blob: 6e5b39b485f98cce4787192154fcc63ee99a103c [file] [log] [blame] [edit]
/*
* 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/.
*/
#ifndef AOM_AV1_COMMON_RECONINTER_H_
#define AOM_AV1_COMMON_RECONINTER_H_
#include "av1/common/av1_common_int.h"
#include "av1/common/convolve.h"
#include "av1/common/filter.h"
#include "av1/common/warped_motion.h"
#include "aom/aom_integer.h"
#if CONFIG_WEDGE_MOD_EXT
#include "av1/encoder/block.h"
#endif
// Work out how many pixels off the edge of a reference frame we're allowed
// to go when forming an inter prediction.
// The outermost row/col of each referernce frame is extended by
// (AOM_BORDER_IN_PIXELS >> subsampling) pixels, but we need to keep
// at least AOM_INTERP_EXTEND pixels within that to account for filtering.
//
// We have to break this up into two macros to keep both clang-format and
// tools/lint-hunks.py happy.
#define AOM_LEFT_TOP_MARGIN_PX(subsampling) \
((AOM_BORDER_IN_PIXELS >> subsampling) - AOM_INTERP_EXTEND)
#define AOM_LEFT_TOP_MARGIN_SCALED(subsampling) \
(AOM_LEFT_TOP_MARGIN_PX(subsampling) << SCALE_SUBPEL_BITS)
#ifdef __cplusplus
extern "C" {
#endif
#if !CONFIG_WEDGE_MOD_EXT
#define MAX_WEDGE_TYPES 16
#endif
#if CONFIG_WEDGE_MOD_EXT
#define MAX_WEDGE_SIZE_LOG2 6 // 64x64
#else
#define MAX_WEDGE_SIZE_LOG2 5 // 32x32
#endif
#define MAX_WEDGE_SIZE (1 << MAX_WEDGE_SIZE_LOG2)
#define MAX_WEDGE_SQUARE (MAX_WEDGE_SIZE * MAX_WEDGE_SIZE)
#define WEDGE_WEIGHT_BITS 6
#define WEDGE_NONE -1
#if CONFIG_WEDGE_MOD_EXT
static const int wedge_angle_dist_2_index[WEDGE_ANGLES][NUM_WEDGE_DIST] = {
{ -1, 0, 1, 2 }, // WEDGE_0
{ 3, 4, 5, 6 }, // WEDGE_14
{ 7, 8, 9, 10 }, // WEDGE_27
{ 11, 12, 13, 14 }, // WEDGE_45
{ 15, 16, 17, 18 }, // WEDGE_63
{ -1, 19, 20, 21 }, // WEDGE_90
{ 22, 23, 24, 25 }, // WEDGE_117
{ 26, 27, 28, 29 }, // WEDGE_135
{ 30, 31, 32, 33 }, // WEDGE_153
{ 34, 35, 36, 37 }, // WEDGE_166
{ -1, 38, 39, 40 }, // WEDGE_180
{ -1, 41, 42, 43 }, // WEDGE_194
{ -1, 44, 45, 46 }, // WEDGE_207
{ -1, 47, 48, 49 }, // WEDGE_225
{ -1, 50, 51, 52 }, // WEDGE_243
{ -1, 53, 54, 55 }, // WEDGE_270
{ -1, 56, 57, 58 }, // WEDGE_297
{ -1, 59, 60, 61 }, // WEDGE_315
{ -1, 62, 63, 64 }, // WEDGE_333
{ -1, 65, 66, 67 }, // WEDGE_346
};
static const int wedge_index_2_angle[MAX_WEDGE_TYPES] = {
WEDGE_0, WEDGE_0, WEDGE_0, // WEDGE_0
WEDGE_14, WEDGE_14, WEDGE_14, WEDGE_14, // WEDGE_14
WEDGE_27, WEDGE_27, WEDGE_27, WEDGE_27, // WEDGE_27
WEDGE_45, WEDGE_45, WEDGE_45, WEDGE_45, // WEDGE_45
WEDGE_63, WEDGE_63, WEDGE_63, WEDGE_63, // WEDGE_63
WEDGE_90, WEDGE_90, WEDGE_90, // WEDGE_90
WEDGE_117, WEDGE_117, WEDGE_117, WEDGE_117, // WEDGE_117
WEDGE_135, WEDGE_135, WEDGE_135, WEDGE_135, // WEDGE_135
WEDGE_153, WEDGE_153, WEDGE_153, WEDGE_153, // WEDGE_153
WEDGE_166, WEDGE_166, WEDGE_166, WEDGE_166, // WEDGE_166
WEDGE_180, WEDGE_180, WEDGE_180, // WEDGE_180
WEDGE_194, WEDGE_194, WEDGE_194, // WEDGE_194
WEDGE_207, WEDGE_207, WEDGE_207, // WEDGE_207
WEDGE_225, WEDGE_225, WEDGE_225, // WEDGE_225
WEDGE_243, WEDGE_243, WEDGE_243, // WEDGE_243
WEDGE_270, WEDGE_270, WEDGE_270, // WEDGE_270
WEDGE_297, WEDGE_297, WEDGE_297, // WEDGE_297
WEDGE_315, WEDGE_315, WEDGE_315, // WEDGE_315
WEDGE_333, WEDGE_333, WEDGE_333, // WEDGE_333
WEDGE_346, WEDGE_346, WEDGE_346 // WEDGE_346
};
static const int wedge_index_2_dist[MAX_WEDGE_TYPES] = {
1, 2, 3, // WEDGE_0
0, 1, 2, 3, // WEDGE_14
0, 1, 2, 3, // WEDGE_27
0, 1, 2, 3, // WEDGE_45
0, 1, 2, 3, // WEDGE_63
1, 2, 3, // WEDGE_90
0, 1, 2, 3, // WEDGE_117
0, 1, 2, 3, // WEDGE_135
0, 1, 2, 3, // WEDGE_153
0, 1, 2, 3, // WEDGE_166
1, 2, 3, // WEDGE_180
1, 2, 3, // WEDGE_194
