plane_start; plane < plane_end; plane++) {
for (int dir = 0; dir < 2; ++dir) {
const int q_ind_plane = q_ind[plane];
const int side_ind_plane = side_ind[plane];
const int q_thr =
df_quant_from_qindex(q_ind_plane, cm->seq_params.bit_depth);
const int side_thr =
df_side_from_qindex(side_ind_plane, cm->seq_params.bit_depth);
lfi->tip_q_thr[plane][dir] = q_thr;
lfi->tip_side_thr[plane][dir] = side_thr;
}
}
}
// Apply loop filtering on TIP frame
void loop_filter_tip_frame(struct AV2Common *cm, MACROBLOCKD *xd,
int plane_start, int plane_end) {
if (!cm->lf.apply_deblocking_filter_tip) return;
for (int plane = plane_start; plane < plane_end; ++plane) {
setup_tip_dst_planes(cm, xd, plane, 0, 0);
TIP_PLANE *const tip = &xd->tip_plane[plane];
struct buf_2d *const dst_buf = &tip->dst;
uint16_t *const dst = dst_buf->buf;
const int dst_stride = dst_buf->stride;
loop_filter_tip_plane(cm, plane, dst, dst_stride, dst_buf->width,
dst_buf->height);
}
}