Add palette coding mode for UV

For 444 videos, a single palette of 3-d colors is
generated for YUV. For 420 videos, there may be two
palettes, one for Y, and the other for UV.

Also fixed a bug when palette and tx-skip are both on.

on derflr
--enable-palette                    +0.00%
with all experiments                +5.87% (was +5.93%)

on screen_content
--enable-palette                    +6.00%
--enable-palette --enable-tx_skip   +15.3%

on screen_content 444 version
--enable-palette                    +6.76%
--enable-palette --enable-tx_skip   +19.5%

Change-Id: I7287090aecc90eebcd4335d132a8c2c3895dfdd4
This commit is contained in:
hui su 2015-02-21 18:41:01 -08:00
parent ee2f0bdfcd
commit 7621c779e5
12 changed files with 899 additions and 211 deletions

View File

@ -233,19 +233,19 @@ typedef struct {
int interinter_wedge_index;
#endif // CONFIG_WEDGE_PARTITION
#if CONFIG_PALETTE
int palette_enabled;
int palette_size;
int palette_enabled[2];
int palette_size[2];
int palette_indexed_size;
int palette_literal_size;
int palette_run_length;
int palette_run_length[2];
int current_palette_size;
int palette_delta_bitdepth;
uint8_t palette_colors[PALETTE_MAX_SIZE];
uint8_t palette_colors[3 * PALETTE_MAX_SIZE];
uint8_t palette_indexed_colors[PALETTE_MAX_SIZE];
int8_t palette_color_delta[PALETTE_MAX_SIZE];
uint8_t palette_literal_colors[PALETTE_MAX_SIZE];
uint16_t palette_runs[PALETTE_MAX_RUNS];
PALETTE_SCAN_ORDER palette_scan_order;
uint16_t palette_runs[2 * PALETTE_MAX_RUNS];
PALETTE_SCAN_ORDER palette_scan_order[2];
#endif // CONFIG_PALETTE
} MB_MODE_INFO;
@ -360,7 +360,7 @@ typedef struct macroblockd {
DECLARE_ALIGNED(16, tran_low_t, dqcoeff[MAX_MB_PLANE][64 * 64]);
#if CONFIG_PALETTE
DECLARE_ALIGNED(16, uint8_t, color_index_map[64 * 64]);
DECLARE_ALIGNED(16, uint8_t, color_index_map[2][64 * 64]);
#endif
ENTROPY_CONTEXT *above_context[MAX_MB_PLANE];

View File

@ -409,6 +409,33 @@ default_palette_run_length_prob[10][PALETTE_RUN_LENGTHS - 1] = {
{ 30, 12, 14, 18, 75, 75},
{ 72, 34, 32, 42, 42, 108},
};
static const vp9_prob default_palette_uv_size_prob[10][PALETTE_SIZES - 1] = {
{ 180, 192, 207, 221, 55, 254},
{ 128, 185, 235, 148, 180, 128},
{ 128, 185, 235, 148, 180, 128},
{ 150, 180, 211, 77, 147, 179},
{ 128, 60, 195, 68, 120, 228},
{ 128, 60, 195, 68, 120, 228},
{ 228, 15, 17, 178, 47, 197},
{ 103, 210, 145, 250, 190, 128},
{ 103, 210, 145, 250, 190, 128},
{ 5, 76, 230, 148, 239, 254},
};
static const vp9_prob
default_palette_uv_run_length_prob[10][PALETTE_RUN_LENGTHS - 1] = {
{ 81, 107, 112, 254, 128, 128},
{ 120, 108, 130, 105, 250, 128},
{ 120, 108, 130, 105, 250, 128},
{ 107, 107, 143, 80, 60, 254},
{ 110, 65, 80, 120, 50, 42},
{ 110, 65, 80, 120, 50, 42},
{ 95, 5, 15, 127, 15, 56},
{ 5, 5, 5, 70, 75, 18},
{ 5, 5, 5, 70, 75, 18},
{ 5, 5, 5, 5, 157, 5},
};
#endif // CONFIG_PALETTE
#if CONFIG_SUPERTX
@ -544,6 +571,8 @@ void vp9_init_mode_probs(FRAME_CONTEXT *fc) {
#if CONFIG_PALETTE
vp9_copy(fc->palette_size_prob, default_palette_size_prob);
vp9_copy(fc->palette_run_length_prob, default_palette_run_length_prob);
vp9_copy(fc->palette_uv_size_prob, default_palette_uv_size_prob);
vp9_copy(fc->palette_uv_run_length_prob, default_palette_uv_run_length_prob);
#endif // CONFIG_PALETTE
#if CONFIG_SUPERTX
vp9_copy(fc->supertx_prob, default_supertx_prob);