1, 2, 3, // WEDGE_207
1, 2, 3, // WEDGE_225
1, 2, 3, // WEDGE_243
1, 2, 3, // WEDGE_270
1, 2, 3, // WEDGE_297
1, 2, 3, // WEDGE_315
1, 2, 3, // WEDGE_333
1, 2, 3, // WEDGE_346
};
#endif // CONFIG_WEDGE_MOD_EXT
#if CONFIG_BAWP
#define BAWP_REF_LINES 1
#endif
// Angles are with respect to horizontal anti-clockwise
#if !CONFIG_WEDGE_MOD_EXT
enum {
WEDGE_HORIZONTAL = 0,
WEDGE_VERTICAL = 1,
WEDGE_OBLIQUE27 = 2,
WEDGE_OBLIQUE63 = 3,
WEDGE_OBLIQUE117 = 4,
WEDGE_OBLIQUE153 = 5,
WEDGE_DIRECTIONS
} UENUM1BYTE(WedgeDirectionType);
#endif
// 3-tuple: {direction, x_offset, y_offset}
typedef struct {
WedgeDirectionType direction;
int x_offset;
int y_offset;
} wedge_code_type;
typedef uint8_t *wedge_masks_type[MAX_WEDGE_TYPES];
typedef struct {
int wedge_types;
const wedge_code_type *codebook;
uint8_t *signflip;
wedge_masks_type *masks;
} wedge_params_type;
extern const wedge_params_type av1_wedge_params_lookup[BLOCK_SIZES_ALL];
typedef struct SubpelParams {
int xs;
int ys;
int subpel_x;
int subpel_y;
} SubpelParams;
struct build_prediction_ctxt {
const AV1_COMMON *cm;
uint16_t **tmp_buf;
int *tmp_width;
int *tmp_height;
int *tmp_stride;
int mb_to_far_edge;
void *dcb; // Decoder-only coding block.
};
typedef enum InterPredMode {
TRANSLATION_PRED,
WARP_PRED,
} InterPredMode;
typedef enum InterCompMode {
UNIFORM_SINGLE,
UNIFORM_COMP,
MASK_COMP,
} InterCompMode;
typedef struct InterPredParams {
InterPredMode mode;
InterCompMode comp_mode;
WarpedMotionParams warp_params;
ConvolveParams conv_params;
const InterpFilterParams *interp_filter_params[2];
int block_width;
int block_height;
#if CONFIG_OPTFLOW_REFINEMENT
// In optical flow refinement, block_width and block_height will pass the
// subblock size into av1_make_inter_predictor, while orig_block_width and
// orig_block_height keep the original block size that is needed by
// calc_subpel_params_func
int orig_block_width;
int orig_block_height;
#endif // CONFIG_OPTFLOW_REFINEMENT
int pix_row;
int pix_col;
struct buf_2d ref_frame_buf;
int subsampling_x;
int subsampling_y;
const struct scale_factors *scale_factors;
int bit_depth;
INTERINTER_COMPOUND_DATA mask_comp;
BLOCK_SIZE sb_type;
int is_intrabc;
#if CONFIG_TIP
/**
* \name Distance of this block from frame edges in 1/8th pixel units.
*/
/**@{*/
int dist_to_left_edge; /*!< Distance from left edge */
int dist_to_right_edge; /*!< Distance from right edge */
int dist_to_top_edge; /*!< Distance from top edge */
int dist_to_bottom_edge; /*!< Distance from bottom edge */
#endif // CONFIG_TIP
} InterPredParams;
#if CONFIG_OPTFLOW_REFINEMENT
// Apply bilinear and bicubic interpolation for subpel gradient to avoid
// calls of build_one_inter_predictor function. Bicubic interpolation
// brings better quality but the speed results are neutral. As such, bilinear
// interpolation is used by default for a better trade-off between quality
// and complexity.
#define OPFL_BILINEAR_GRAD 0
#define OPFL_BICUBIC_GRAD 1
// Use downsampled gradient arrays to compute MV offsets
#define OPFL_DOWNSAMP_QUINCUNX 1
// Delta to use for computing gradients in bits, with 0 referring to
// integer-pel. The actual delta value used from the 1/8-pel original MVs
// is 2^(3 - SUBPEL_GRAD_DELTA_BITS). The max value of this macro is 3.
#define SUBPEL_GRAD_DELTA_BITS 3
// Combine computations of interpolated gradients and the least squares
// solver. The basic idea is that, typically we would compute the following:
// 1. d0, d1, P0 and P1
// 2. Gradients of P0 and P1: gx0, gx1, gy0, and gy1
// 3. Solving least squares for vx and vy, which requires d0*gx0-d1*gx1,
// d0*gy0-d1*gy1, and P0-P1.
