hashirama/uvg266
1
0
Fork 0
mirror of https://github.com/ultravideo/uvg266.git synced 2024-11-23 18:14:06 +00:00
Code Issues Actions Projects Releases Wiki Activity

[lfnst] change kvz_ to uvg_

Browse source
This commit is contained in:
Joose Sainio 2022-06-13 10:00:46 +03:00
parent 7abbf14d50
commit 480a907c07
7 changed files with 34 additions and 34 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
src/encode_coding_tree.c
View file

@ -130,11 +130,11 @@ static int get_isp_split_dim(const int width, const int height, const int isp_sp
non_split_dim_size = height;
}
const int min_num_samples_cu = 1 << ((kvz_math_floor_log2(MIN_TB_SIZE_Y) << 1));
const int factor_to_min_samples = non_split_dim_size < min_num_samples_cu ? min_num_samples_cu >> kvz_math_floor_log2(non_split_dim_size) : 1;
const int min_num_samples_cu = 1 << ((uvg_math_floor_log2(MIN_TB_SIZE_Y) << 1));
const int factor_to_min_samples = non_split_dim_size < min_num_samples_cu ? min_num_samples_cu >> uvg_math_floor_log2(non_split_dim_size) : 1;
partition_size = (split_dim_size >> div_shift) < factor_to_min_samples ? factor_to_min_samples : (split_dim_size >> div_shift);
assert(!(kvz_math_floor_log2(partition_size) + kvz_math_floor_log2(non_split_dim_size) < kvz_math_floor_log2(min_num_samples_cu)) && "Partition has less than minimum amount of samples.");
assert(!(uvg_math_floor_log2(partition_size) + uvg_math_floor_log2(non_split_dim_size) < uvg_math_floor_log2(min_num_samples_cu)) && "Partition has less than minimum amount of samples.");
return partition_size;
}
@ -217,7 +217,7 @@ static bool encode_lfnst_idx(encoder_state_t * const state, cabac_data_t * const
// TODO: this works only for square blocks
const int pu_x = x + ((i % tu_row_length) * tu_width);
const int pu_y = y + ((i / tu_row_length) * tu_height);
const cu_info_t* cur_tu = kvz_cu_array_at_const(frame->cu_array, pu_x, pu_y);
const cu_info_t* cur_tu = uvg_cu_array_at_const(frame->cu_array, pu_x, pu_y);
assert(cur_tu != NULL && "NULL transform unit.");
bool cbf_set = cbf_is_set(cur_tu->cbf, tr_depth, COLOR_Y);
@ -1207,7 +1207,7 @@ void uvg_encode_intra_luma_coding_unit(const encoder_state_t * const state,
}
}
kvz_cabac_encode_trunc_bin(cabac, tmp_pred, 67 - INTRA_MPM_COUNT, bits_out);
uvg_cabac_encode_trunc_bin(cabac, tmp_pred, 67 - INTRA_MPM_COUNT, bits_out);
}
if (cabac->only_count && bits_out) *bits_out += bits;
}
@ -1658,7 +1658,7 @@ void uvg_encode_coding_tree(encoder_state_t * const state,
if (state->encoder_control->chroma_format != UVG_CSP_400 && depth == 4 && x % 8 && y % 8) {
encode_chroma_intra_cu(cabac, cur_cu, state->encoder_control->cfg.cclm);
// LFNST constraints must be reset here. Otherwise the left over values will interfere when calculating new constraints
cu_info_t* tmp = kvz_cu_array_at(frame->cu_array, x, y);
cu_info_t* tmp = uvg_cu_array_at(frame->cu_array, x, y);
tmp->violates_lfnst_constrained[0] = false;
tmp->violates_lfnst_constrained[1] = false;
tmp->lfnst_last_scan_pos = false;