View File

@ -69,6 +69,8 @@ typedef struct frame_contexts {
#if CONFIG_PALETTE
vp9_prob palette_size_prob[10][PALETTE_SIZES - 1];
vp9_prob palette_run_length_prob[10][PALETTE_RUN_LENGTHS - 1];
vp9_prob palette_uv_size_prob[10][PALETTE_SIZES - 1];
vp9_prob palette_uv_run_length_prob[10][PALETTE_RUN_LENGTHS - 1];
#endif // CONFIG_PALETTE
#if CONFIG_SUPERTX
vp9_prob supertx_prob[PARTITION_SUPERTX_CONTEXTS][TX_SIZES];

View File

@ -260,7 +260,9 @@ static INLINE void init_macroblockd(VP9_COMMON *cm, MACROBLOCKD *xd) {
}
#if CONFIG_PALETTE
xd->plane[0].color_index_map = xd->color_index_map;
for (i = 0; i < 2; ++i) {
xd->plane[i].color_index_map = xd->color_index_map[i];
}
#endif
xd->above_seg_context = cm->above_seg_context;

View File

@ -107,7 +107,7 @@ void transpose_block(uint8_t *seq_in, uint8_t *seq_out, int rows, int cols) {
void palette_color_insertion(uint8_t *old_colors, int *m, int *count,
MB_MODE_INFO *mbmi) {
int k = *m, n = mbmi->palette_literal_size;
int i, j, l, idx, min_idx = -1;
int i, j, l, min_idx = -1;
uint8_t *new_colors = mbmi->palette_literal_colors;
uint8_t val;
@ -129,48 +129,27 @@ void palette_color_insertion(uint8_t *old_colors, int *m, int *count,
for (i = 0; i < n; i++) {
val = new_colors[i];
j = 0;
while (val > old_colors[j] && j < k)
while (val != old_colors[j] && j < k)
j++;
if (j < k && val == old_colors[j]) {
count[j] += 8;
continue;
}
idx = j;
k++;
if (k > PALETTE_BUF_SIZE) {
k--;
if (k + 1 > PALETTE_BUF_SIZE) {
min_idx = 0;
for (l = 1; l < k; l++) {
for (l = 1; l < k; l++)
if (count[l] < count[min_idx])
min_idx = l;
}
l = min_idx;
while (l < k - 1) {
old_colors[l] = old_colors[l + 1];
count[l] = count[l + 1];
l++;
}
}
if (min_idx < 0 || idx <= min_idx)
j = idx;
else
j = idx - 1;
if (j == k - 1) {
old_colors[k - 1] = val;
count[k - 1] = 8;
old_colors[min_idx] = val;
count[min_idx] = 8;
} else {
for (l = k - 1; l > j; l--) {
old_colors[l] = old_colors[l - 1];
count[l] = count[l - 1];
}
old_colors[j] = val;
count[j] = 8;
old_colors[k] = val;
count[k] = 8;
k++;
}
}
*m = k;
}
@ -209,7 +188,7 @@ int get_bit_depth(int n) {
int palette_max_run(BLOCK_SIZE bsize) {
int table[BLOCK_SIZES] = {
16, 16, 16, 32, // BLOCK_4X4, BLOCK_4X8, BLOCK_8X4, BLOCK_8X8
32, 32, 32, 32, // BLOCK_4X4, BLOCK_4X8, BLOCK_8X4, BLOCK_8X8
64, 64, 64, 64, // BLOCK_8X16, BLOCK_16X8, BLOCK_16X16, BLOCK_16X32
64, 64, 64, 64, // BLOCK_32X16, BLOCK_32X32, BLOCK_32X64, BLOCK_64X32
64 // BLOCK_64X64