// When this flag is turned on, we compute the following
// 1. d0, d1, P0 and P1
// 2. tmp0 = d0*P0-d1*P1 and tmp1 = P0-P1
// 3. Gradients of tmp0: gx and gy
// 4. Solving least squares for vx and vy using gx, gy and tmp1
// Note that this only requires 2 gradient operators instead of 4 and thus
// reduces the complexity. However, it is only feasible when gradients are
// obtained using bilinear or bicubic interpolation. Thus, this flag should
// only be on when either of OPFL_BILINEAR_GRAD and OPFL_BICUBIC_GRAD is on.
#define OPFL_COMBINE_INTERP_GRAD_LS 1
// Bilinear and bicubic coefficients. Note that, at boundary, we apply
// coefficients that are doubled because spatial distance between the two
// interpolated pixels is halved. In other words, instead of computing
// coeff * (v[delta] - v[-delta]) / (2 * delta),
// we are practically computing
// coeff * (v[delta] - v[0]) / (2 * delta).
// Thus, coeff is doubled to get a better gradient quality.
#if OPFL_BILINEAR_GRAD
static const int bilinear_bits = 3;
static const int32_t coeffs_bilinear[4][2] = {
{ 8, 16 }, // delta = 1 (SUBPEL_GRAD_DELTA_BITS = 0)
{ 4, 8 }, // delta = 0.5 (SUBPEL_GRAD_DELTA_BITS = 1)
{ 2, 4 }, // delta = 0.25 (SUBPEL_GRAD_DELTA_BITS = 2)
{ 1, 2 }, // delta = 0.125 (SUBPEL_GRAD_DELTA_BITS = 3)
};
#endif
#if OPFL_BICUBIC_GRAD
static const int bicubic_bits = 7;
static const int32_t coeffs_bicubic[4][2][2] = {
{ { 128, 256 }, { 0, 0 } }, // delta = 1 (SUBPEL_GRAD_DELTA_BITS = 0)
{ { 80, 160 }, { -8, -16 } }, // delta = 0.5 (SUBPEL_GRAD_DELTA_BITS = 1)
{ { 42, 84 }, { -5, -10 } }, // delta = 0.25 (SUBPEL_GRAD_DELTA_BITS = 2)
{ { 21, 42 }, { -3, -6 } }, // delta = 0.125 (SUBPEL_GRAD_DELTA_BITS = 3)
};
#endif
#endif // CONFIG_OPTFLOW_REFINEMENT
void av1_init_inter_params(InterPredParams *inter_pred_params, int block_width,
int block_height, int pix_row, int pix_col,
int subsampling_x, int subsampling_y, int bit_depth,
int is_intrabc, const struct scale_factors *sf,
const struct buf_2d *ref_buf,
InterpFilter interp_filter);
#if CONFIG_WARP_REF_LIST
// Check if the signaling of the warp delta parameters are allowed
static INLINE int allow_warp_parameter_signaling(const MB_MODE_INFO *mbmi) {
return (
#if CONFIG_WARPMV
mbmi->mode != WARPMV &&
#endif // CONFIG_WARPMV
mbmi->motion_mode == WARP_DELTA && mbmi->warp_ref_idx == 1);
}
#endif // CONFIG_WARP_REF_LIST
#if CONFIG_CWP
// Map the index to weighting factor for compound weighted prediction
static INLINE int get_cwp_coding_idx(int val, int encode,
const AV1_COMMON *const cm,
const MB_MODE_INFO *const mbmi) {
int is_same_side = 0;
int cur_ref_side = 0;
int other_ref_side = 0;
if (has_second_ref(mbmi)) {
cur_ref_side = cm->ref_frame_side[mbmi->ref_frame[0]];
other_ref_side = cm->ref_frame_side[mbmi->ref_frame[1]];
is_same_side = (cur_ref_side > 0 && other_ref_side > 0) ||
(cur_ref_side == 0 && other_ref_side == 0);
}
if (encode) {
for (int i = 0; i < MAX_CWP_NUM; i++) {
if (cwp_weighting_factor[is_same_side][i] == val) return i;
}
return 0;
} else {
return cwp_weighting_factor[is_same_side][val];
}
}
#endif // CONFIG_CWP
#if CONFIG_ADAPTIVE_MVD
static INLINE int enable_adaptive_mvd_resolution(const AV1_COMMON *const cm,
const MB_MODE_INFO *mbmi) {
const int mode = mbmi->mode;
return (mode == NEAR_NEWMV || mode == NEW_NEARMV
#if CONFIG_OPTFLOW_REFINEMENT
|| mode == NEAR_NEWMV_OPTFLOW || mode == NEW_NEARMV_OPTFLOW
#if IMPROVED_AMVD && CONFIG_JOINT_MVD
|| mode == JOINT_AMVDNEWMV_OPTFLOW
#endif // IMPROVED_AMVD && CONFIG_JOINT_MVD
#endif
#if IMPROVED_AMVD
|| mode == AMVDNEWMV
#endif // IMPROVED_AMVD
#if IMPROVED_AMVD && CONFIG_JOINT_MVD
|| mode == JOINT_AMVDNEWMV
#endif
) &&
cm->seq_params.enable_adaptive_mvd;
}
#endif // CONFIG_ADAPTIVE_MVD
#if CONFIG_JOINT_MVD
// get the base reference frame list for joint MVD coding, the MVD for base
// reference frame is the same as the joint MVD, the MVD for the other reference
// frame is scaled from the joint MVD.