8
src/lfnst_tables.h
View file

@ -48,12 +48,12 @@
#define NUM_EXT_LUMA_MODE 28
#define DIA_IDX 34
const uint8_t kvz_lfnst_lut[NUM_INTRA_MODE + NUM_EXT_LUMA_MODE - 1] =
const uint8_t uvg_lfnst_lut[NUM_INTRA_MODE + NUM_EXT_LUMA_MODE - 1] =
{//0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94
0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1
};
const int8_t kvz_lfnst_8x8[4][2][16][48] = {
const int8_t uvg_lfnst_8x8[4][2][16][48] = {
{ //0
{
{ -117, 28, 18, 2, 4, 1, 2, 1, 32, -18, -2, 0, -1, 0, 0, 0, 14, -1, -3, 0, -1, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 3, 0, -1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0 },
@ -208,7 +208,7 @@ const int8_t kvz_lfnst_8x8[4][2][16][48] = {
}
};
const int8_t kvz_lfnst_4x4[4][2][16][16] = {
const int8_t uvg_lfnst_4x4[4][2][16][16] = {
{ //0
{
{ 108, -44, -15, 1, -44, 19, 7, -1, -11, 6, 2, -1, 0, -1, -1, 0 },
@ -363,7 +363,7 @@ const int8_t kvz_lfnst_4x4[4][2][16][16] = {
}
};
const uint32_t kvz_coef_top_left_diag_scan_8x8[8][64] = {
const uint32_t uvg_coef_top_left_diag_scan_8x8[8][64] = {
{0, 1, 1, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 5, 5, 6, 4, 5, 5, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8, 9, 9, 10, 4, 5, 5, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8, 9, 9, 10, 8, 9, 9, 10, 10, 10, 11, 11, 11, 11, 12, 12, 12, 13, 13, 14},
{0, 2, 1, 4, 3, 2, 6, 5, 4, 3, 7, 6, 5, 8, 7, 9, 8, 10, 9, 12, 11, 10, 14, 13, 12, 11, 15, 14, 13, 16, 15, 17, 4, 6, 5, 8, 7, 6, 10, 9, 8, 7, 11, 10, 9, 12, 11, 13, 12, 14, 13, 16, 15, 14, 18, 17, 16, 15, 19, 18, 17, 20, 19, 21},
{0, 4, 1, 8, 5, 2, 12, 9, 6, 3, 13, 10, 7, 14, 11, 15, 16, 20, 17, 24, 21, 18, 28, 25, 22, 19, 29, 26, 23, 30, 27, 31, 4, 8, 5, 12, 9, 6, 16, 13, 10, 7, 17, 14, 11, 18, 15, 19, 20, 24, 21, 28, 25, 22, 32, 29, 26, 23, 33, 30, 27, 34, 31, 35},

8
src/search_intra.c
View file

@ -259,11 +259,11 @@ static void derive_lfnst_constraints(cu_info_t* const pred_cu,
const int lcu_x = (lcu_px.x >> shift) & ~3;
const int lcu_y = (lcu_px.y >> shift) & ~3;
int8_t scan_idx = kvz_get_scan_order(pred_cu->type, pred_cu->intra.mode, depth);
int8_t scan_idx = uvg_get_scan_order(pred_cu->type, pred_cu->intra.mode, depth);
// ToDo: large block support in VVC?
const uint32_t log2_block_size = kvz_g_convert_to_bit[width] + 2;
const uint32_t* scan = kvz_g_sig_last_scan[scan_idx][log2_block_size - 1];
const uint32_t log2_block_size = uvg_g_convert_to_bit[width] + 2;
const uint32_t* scan = uvg_g_sig_last_scan[scan_idx][log2_block_size - 1];
const coeff_t* coeff;
switch (color) {
case COLOR_Y:
@ -462,7 +462,7 @@ static double search_intra_trdepth(
depth, search_data,
pred_cu,
lcu);
best_rd_cost += kvz_cu_rd_cost_chroma(state, lcu_px.x, lcu_px.y, depth, pred_cu, lcu);
best_rd_cost += uvg_cu_rd_cost_chroma(state, lcu_px.x, lcu_px.y, depth, pred_cu, lcu);
pred_cu->intra.mode = luma_mode;
// Check lfnst constraints for chroma