View File

@ -1382,11 +1382,18 @@ void vp9_predict_intra_block(const MACROBLOCKD *xd, int block_idx, int bwl_in,
if (!filterflag) {
#endif // CONFIG_FILTERINTRA
#if CONFIG_PALETTE
if (xd->mi[0].src_mi->mbmi.palette_enabled && !plane) {
uint8_t *palette = xd->mi[0].src_mi->mbmi.palette_colors;
if (xd->mi[0].src_mi->mbmi.palette_enabled[plane !=0 ]) {
uint8_t *palette = xd->mi[0].src_mi->mbmi.palette_colors +
plane * PALETTE_MAX_SIZE;
int bs = 4 * (1 << tx_size);
uint8_t *map = xd->plane[0].color_index_map;
int r, c, stride = 4 * (1 << bwl_in);
int stride = 4 * (1 << bwl_in);
int r, c;
uint8_t *map = NULL;
if (xd->plane[1].subsampling_x && xd->plane[1].subsampling_y)
map = xd->plane[plane != 0].color_index_map;
else if (!xd->plane[1].subsampling_x && !xd->plane[1].subsampling_y)
map = xd->plane[0].color_index_map;
for (r = 0; r < bs; r++) {
for (c = 0; c < bs; c++) {

View File

@ -223,28 +223,34 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
mbmi->segment_id = read_intra_segment_id(cm, xd, mi_row, mi_col, r);
mbmi->skip = read_skip(cm, xd, mbmi->segment_id, r);
#if CONFIG_PALETTE
if (bsize >= BLOCK_8X8 && cm->allow_palette_mode)
mbmi->palette_enabled = vp9_read_bit(r);
else
mbmi->palette_enabled = 0;
if (mbmi->palette_enabled) {
int i, j, k;
int m1, m2, d;
if (bsize >= BLOCK_8X8 && cm->allow_palette_mode) {
mbmi->palette_enabled[0] = vp9_read_bit(r);
mbmi->palette_enabled[1] = vp9_read_bit(r);
} else {
mbmi->palette_enabled[0] = 0;
mbmi->palette_enabled[1] = 0;
}
if (mbmi->palette_enabled[0]) {
int i, m1, m2, d, val;
int rows = 4 * num_4x4_blocks_high_lookup[bsize];
int cols = 4 * num_4x4_blocks_wide_lookup[bsize];
int c1, c2;
int scan_order[4096];
uint8_t map[4096];
PALETTE_RUN_LENGTH bits;
mbmi->mode = DC_PRED;
mbmi->palette_size =
mbmi->palette_size[0] =
vp9_read_tree(r, vp9_palette_size_tree,
cm->fc.palette_size_prob[bsize - BLOCK_8X8]);
mbmi->palette_size += 2;
mbmi->palette_indexed_size =
vp9_decode_uniform(r, MIN(mbmi->palette_size + 1, 8));
mbmi->palette_literal_size = mbmi->palette_size -
mbmi->palette_size[0] += 2;
if ((xd->plane[1].subsampling_x && xd->plane[1].subsampling_y)
|| !mbmi->palette_enabled[1])
mbmi->palette_indexed_size =
vp9_decode_uniform(r, MIN(mbmi->palette_size[0] + 1, 8));
else
mbmi->palette_indexed_size = 0;
mbmi->palette_literal_size = mbmi->palette_size[0] -
mbmi->palette_indexed_size;
if (PALETTE_DELTA_BIT)
@ -253,10 +259,10 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
else
mbmi->palette_delta_bitdepth = 0;
mbmi->palette_run_length =
mbmi->palette_run_length[0] =
vp9_read_literal(r, get_bit_depth(palette_max_run(bsize)));
mbmi->palette_run_length = (mbmi->palette_run_length) << 1;
mbmi->palette_scan_order = vp9_read_literal(r, 2);
mbmi->palette_run_length[0] = (mbmi->palette_run_length[0]) << 1;
mbmi->palette_scan_order[0] = vp9_read_literal(r, 2);