static INLINE int get_joint_mvd_base_ref_list(const AV1_COMMON *const cm,
const MB_MODE_INFO *mbmi) {
int base_ref_list = 0;
int first_ref_dist = 0;
int sec_ref_dist = 0;
if (has_second_ref(mbmi)) {
first_ref_dist = cm->ref_frame_relative_dist[mbmi->ref_frame[0]];
sec_ref_dist = cm->ref_frame_relative_dist[mbmi->ref_frame[1]];
if (first_ref_dist >= sec_ref_dist) {
base_ref_list = 0;
} else {
base_ref_list = 1;
}
}
return base_ref_list;
}
// check whether the direction of two reference frames are from same side
static INLINE int is_ref_frame_same_side(const AV1_COMMON *const cm,
const MB_MODE_INFO *mbmi) {
int is_same_side = 0;
int cur_ref_side = 0;
int other_ref_side = 0;
if (has_second_ref(mbmi)) {
cur_ref_side = cm->ref_frame_side[mbmi->ref_frame[0]];
other_ref_side = cm->ref_frame_side[mbmi->ref_frame[1]];
is_same_side = (cur_ref_side > 0 && other_ref_side > 0) ||
(cur_ref_side == 0 && other_ref_side == 0);
}
return is_same_side;
}
#endif // CONFIG_JOINT_MVD
void av1_init_comp_mode(InterPredParams *inter_pred_params);
void av1_init_warp_params(InterPredParams *inter_pred_params,
const WarpTypesAllowed *warp_types, int ref,
const MACROBLOCKD *xd, const MB_MODE_INFO *mi);
static INLINE int has_scale(int xs, int ys) {
return xs != SCALE_SUBPEL_SHIFTS || ys != SCALE_SUBPEL_SHIFTS;
}
static INLINE void revert_scale_extra_bits(SubpelParams *sp) {
sp->subpel_x >>= SCALE_EXTRA_BITS;
sp->subpel_y >>= SCALE_EXTRA_BITS;
sp->xs >>= SCALE_EXTRA_BITS;
sp->ys >>= SCALE_EXTRA_BITS;
assert(sp->subpel_x < SUBPEL_SHIFTS);
assert(sp->subpel_y < SUBPEL_SHIFTS);
assert(sp->xs <= SUBPEL_SHIFTS);
assert(sp->ys <= SUBPEL_SHIFTS);
}
static INLINE void highbd_inter_predictor(
const uint16_t *src, int src_stride, uint16_t *dst, int dst_stride,
const SubpelParams *subpel_params, int w, int h,
ConvolveParams *conv_params, const InterpFilterParams *interp_filters[2],
int bd) {
assert(conv_params->do_average == 0 || conv_params->do_average == 1);
const int is_scaled = has_scale(subpel_params->xs, subpel_params->ys);
if (is_scaled) {
av1_highbd_convolve_2d_facade(src, src_stride, dst, dst_stride, w, h,
interp_filters, subpel_params->subpel_x,
subpel_params->xs, subpel_params->subpel_y,
subpel_params->ys, 1, conv_params, bd);
} else {
SubpelParams sp = *subpel_params;
revert_scale_extra_bits(&sp);
av1_highbd_convolve_2d_facade(src, src_stride, dst, dst_stride, w, h,
interp_filters, sp.subpel_x, sp.xs,
sp.subpel_y, sp.ys, 0, conv_params, bd);
}
}
void av1_modify_neighbor_predictor_for_obmc(MB_MODE_INFO *mbmi);
int av1_skip_u4x4_pred_in_obmc(BLOCK_SIZE bsize,
const struct macroblockd_plane *pd, int dir);
static INLINE int is_interinter_compound_used(COMPOUND_TYPE type,
BLOCK_SIZE sb_type) {
const int comp_allowed = is_comp_ref_allowed(sb_type);
switch (type) {
case COMPOUND_AVERAGE:
case COMPOUND_DIFFWTD: return comp_allowed;
case COMPOUND_WEDGE:
return comp_allowed && av1_wedge_params_lookup[sb_type].wedge_types > 0;
default: assert(0); return 0;
}
}
static INLINE int is_any_masked_compound_used(BLOCK_SIZE sb_type) {
COMPOUND_TYPE comp_type;
int i;
if (!is_comp_ref_allowed(sb_type)) return 0;
for (i = 0; i < COMPOUND_TYPES; i++) {
comp_type = (COMPOUND_TYPE)i;
if (is_masked_compound_type(comp_type) &&
is_interinter_compound_used(comp_type, sb_type))
return 1;
}
return 0;
}
static INLINE int get_wedge_types_lookup(BLOCK_SIZE sb_type) {
return av1_wedge_params_lookup[sb_type].wedge_types;
}
static INLINE int av1_is_wedge_used(BLOCK_SIZE sb_type) {
return av1_wedge_params_lookup[sb_type].wedge_types > 0;
}
void av1_make_inter_predictor(const uint16_t *src, int src_stride,
uint16_t *dst, int dst_stride,
InterPredParams *inter_pred_params,
const SubpelParams *subpel_params);
typedef void (*CalcSubpelParamsFunc)(const MV *const src_mv,
InterPredParams *const inter_pred_params,
MACROBLOCKD *xd, int mi_x, int mi_y,
int ref,
#if CONFIG_OPTFLOW_REFINEMENT
int use_optflow_refinement,
#endif // CONFIG_OPTFLOW_REFINEMENT
uint16_t **mc_buf, uint16_t **pre,
SubpelParams *subpel_params,
int *src_stride);
void av1_build_one_inter_predictor(
uint16_t *dst, int dst_stride, const MV *const src_mv,
InterPredParams *inter_pred_params, MACROBLOCKD *xd, int mi_x, int mi_y,
int ref, uint16_t **mc_buf, CalcSubpelParamsFunc calc_subpel_params_func);
void av1_build_inter_predictors(const AV1_COMMON *cm, MACROBLOCKD *xd,
int plane, MB_MODE_INFO *mi,
#if CONFIG_BAWP
const BUFFER_SET *dst_orig,
#endif
int build_for_obmc, int bw, int bh, int mi_x,
int mi_y, uint16_t **mc_buf,
CalcSubpelParamsFunc calc_subpel_params_func);
#if CONFIG_OPTFLOW_REFINEMENT
// This parameter k=OPFL_DIST_RATIO_THR is used to prune MV refinement for the
// case where d0 and d1 are very different. Assuming a = max(|d0|, |d1|) and
// b = min(|d0|, |d1|), MV refinement will only be allowed only if a/b <= k.