4
src/strategies/avx2/quant-avx2.c
View file

@ -707,7 +707,7 @@ int uvg_quantize_residual_avx2(encoder_state_t *const state,
if (state->encoder_control->cfg.lfnst && cur_cu->type == CU_INTRA) {
// Forward low frequency non-separable transform
kvz_fwd_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
uvg_fwd_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
}
// Quantize coeffs. (coeff -> coeff_out)
@ -745,7 +745,7 @@ int uvg_quantize_residual_avx2(encoder_state_t *const state,
if (state->encoder_control->cfg.lfnst && cur_cu->type == CU_INTRA) {
// Inverse low frequency non-separable transform
kvz_inv_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
uvg_inv_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
}
if (use_trskip) {
uvg_itransformskip(state->encoder_control, residual, coeff, width);

4
src/strategies/generic/quant-generic.c
View file

@ -463,7 +463,7 @@ int uvg_quantize_residual_generic(encoder_state_t *const state,
if (state->encoder_control->cfg.lfnst && cur_cu->type == CU_INTRA) {
// Forward low frequency non-separable transform
kvz_fwd_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
uvg_fwd_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
}
@ -507,7 +507,7 @@ int uvg_quantize_residual_generic(encoder_state_t *const state,
if (state->encoder_control->cfg.lfnst && cur_cu->type == CU_INTRA) {
// Inverse low frequency non-separable transform
kvz_inv_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
uvg_inv_lfnst(cur_cu, width, height, color, lfnst_index, coeff);
}
if (use_trskip) {
uvg_itransformskip(state->encoder_control, residual, coeff, width);