m1 = mbmi->palette_indexed_size;
m2 = mbmi->palette_literal_size;
@ -276,40 +282,22 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
memset(mbmi->palette_color_delta, 0,
m1 * sizeof(mbmi->palette_color_delta[0]));
}
}
if (m2 > 0) {
for (i = 0; i < m2; i++)
mbmi->palette_literal_colors[i] = vp9_read_literal(r, 8);
}
j = 0;
k = 0;
for (i = 0; i < mbmi->palette_size; i++) {
if (j < m1) {
c1 = cm->current_palette_colors[mbmi->palette_indexed_colors[j]];
if (mbmi->palette_color_delta[j])
c1 += mbmi->palette_color_delta[j];
} else {
c1 = 256;
for (i = 0; i < m1; i++) {
val = cm->current_palette_colors[mbmi->palette_indexed_colors[i]];
if (mbmi->palette_color_delta[i])
val += mbmi->palette_color_delta[i];
mbmi->palette_colors[i] = val;
}
if (k < m2)
c2 = mbmi->palette_literal_colors[k];
else
c2 = 256;
if (c1 < c2) {
mbmi->palette_colors[i] = c1;
j++;
} else if (c2 < c1) {
mbmi->palette_colors[i] = c2;
k++;
}
if (m2 > 0) {
for (i = 0; i < m2; i++) {
mbmi->palette_literal_colors[i] = vp9_read_literal(r, 8);
mbmi->palette_colors[m1 + i] = mbmi->palette_literal_colors[i];
}
}
d = get_bit_depth(rows * cols);
for (i = 0; i < mbmi->palette_run_length; i += 2) {
for (i = 0; i < mbmi->palette_run_length[0]; i += 2) {
mbmi->palette_runs[i] =
vp9_read_literal(r, get_bit_depth(m1 + m2));
@ -326,8 +314,8 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
&cm ->current_palette_size,
cm->current_palette_count, mbmi);
run_lengh_decoding(mbmi->palette_runs, mbmi->palette_run_length, map);
switch (mbmi->palette_scan_order) {
run_lengh_decoding(mbmi->palette_runs, mbmi->palette_run_length[0], map);
switch (mbmi->palette_scan_order[0]) {
case H_SCAN:
memcpy(xd->plane[0].color_index_map, map, rows * cols);
break;
@ -349,7 +337,81 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
}
mbmi->tx_size = MIN(max_txsize_lookup[bsize],
tx_mode_to_biggest_tx_size[cm->tx_mode]);
} else {
}
if (mbmi->palette_enabled[1]) {
int i, d;
int rows = 4 * num_4x4_blocks_high_lookup[bsize] >>
xd->plane[1].subsampling_y;
int cols = 4 * num_4x4_blocks_wide_lookup[bsize] >>
xd->plane[1].subsampling_x;
int scan_order[4096];
uint8_t map[4096];
PALETTE_RUN_LENGTH bits;
BLOCK_SIZE uv_bsize = get_plane_block_size(bsize, &xd->plane[1]);
mbmi->uv_mode = DC_PRED;
if (xd->plane[1].subsampling_x && xd->plane[1].subsampling_y) {
mbmi->palette_size[1] =
vp9_read_tree(r, vp9_palette_size_tree,
cm->fc.palette_uv_size_prob[uv_bsize - BLOCK_4X4]);
mbmi->palette_size[1] += 2;
mbmi->palette_run_length[1] =
vp9_read_literal(r, get_bit_depth(palette_max_run(uv_bsize)));
mbmi->palette_run_length[1] = (mbmi->palette_run_length[1]) << 1;
mbmi->palette_scan_order[1] = vp9_read_literal(r, 2);
} else {
mbmi->palette_size[1] = mbmi->palette_size[0];
}
for (i = 0; i < mbmi->palette_size[1]; i++)
mbmi->palette_colors[PALETTE_MAX_SIZE + i] = vp9_read_literal(r, 8);
for (i = 0; i < mbmi->palette_size[1]; i++)