// If k is set to 0, refinement will always be enabled.
// If k is set to 1, refinement will only be enabled when |d0|=|d1|.
#define OPFL_DIST_RATIO_THR 0
// Apply regularized least squares (RLS). The RLS parameter is bw * bh * 2^(b-4)
// where b = OPFL_RLS_PARAM_BITS.
#define OPFL_REGULARIZED_LS 1
#define OPFL_RLS_PARAM_BITS 4
// Number of bits allowed for covariance matrix elements (su2, sv2, suv, suw
// and svw) so that det, det_x, and det_y does not cause overflow issue in
// int64_t. Its value must be <= (64 - mv_prec_bits - grad_prec_bits) / 2.
#define OPFL_COV_CLAMP_BITS 28
#define OPFL_COV_CLAMP_VAL (1 << OPFL_COV_CLAMP_BITS)
// Precision of refined MV returned, 0 being integer pel. For now, only 1/8 or
// 1/16-pel can be used.
#define MV_REFINE_PREC_BITS 4 // (1/16-pel)
void av1_opfl_mv_refinement_highbd(const uint16_t *p0, int pstride0,
const uint16_t *p1, int pstride1,
const int16_t *gx0, const int16_t *gy0,
const int16_t *gx1, const int16_t *gy1,
int gstride, int bw, int bh, int d0, int d1,
int grad_prec_bits, int mv_prec_bits,
int *vx0, int *vy0, int *vx1, int *vy1);
void av1_opfl_build_inter_predictor(
const AV1_COMMON *cm, MACROBLOCKD *xd, int plane, const MB_MODE_INFO *mi,
int bw, int bh, int mi_x, int mi_y, uint16_t **mc_buf,
InterPredParams *inter_pred_params,
CalcSubpelParamsFunc calc_subpel_params_func, int ref, uint16_t *pred_dst);
static INLINE int is_opfl_refine_allowed(const AV1_COMMON *cm,
const MB_MODE_INFO *mbmi) {
if (cm->seq_params.enable_opfl_refine == AOM_OPFL_REFINE_NONE ||
cm->features.opfl_refine_type == REFINE_NONE)
return 0;
const unsigned int cur_index = cm->cur_frame->order_hint;
int d0, d1;
#if CONFIG_OPTFLOW_ON_TIP
if (mbmi->ref_frame[0] == TIP_FRAME) {
d0 = cm->tip_ref.ref_offset[0];
d1 = cm->tip_ref.ref_offset[1];
} else {
#endif // CONFIG_OPTFLOW_ON_TIP
if (!mbmi->ref_frame[1]) return 0;
const RefCntBuffer *const ref0 = get_ref_frame_buf(cm, mbmi->ref_frame[0]);
const RefCntBuffer *const ref1 = get_ref_frame_buf(cm, mbmi->ref_frame[1]);
d0 = (int)cur_index - (int)ref0->order_hint;
d1 = (int)cur_index - (int)ref1->order_hint;
#if CONFIG_OPTFLOW_ON_TIP
}
#endif // CONFIG_OPTFLOW_ON_TIP
if (!((d0 <= 0) ^ (d1 <= 0))) return 0;
return OPFL_DIST_RATIO_THR == 0 ||
(AOMMAX(abs(d0), abs(d1)) <=
OPFL_DIST_RATIO_THR * AOMMIN(abs(d0), abs(d1)));
}
// Generate refined MVs using optflow refinement
int av1_get_optflow_based_mv_highbd(
const AV1_COMMON *cm, MACROBLOCKD *xd, int plane, const MB_MODE_INFO *mbmi,
int_mv *mv_refined, int bw, int bh, int mi_x, int mi_y, uint16_t **mc_buf,
CalcSubpelParamsFunc calc_subpel_params_func, int16_t *gx0, int16_t *gy0,
int16_t *gx1, int16_t *gy1, int *vx0, int *vy0, int *vx1, int *vy1,
uint16_t *dst0, uint16_t *dst1
#if CONFIG_OPTFLOW_ON_TIP
,
int do_pred, int use_4x4
#endif // CONFIG_OPTFLOW_ON_TIP
);
// With the refined MVs, generate the inter prediction for the block.
void av1_opfl_rebuild_inter_predictor(
uint16_t *dst, int dst_stride, int plane, int_mv *const mv_refined,
InterPredParams *inter_pred_params, MACROBLOCKD *xd, int mi_x, int mi_y,
int ref, uint16_t **mc_buf, CalcSubpelParamsFunc calc_subpel_params_func
#if CONFIG_OPTFLOW_ON_TIP
,
int use_4x4
#endif // CONFIG_OPTFLOW_ON_TIP
);
// Integer division based on lookup table.