28
src/transform.c
View file

@ -232,9 +232,9 @@ void uvg_itransform2d(const encoder_control_t * const encoder,
}
}
void kvz_fwd_lfnst_NxN(coeff_t *src, coeff_t *dst, const int8_t mode, const int8_t index, const int8_t size, int zero_out_size)
void uvg_fwd_lfnst_NxN(coeff_t *src, coeff_t *dst, const int8_t mode, const int8_t index, const int8_t size, int zero_out_size)
{
const int8_t *tr_mat = (size > 4) ? kvz_lfnst_8x8[mode][index][0] : kvz_lfnst_4x4[mode][index][0];
const int8_t *tr_mat = (size > 4) ? uvg_lfnst_8x8[mode][index][0] : uvg_lfnst_4x4[mode][index][0];
const int tr_size = (size > 4) ? 48 : 16;
int coef;
coeff_t *out = dst;
@ -277,7 +277,7 @@ static inline bool get_transpose_flag(const int8_t intra_mode)
((intra_mode < NUM_LUMA_MODE) && (intra_mode > DIA_IDX));
}
void kvz_fwd_lfnst(const cu_info_t* const cur_cu,
void uvg_fwd_lfnst(const cu_info_t* const cur_cu,
const int width, const int height,
const uint8_t color,
const uint16_t lfnst_idx,
@ -295,14 +295,14 @@ void kvz_fwd_lfnst(const cu_info_t* const cur_cu,
const int cu_type = cur_cu->type;
const int scan_order = kvz_get_scan_order(cu_type, intra_mode, depth);
const int scan_order = uvg_get_scan_order(cu_type, intra_mode, depth);
if (lfnst_index && !mts_skip && (is_separate_tree || color == COLOR_Y))
{
const uint32_t log2_block_size = kvz_g_convert_to_bit[width] + 2;
const uint32_t log2_block_size = uvg_g_convert_to_bit[width] + 2;
assert(log2_block_size != -1 && "LFNST: invalid block width.");
const bool whge3 = width >= 8 && height >= 8;
const uint32_t* scan = whge3 ? kvz_coef_top_left_diag_scan_8x8[log2_block_size] : kvz_g_sig_last_scan[scan_order][log2_block_size - 1];
const uint32_t* scan = whge3 ? uvg_coef_top_left_diag_scan_8x8[log2_block_size] : uvg_g_sig_last_scan[scan_order][log2_block_size - 1];
if (is_cclm_mode) {
intra_mode = cur_cu->intra.mode;
@ -367,7 +367,7 @@ void kvz_fwd_lfnst(const cu_info_t* const cur_cu,
}
}
kvz_fwd_lfnst_NxN(tmp_in_matrix, tmp_out_matrix, kvz_lfnst_lut[intra_mode], lfnst_index - 1, sb_size,
uvg_fwd_lfnst_NxN(tmp_in_matrix, tmp_out_matrix, uvg_lfnst_lut[intra_mode], lfnst_index - 1, sb_size,
(tu_4x4 || tu_8x8) ? 8 : 16);
lfnst_tmp = tmp_out_matrix; // forward spectral rearrangement
@ -383,11 +383,11 @@ void kvz_fwd_lfnst(const cu_info_t* const cur_cu,
}
}
void kvz_inv_lfnst_NxN(coeff_t *src, coeff_t *dst, const uint32_t mode, const uint32_t index, const uint32_t size, int zero_out_size, const int max_log2_tr_dyn_range)
void uvg_inv_lfnst_NxN(coeff_t *src, coeff_t *dst, const uint32_t mode, const uint32_t index, const uint32_t size, int zero_out_size, const int max_log2_tr_dyn_range)
{
const coeff_t output_min = -(1 << max_log2_tr_dyn_range);
const coeff_t output_max = (1 << max_log2_tr_dyn_range) - 1;
const int8_t *tr_mat = (size > 4) ? kvz_lfnst_8x8[mode][index][0] : kvz_lfnst_4x4[mode][index][0];
const int8_t *tr_mat = (size > 4) ? uvg_lfnst_8x8[mode][index][0] : uvg_lfnst_4x4[mode][index][0];
const int tr_size = (size > 4) ? 48 : 16;
int resi;
coeff_t *out = dst;
@ -408,7 +408,7 @@ void kvz_inv_lfnst_NxN(coeff_t *src, coeff_t *dst, const uint32_t mode, const ui
}
}
void kvz_inv_lfnst(const cu_info_t *cur_cu,
void uvg_inv_lfnst(const cu_info_t *cur_cu,
const int width, const int height,
const uint8_t color,
const uint16_t lfnst_idx,
@ -429,12 +429,12 @@ void kvz_inv_lfnst(const cu_info_t *cur_cu,
const int cu_type = cur_cu->type;
const int scan_order = kvz_get_scan_order(cu_type, intra_mode, depth);
const int scan_order = uvg_get_scan_order(cu_type, intra_mode, depth);
if (lfnst_index && !mts_skip && (is_separate_tree || color == COLOR_Y)) {
const uint32_t log2_block_size = kvz_g_convert_to_bit[width] + 2;
const uint32_t log2_block_size = uvg_g_convert_to_bit[width] + 2;
const bool whge3 = width >= 8 && height >= 8;
const uint32_t* scan = whge3 ? kvz_coef_top_left_diag_scan_8x8[log2_block_size] : kvz_g_sig_last_scan[scan_order][log2_block_size - 1];
const uint32_t* scan = whge3 ? uvg_coef_top_left_diag_scan_8x8[log2_block_size] : uvg_g_sig_last_scan[scan_order][log2_block_size - 1];
if (is_cclm_mode) {
intra_mode = cur_cu->intra.mode;
@ -469,7 +469,7 @@ void kvz_inv_lfnst(const cu_info_t *cur_cu,
scan_ptr++;
}
kvz_inv_lfnst_NxN(tmp_in_matrix, tmp_out_matrix, kvz_lfnst_lut[intra_mode], lfnst_index - 1, sb_size,
uvg_inv_lfnst_NxN(tmp_in_matrix, tmp_out_matrix, uvg_lfnst_lut[intra_mode], lfnst_index - 1, sb_size,
(tu_4x4_flag || tu_8x8_flag) ? 8 : 16, max_log2_dyn_range);
lfnst_tmp = tmp_out_matrix; // inverse low frequency non-separale transform

4
src/transform.h
View file

@ -79,13 +79,13 @@ void uvg_quantize_lcu_residual(
lcu_t* lcu,
bool early_skip);
void kvz_fwd_lfnst(const cu_info_t* const cur_cu,
void uvg_fwd_lfnst(const cu_info_t* const cur_cu,
const int width, const int height,
const uint8_t color,
const uint16_t lfnst_idx,
coeff_t *coeffs);
void kvz_inv_lfnst(const cu_info_t* cur_cu,
void uvg_inv_lfnst(const cu_info_t* cur_cu,
const int width, const int height,
const uint8_t color,
const uint16_t lfnst_idx,