mbmi->palette_colors[2 * PALETTE_MAX_SIZE + i] = vp9_read_literal(r, 8);
if (xd->plane[1].subsampling_x && xd->plane[1].subsampling_y) {
d = get_bit_depth(rows * cols);
for (i = 0; i < mbmi->palette_run_length[1]; i += 2) {
mbmi->palette_runs[PALETTE_MAX_RUNS + i] =
vp9_read_literal(r, get_bit_depth(mbmi->palette_size[1]));
bits = vp9_read_tree(r, vp9_palette_run_length_tree,
cm->fc.
palette_uv_run_length_prob[uv_bsize - BLOCK_4X4]);
if (bits == MAX_BITS)
mbmi->palette_runs[PALETTE_MAX_RUNS + i + 1] = vp9_read_literal(r, d);
else
mbmi->palette_runs[PALETTE_MAX_RUNS + i + 1] =
vp9_read_literal(r, bits - ONE_BITS + 1);
mbmi->palette_runs[ PALETTE_MAX_RUNS + i + 1] += 1;
}
run_lengh_decoding(mbmi->palette_runs + PALETTE_MAX_RUNS,
mbmi->palette_run_length[1], map);
switch (mbmi->palette_scan_order[1]) {
case H_SCAN:
memcpy(xd->plane[1].color_index_map, map, rows * cols);
break;
case V_SCAN:
transpose_block(map, xd->plane[1].color_index_map, cols, rows);
break;
case SPIN_SCAN:
spin_scan_order(scan_order, rows, cols);
for (i = 0; i < rows * cols; i++)
xd->plane[1].color_index_map[scan_order[i]] = map[i];
break;
case ZZ_SCAN:
zz_scan_order(scan_order, rows, cols);
for (i = 0; i < rows * cols; i++)
xd->plane[1].color_index_map[scan_order[i]] = map[i];
break;
default:
break;
}
}
}
if (!mbmi->palette_enabled[0]) {
mbmi->tx_size = read_tx_size(cm, xd, cm->tx_mode, bsize, 1, r);
}
#else
@ -451,7 +513,7 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
break;
default:
#if CONFIG_PALETTE
if (!mbmi->palette_enabled)
if (!mbmi->palette_enabled[0])
mbmi->mode = read_intra_mode(r,
get_y_mode_probs(mi, above_mi, left_mi, 0));
#else
@ -461,7 +523,7 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
#if CONFIG_FILTERINTRA
if (is_filter_enabled(mbmi->tx_size) && is_filter_allowed(mbmi->mode)
#if CONFIG_PALETTE
&& !mbmi->palette_enabled
&& !mbmi->palette_enabled[0]
#endif // CONFIG_PALETTE
)
mbmi->filterbit = vp9_read(r,
@ -471,10 +533,19 @@ static void read_intra_frame_mode_info(VP9_COMMON *const cm,
#endif // CONFIG_FILTERINTRA
}
#if CONFIG_PALETTE
if (!mbmi->palette_enabled[1])
mbmi->uv_mode = read_intra_mode(r, vp9_kf_uv_mode_prob[mbmi->mode]);
#else
mbmi->uv_mode = read_intra_mode(r, vp9_kf_uv_mode_prob[mbmi->mode]);
#endif
#if CONFIG_FILTERINTRA
if (is_filter_enabled(get_uv_tx_size(mbmi, &xd->plane[1])) &&
is_filter_allowed(mbmi->uv_mode))
is_filter_allowed(mbmi->uv_mode)
#if CONFIG_PALETTE
&& !mbmi->palette_enabled[1]
#endif // CONFIG_PALETTE
)
mbmi->uv_filterbit = vp9_read(r,
cm->fc.filterintra_prob[get_uv_tx_size(mbmi, &xd->plane[1])][mbmi->uv_mode]);
else
@ -712,7 +783,9 @@ static void read_intra_block_mode_info(VP9_COMMON *const cm, MODE_INFO *mi,
#endif // CONFIG_FILTERINTRA
#if CONFIG_PALETTE
mbmi->palette_enabled = 0;
for (i = 0; i < 2; ++i) {
mbmi->palette_enabled[i] = 0;
}
#endif // CONFIG_PALETTE
}