// num: numerator
// den: denominator
// out: output result (num / den)
static INLINE int32_t divide_and_round_signed(int64_t num, int64_t den) {
if (llabs(den) == 1) return (int32_t)(den < 0 ? -num : num);
const int optflow_prec_bits = 16;
int16_t shift;
const int sign_den = (den < 0 ? -1 : 1);
uint16_t inverse_den = resolve_divisor_64(llabs(den), &shift);
shift -= optflow_prec_bits;
if (shift < 0) {
inverse_den <<= (-shift);
shift = 0;
}
int32_t out;
// Make sure 1) the bits for right shift is < 63 and 2) the bit depth
// of num is < 48 to avoid overflow in num * inverse_den
if (optflow_prec_bits + shift >= 63 ||
ROUND_POWER_OF_TWO_SIGNED_64(num, 63 - optflow_prec_bits) != 0) {
int64_t out_tmp = ROUND_POWER_OF_TWO_SIGNED_64(num, optflow_prec_bits);
out = (int32_t)ROUND_POWER_OF_TWO_SIGNED_64(
out_tmp * (int64_t)inverse_den * sign_den, shift);
} else {
out = (int32_t)ROUND_POWER_OF_TWO_SIGNED_64(
num * (int64_t)inverse_den * sign_den, optflow_prec_bits + shift);
}
#ifndef NDEBUG
// Verify that the result is consistent with built-in division.
// Quick overflow check
int32_t out_div = (llabs(num) + llabs(den) < 0)
? (int32_t)DIVIDE_AND_ROUND_SIGNED(
ROUND_POWER_OF_TWO_SIGNED_64(num, 2),
ROUND_POWER_OF_TWO_SIGNED_64(den, 2))
: (int32_t)DIVIDE_AND_ROUND_SIGNED(num, den);
// check if error is at most 1 at usable values of out_div
if (abs(out_div - out) > 1 && abs(out_div) <= 64) {
printf("Warning: num = %" PRId64 ", den = %" PRId64
", inverse_den = %d, shift = %d, v0 = %d, v = %d\n",
num, den, inverse_den, shift, out_div, out);
}
#endif // NDEBUG
return out;
}
// Return 1 if current frame is REFINE_ALL and the current block uses optical
// flow refinement, i.e., inter mode is in {NEAR_NEARMV, NEAR_NEWMV,
// NEW_NEARMV, NEW_NEWMV}, and compound type is simple compound average.
static INLINE int use_opfl_refine_all(const AV1_COMMON *cm,
const MB_MODE_INFO *mbmi) {
return cm->features.opfl_refine_type == REFINE_ALL &&
mbmi->mode >= COMP_INTER_MODE_START &&
mbmi->mode < COMP_OPTFLOW_MODE_START &&
mbmi->mode != GLOBAL_GLOBALMV &&
mbmi->interinter_comp.type == COMPOUND_AVERAGE;
}
#endif // CONFIG_OPTFLOW_REFINEMENT
// TODO(jkoleszar): yet another mv clamping function :-(
static INLINE MV clamp_mv_to_umv_border_sb(const MACROBLOCKD *xd,
const MV *src_mv, int bw, int bh,
#if CONFIG_OPTFLOW_REFINEMENT
int use_optflow_refinement,
#endif // CONFIG_OPTFLOW_REFINEMENT
int ss_x, int ss_y) {
// If the MV points so far into the UMV border that no visible pixels
// are used for reconstruction, the subpel part of the MV can be
// discarded and the MV limited to 16 pixels with equivalent results.
const int spel_left = (AOM_INTERP_EXTEND + bw) << SUBPEL_BITS;
const int spel_right = spel_left - SUBPEL_SHIFTS;
const int spel_top = (AOM_INTERP_EXTEND + bh) << SUBPEL_BITS;
const int spel_bottom = spel_top - SUBPEL_SHIFTS;
#if CONFIG_OPTFLOW_REFINEMENT
MV clamped_mv;
if (use_optflow_refinement) {
// optflow refinement always returns MVs with 1/16 precision so it is not
// necessary to shift the MV before clamping
clamped_mv.row = (int16_t)ROUND_POWER_OF_TWO_SIGNED(
src_mv->row * (1 << SUBPEL_BITS), MV_REFINE_PREC_BITS + ss_y);
clamped_mv.col = (int16_t)ROUND_POWER_OF_TWO_SIGNED(
src_mv->col * (1 << SUBPEL_BITS), MV_REFINE_PREC_BITS + ss_x);