View File

@ -735,26 +735,30 @@ static void write_mb_modes_kf(const VP9_COMMON *cm, const MACROBLOCKD *xd,
#if CONFIG_PALETTE
if (bsize >= BLOCK_8X8 && cm->allow_palette_mode) {
int l, m1, m2, i, d, bits;
int rows = 4 * num_4x4_blocks_high_lookup[bsize];
int cols = 4 * num_4x4_blocks_wide_lookup[bsize];
vp9_write_bit(w, mbmi->palette_enabled);
if (mbmi->palette_enabled) {
vp9_write_bit(w, mbmi->palette_enabled[0]);
vp9_write_bit(w, mbmi->palette_enabled[1]);
if (mbmi->palette_enabled[0]) {
int rows = 4 * num_4x4_blocks_high_lookup[bsize];
int cols = 4 * num_4x4_blocks_wide_lookup[bsize];
m1 = mbmi->palette_indexed_size;
m2 = mbmi->palette_literal_size;
l = mbmi->palette_run_length;
l = mbmi->palette_run_length[0];
vp9_write_token(w, vp9_palette_size_tree,
cm->fc.palette_size_prob[bsize - BLOCK_8X8],
&palette_size_encodings[mbmi->palette_size - 2]);
vp9_encode_uniform(w, MIN(mbmi->palette_size + 1, 8),
mbmi->palette_indexed_size);
&palette_size_encodings[mbmi->palette_size[0] - 2]);
if ((xd->plane[1].subsampling_x && xd->plane[1].subsampling_y)
|| !mbmi->palette_enabled[1])
vp9_encode_uniform(w, MIN(mbmi->palette_size[0] + 1, 8),
mbmi->palette_indexed_size);
if (PALETTE_DELTA_BIT)
vp9_write_literal(w, mbmi->palette_delta_bitdepth, PALETTE_DELTA_BIT);
vp9_write_literal(w, (l >> 1),
get_bit_depth(palette_max_run(bsize)));
vp9_write_literal(w, mbmi->palette_scan_order, 2);
vp9_write_literal(w, mbmi->palette_scan_order[0], 2);
if (m1 > 0) {
for (i = 0; i < m1; i++)
@ -768,7 +772,6 @@ static void write_mb_modes_kf(const VP9_COMMON *cm, const MACROBLOCKD *xd,
}
}
}
if (m2 > 0) {
for (i = 0; i < m2; i++)
vp9_write_literal(w, mbmi->palette_literal_colors[i], 8);
@ -787,10 +790,50 @@ static void write_mb_modes_kf(const VP9_COMMON *cm, const MACROBLOCKD *xd,
bits > 6 ? d : bits);
}
}
if (mbmi->palette_enabled[1]) {
const uint16_t *runs = mbmi->palette_runs + PALETTE_MAX_RUNS;
int rows = 4 * num_4x4_blocks_high_lookup[bsize] >>
xd->plane[1].subsampling_y;
int cols = 4 * num_4x4_blocks_wide_lookup[bsize] >>
xd->plane[1].subsampling_x;
BLOCK_SIZE uv_bsize = get_plane_block_size(bsize, &xd->plane[1]);
l = mbmi->palette_run_length[1];
if (xd->plane[1].subsampling_x && xd->plane[1].subsampling_y) {
vp9_write_token(w, vp9_palette_size_tree,
cm->fc.palette_uv_size_prob[uv_bsize - BLOCK_4X4],
&palette_size_encodings[mbmi->palette_size[1] - 2]);
vp9_write_literal(w, (l >> 1),
get_bit_depth(palette_max_run(uv_bsize)));
vp9_write_literal(w, mbmi->palette_scan_order[1], 2);
}
for (i = 0; i < mbmi->palette_size[1]; i++)
vp9_write_literal(w, mbmi->palette_colors[PALETTE_MAX_SIZE + i], 8);
for (i = 0; i < mbmi->palette_size[1]; i++)
vp9_write_literal(w, mbmi->palette_colors[2 * PALETTE_MAX_SIZE + i], 8);
if (xd->plane[1].subsampling_x && xd->plane[1].subsampling_y) {
d = get_bit_depth(rows * cols);
for (i = 0; i < l; i += 2) {
vp9_write_literal(w, runs[i],
get_bit_depth(mbmi->palette_size[1]));
bits = get_bit_depth(runs[i + 1]);
vp9_write_token(w, vp9_palette_run_length_tree,
cm->fc.palette_uv_run_length_prob[uv_bsize -
BLOCK_4X4],
&palette_run_length_encodings[bits > 6 ?
6 : bits - 1]);
vp9_write_literal(w, runs[i + 1] - 1,
bits > 6 ? d : bits);
}
}
}
}
if (bsize >= BLOCK_8X8 && cm->tx_mode == TX_MODE_SELECT &&
!mbmi->palette_enabled)
!mbmi->palette_enabled[0])
#else
if (bsize >= BLOCK_8X8 && cm->tx_mode == TX_MODE_SELECT)
#endif
@ -818,7 +861,7 @@ static void write_mb_modes_kf(const VP9_COMMON *cm, const MACROBLOCKD *xd,
if (bsize >= BLOCK_8X8) {
#if CONFIG_PALETTE
if (!mbmi->palette_enabled)
if (!mbmi->palette_enabled[0])
write_intra_mode(w, mbmi->mode,
get_y_mode_probs(mi, above_mi, left_mi, 0));
#else
@ -827,7 +870,7 @@ static void write_mb_modes_kf(const VP9_COMMON *cm, const MACROBLOCKD *xd,
#if CONFIG_FILTERINTRA
if (is_filter_allowed(mbmi->mode) && is_filter_enabled(mbmi->tx_size)
#if CONFIG_PALETTE
&& !mbmi->palette_enabled
&& !mbmi->palette_enabled[0]
#endif // CONFIG_PALETTE
)
vp9_write(w, mbmi->filterbit,
@ -852,13 +895,22 @@ static void write_mb_modes_kf(const VP9_COMMON *cm, const MACROBLOCKD *xd,
}
}
#if CONFIG_PALETTE
if (!mbmi->palette_enabled[1])
write_intra_mode(w, mbmi->uv_mode, vp9_kf_uv_mode_prob[mbmi->mode]);
#else
write_intra_mode(w, mbmi->uv_mode, vp9_kf_uv_mode_prob[mbmi->mode]);
#endif // CONFIG_PALETTE
#if CONFIG_FILTERINTRA
if (is_filter_allowed(mbmi->uv_mode) &&
is_filter_enabled(get_uv_tx_size(mbmi, &xd->plane[1])))
is_filter_enabled(get_uv_tx_size(mbmi, &xd->plane[1]))
#if CONFIG_PALETTE
&& !mbmi->palette_enabled[1]
#endif // CONFIG_PALETTE
)
vp9_write(w, mbmi->uv_filterbit,
cm->fc.filterintra_prob[get_uv_tx_size(mbmi, &xd->plane[1])][mbmi->uv_mode]);
#endif
#endif // CONFIG_FILTERINTRA
}
static void write_modes_b(VP9_COMP *cpi, const TileInfo *const tile,