} else {
clamped_mv.row = (int16_t)(src_mv->row * (1 << (1 - ss_y)));
clamped_mv.col = (int16_t)(src_mv->col * (1 << (1 - ss_x)));
}
#else
MV clamped_mv = { (int16_t)(src_mv->row * (1 << (1 - ss_y))),
(int16_t)(src_mv->col * (1 << (1 - ss_x))) };
#endif // CONFIG_OPTFLOW_REFINEMENT
assert(ss_x <= 1);
assert(ss_y <= 1);
const SubpelMvLimits mv_limits = {
xd->mb_to_left_edge * (1 << (1 - ss_x)) - spel_left,
xd->mb_to_right_edge * (1 << (1 - ss_x)) + spel_right,
xd->mb_to_top_edge * (1 << (1 - ss_y)) - spel_top,
xd->mb_to_bottom_edge * (1 << (1 - ss_y)) + spel_bottom
};
clamp_mv(&clamped_mv, &mv_limits);
return clamped_mv;
}
static INLINE int64_t scaled_buffer_offset(int x_offset, int y_offset,
int stride,
const struct scale_factors *sf) {
const int x =
sf ? sf->scale_value_x(x_offset, sf) >> SCALE_EXTRA_BITS : x_offset;
const int y =
sf ? sf->scale_value_y(y_offset, sf) >> SCALE_EXTRA_BITS : y_offset;
return (int64_t)y * stride + x;
}
static INLINE void setup_pred_plane(struct buf_2d *dst, uint16_t *src,
int width, int height, int stride,
int mi_row, int mi_col,
const struct scale_factors *scale,
int subsampling_x, int subsampling_y,
const CHROMA_REF_INFO *chroma_ref_info) {
// Offset the buffer pointer
if (chroma_ref_info && (subsampling_x || subsampling_y)) {
mi_row = chroma_ref_info->mi_row_chroma_base;
mi_col = chroma_ref_info->mi_col_chroma_base;
}
const int x = (MI_SIZE * mi_col) >> subsampling_x;
const int y = (MI_SIZE * mi_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;
}
void av1_setup_dst_planes(struct macroblockd_plane *planes,
const YV12_BUFFER_CONFIG *src, int mi_row, int mi_col,
const int plane_start, const int plane_end,
const CHROMA_REF_INFO *chroma_ref_info);
void av1_setup_pre_planes(MACROBLOCKD *xd, int idx,
const YV12_BUFFER_CONFIG *src, int mi_row, int mi_col,
const struct scale_factors *sf, const int num_planes,
const CHROMA_REF_INFO *chroma_ref_info);
static INLINE void set_default_interp_filters(
MB_MODE_INFO *const mbmi,
#if CONFIG_OPTFLOW_REFINEMENT
const AV1_COMMON *cm,
#endif // CONFIG_OPTFLOW_REFINEMENT
InterpFilter frame_interp_filter) {
#if CONFIG_SKIP_MODE_ENHANCEMENT
if (mbmi->skip_mode) {
mbmi->interp_fltr = MULTITAP_SHARP;
return;
}
#endif // CONFIG_SKIP_MODE_ENHANCEMENT
#if CONFIG_OPTFLOW_REFINEMENT
mbmi->interp_fltr =
(mbmi->mode >= NEAR_NEARMV_OPTFLOW || use_opfl_refine_all(cm, mbmi))
? MULTITAP_SHARP
: av1_unswitchable_filter(frame_interp_filter);
#else
mbmi->interp_fltr = av1_unswitchable_filter(frame_interp_filter);
#endif // CONFIG_OPTFLOW_REFINEMENT
}
static INLINE int av1_is_interp_needed(const AV1_COMMON *const cm,
const MACROBLOCKD *const xd) {
(void)cm;
const MB_MODE_INFO *const mbmi = xd->mi[0];
if (mbmi->skip_mode) return 0;
#if CONFIG_WARPMV
if (mbmi->mode == WARPMV) return 0;
#endif // CONFIG_WARPMV
#if CONFIG_OPTFLOW_REFINEMENT
// No interpolation filter search when optical flow MV refinement is used.
if (mbmi->mode >= NEAR_NEARMV_OPTFLOW || use_opfl_refine_all(cm, mbmi))
return 0;
#endif // CONFIG_OPTFLOW_REFINEMENT
if (is_warp_mode(mbmi->motion_mode)) return 0;
if (is_nontrans_global_motion(xd, xd->mi[0])) return 0;
return 1;
}
// Sets up buffers 'dst_buf1' and 'dst_buf2' from relevant buffers in 'xd' for
// subsequent use in OBMC prediction.
void av1_setup_obmc_dst_bufs(MACROBLOCKD *xd, uint16_t **dst_buf1,
uint16_t **dst_buf2);
void av1_setup_build_prediction_by_above_pred(
MACROBLOCKD *xd, int rel_mi_col, uint8_t above_mi_width,
MB_MODE_INFO *above_mbmi, struct build_prediction_ctxt *ctxt,
const int num_planes);
void av1_setup_build_prediction_by_left_pred(MACROBLOCKD *xd, int rel_mi_row,
uint8_t left_mi_height,
MB_MODE_INFO *left_mbmi,
struct build_prediction_ctxt *ctxt,
const int num_planes);
void av1_build_obmc_inter_prediction(const AV1_COMMON *cm, MACROBLOCKD *xd,
uint16_t *above[MAX_MB_PLANE],
int above_stride[MAX_MB_PLANE],
uint16_t *left[MAX_MB_PLANE],
int left_stride[MAX_MB_PLANE]);
const uint8_t *av1_get_obmc_mask(int length);
void av1_count_overlappable_neighbors(const AV1_COMMON *cm, MACROBLOCKD *xd);
#define MASK_MASTER_SIZE ((MAX_WEDGE_SIZE) << 1)
#define MASK_MASTER_STRIDE (MASK_MASTER_SIZE)
void av1_init_wedge_masks();
static INLINE const uint8_t *av1_get_contiguous_soft_mask(int8_t wedge_index,
int8_t wedge_sign,
BLOCK_SIZE sb_type) {
return av1_wedge_params_lookup[sb_type].masks[wedge_sign][wedge_index];
}
const uint8_t *av1_get_compound_type_mask(
const INTERINTER_COMPOUND_DATA *const comp_data, BLOCK_SIZE sb_type);
#if CONFIG_CWP
// Init the masks for compound weighted prediction
void init_cwp_masks();
// Get the mask for compound weighted prediction
const int8_t *av1_get_cwp_mask(int list_idx, int idx);
#endif // CONFIG_CWP
// build interintra_predictors for one plane
void av1_build_interintra_predictor(const AV1_COMMON *cm, MACROBLOCKD *xd,
uint16_t *pred, int stride,
const BUFFER_SET *ctx, int plane,
BLOCK_SIZE bsize);
#if CONFIG_EXT_RECUR_PARTITIONS
void av1_build_intra_predictors_for_interintra(const AV1_COMMON *cm,
MACROBLOCKD *xd, int plane,
const BUFFER_SET *ctx,
uint16_t *dst, int dst_stride);
#else
void av1_build_intra_predictors_for_interintra(const AV1_COMMON *cm,
MACROBLOCKD *xd,
BLOCK_SIZE bsize, int plane,
const BUFFER_SET *ctx,
uint16_t *dst, int dst_stride);
#endif // CONFIG_EXT_RECUR_PARTITIONS
void av1_combine_interintra(MACROBLOCKD *xd, BLOCK_SIZE bsize, int plane,
const uint16_t *inter_pred, int inter_stride,
const uint16_t *intra_pred, int intra_stride);
int av1_allow_warp(const MB_MODE_INFO *const mbmi,
const WarpTypesAllowed *const warp_types,
const WarpedMotionParams *const gm_params,
#if CONFIG_EXTENDED_WARP_PREDICTION
int ref,
#endif // CONFIG_EXTENDED_WARP_PREDICTION
int build_for_obmc, const struct scale_factors *const sf,
WarpedMotionParams *final_warp_params);
#if CONFIG_FLEX_MVRES
// derive the context of the mpp_flag
int av1_get_mpp_flag_context(const AV1_COMMON *cm, const MACROBLOCKD *xd);
// derive the context of the precision signaling
int av1_get_pb_mv_precision_down_context(const AV1_COMMON *cm,
const MACROBLOCKD *xd);
// derive the context of the mv class
int av1_get_mv_class_context(const MvSubpelPrecision pb_mv_precision);
// set the precision of a block to the precision
void set_mv_precision(MB_MODE_INFO *mbmi, MvSubpelPrecision precision);
#if BUGFIX_AMVD_AMVR
void set_amvd_mv_precision(MB_MODE_INFO *mbmi, MvSubpelPrecision precision);
#endif // BUGFIX_AMVD_AMVR
// set the most probable mv precision of the block
// Currently, the most probable MV precision is same as the maximum precision of
// the block.
void set_most_probable_mv_precision(const AV1_COMMON *const cm,
MB_MODE_INFO *mbmi, const BLOCK_SIZE bsize);
// Set the default value fo the precision set. Currently the value is always 0.
void set_default_precision_set(const AV1_COMMON *const cm, MB_MODE_INFO *mbmi,
const BLOCK_SIZE bsize);
// Set the precision set of the block. Currently, the value is 0.
void set_precision_set(const AV1_COMMON *const cm, MACROBLOCKD *const xd,
MB_MODE_INFO *mbmi, const BLOCK_SIZE bsize,
uint8_t ref_mv_idx);
// Get the index of the precision
// this index is signalled when precision is not same as the most probable
// precision
int av1_get_pb_mv_precision_index(const MB_MODE_INFO *mbmi);
// get the actual precision value from the signalled index
MvSubpelPrecision av1_get_precision_from_index(MB_MODE_INFO *mbmi,
int precision_idx_coded_value);
// Set the maximum precision to the default value
void set_default_max_mv_precision(MB_MODE_INFO *mbmi,
MvSubpelPrecision precision);
// get the maximum allowed precision value of the block
MvSubpelPrecision av1_get_mbmi_max_mv_precision(const AV1_COMMON *const cm,
const SB_INFO *sbi,
const MB_MODE_INFO *mbmi);
// check if pb_mv_precision is allowed or not
int is_pb_mv_precision_active(const AV1_COMMON *const cm,
const MB_MODE_INFO *mbmi, const BLOCK_SIZE bsize);
#endif
#if CONFIG_WARPMV
// check if the WARPMV mode is allwed for a given blocksize
static INLINE int is_warpmv_allowed_bsize(BLOCK_SIZE bsize) {
assert(bsize < BLOCK_SIZES_ALL);
return AOMMIN(block_size_wide[bsize], block_size_high[bsize]) >= 8;
}
// check if WARPMV mode is allowed
static INLINE int is_warpmv_mode_allowed(const AV1_COMMON *const cm,
const MB_MODE_INFO *mbmi,
BLOCK_SIZE bsize) {
if (has_second_ref(mbmi) || !cm->features.enabled_motion_modes
#if CONFIG_TIP
|| is_tip_ref_frame(mbmi->ref_frame[0])
#endif // CONFIG_TIP
)
return 0;
int frame_warp_delta_allowed =
cm->features.enabled_motion_modes & (1 << WARP_DELTA);
return frame_warp_delta_allowed && is_warpmv_allowed_bsize(bsize);
}
#endif // CONFIG_WARPMV
#ifdef __cplusplus
} // extern "C"
#endif
#endif // AOM_AV1_COMMON_RECONINTER_H_