View File

@ -44,8 +44,11 @@ static void alloc_mode_context(VP9_COMMON *cm, int num_4x4_blk,
}
}
#if CONFIG_PALETTE
CHECK_MEM_ERROR(cm, ctx->color_index_map,
vpx_memalign(16, num_pix * sizeof(ctx->color_index_map[0])));
for (i = 0; i < 2; i++) {
CHECK_MEM_ERROR(cm, ctx->color_index_map[i],
vpx_memalign(16, num_pix *
sizeof(ctx->color_index_map[i][0])));
}
#endif
}
@ -67,8 +70,10 @@ static void free_mode_context(PICK_MODE_CONTEXT *ctx) {
}
#if CONFIG_PALETTE
vpx_free(ctx->color_index_map);
ctx->color_index_map = 0;
for (i = 0; i < 2; i++) {
vpx_free(ctx->color_index_map[i]);
ctx->color_index_map[i] = 0;
}
#endif
}

View File

@ -31,7 +31,7 @@ typedef struct {
tran_low_t *dqcoeff_pbuf[MAX_MB_PLANE][3];
uint16_t *eobs_pbuf[MAX_MB_PLANE][3];
#if CONFIG_PALETTE
uint8_t *color_index_map;
uint8_t *color_index_map[2];
uint8_t palette_colors_buf[PALETTE_BUF_SIZE];
int palette_buf_size;
int palette_count_buf[PALETTE_BUF_SIZE];

View File

@ -783,7 +783,9 @@ static void update_state(VP9_COMP *cpi, PICK_MODE_CONTEXT *ctx,
}
#if CONFIG_PALETTE
pd[0].color_index_map = ctx->color_index_map;
for (i = 0; i < 2; i++) {
pd[i].color_index_map = ctx->color_index_map[i];
}
#endif
// Restore the coding context of the MB to that that was in place
@ -1285,7 +1287,9 @@ static void rd_pick_sb_modes(VP9_COMP *cpi, const TileInfo *const tile,
p[i].eobs = ctx->eobs_pbuf[i][0];
}
#if CONFIG_PALETTE
pd[0].color_index_map = ctx->color_index_map;
for (i = 0; i < 2; ++i) {
pd[i].color_index_map = ctx->color_index_map[i];
}
#endif
ctx->is_coded = 0;
ctx->skippable = 0;
@ -4964,14 +4968,14 @@ static void encode_superblock(VP9_COMP *cpi, TOKENEXTRA **t, int output_enabled,
mi);
vp9_tokenize_sb(cpi, t, !output_enabled, MAX(bsize, BLOCK_8X8));
#if CONFIG_PALETTE
if (mbmi->palette_enabled && output_enabled) {
if (mbmi->palette_enabled[0] && output_enabled) {
palette_color_insertion(cm->current_palette_colors,
&cm ->current_palette_size,
cm->current_palette_count, mbmi);
}
if (frame_is_intra_only(cm) && output_enabled && bsize >= BLOCK_8X8) {
cm->block_counter++;
if (mbmi->palette_enabled)
if (mbmi->palette_enabled[0])
cm->palette_counter++;
}
#endif

View File

@ -873,13 +873,13 @@ static void super_block_yrd(VP9_COMP *cpi, MACROBLOCK *x, int *rate,
(void) xd;
if (cpi->sf.tx_size_search_method == USE_LARGESTALL
#if CONFIG_PALETTE
|| xd->mi[0].src_mi->mbmi.palette_enabled
|| xd->mi[0].src_mi->mbmi.palette_enabled[0]
#endif // CONFIG_PALETTE
) {
#else
if (cpi->sf.tx_size_search_method == USE_LARGESTALL || xd->lossless
#if CONFIG_PALETTE
|| xd->mi[0].src_mi->mbmi.palette_enabled
|| xd->mi[0].src_mi->mbmi.palette_enabled[0]
#endif // CONFIG_PALETTE
) {
#endif // CONFIG_TX_SKIP
@ -1124,7 +1124,8 @@ static int64_t rd_pick_intra4x4block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
xd->mi[0].src_mi->mbmi.tx_skip[1] = 0;
#endif // CONFIG_TX_SKIP
#if CONFIG_PALETTE
xd->mi[0].src_mi->mbmi.palette_enabled = 0;
xd->mi[0].src_mi->mbmi.palette_enabled[0] = 0;
xd->mi[0].src_mi->mbmi.palette_enabled[1] = 0;
#endif // CONFIG_PALETTE
for (idy = 0; idy < num_4x4_blocks_high; ++idy) {
for (idx = 0; idx < num_4x4_blocks_wide; ++idx) {
@ -1232,6 +1233,9 @@ static int64_t rd_pick_intra_sub_8x8_y_mode(VP9_COMP *cpi, MACROBLOCK *mb,
vpx_memcpy(t_above, xd->plane[0].above_context, sizeof(t_above));
vpx_memcpy(t_left, xd->plane[0].left_context, sizeof(t_left));
#if CONFIG_PALETTE
mic->mbmi.palette_enabled[0] = 0;
#endif
// Pick modes for each sub-block (of size 4x4, 4x8, or 8x4) in an 8x8 block.
for (idy = 0; idy < 2; idy += num_4x4_blocks_high) {
@ -1312,7 +1316,7 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
int *bmode_costs;
#if CONFIG_FILTERINTRA
int mode_ext, fbit, fbit_selected = 0;
#endif
#endif // CONFIG_FILTERINTRA
const MODE_INFO *above_mi = xd->mi[-xd->mi_stride].src_mi;
const MODE_INFO *left_mi = xd->left_available ? xd->mi[-1].src_mi : NULL;
const PREDICTION_MODE A = vp9_above_block_mode(mic, above_mi, 0);
@ -1322,7 +1326,7 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
int q_idx = vp9_get_qindex(&cpi->common.seg, mic->mbmi.segment_id,
cpi->common.base_qindex);
int try_tx_skip = q_idx <= TX_SKIP_Q_THRESH_INTRA;
#endif
#endif // CONFIG_TX_SKIP
#if CONFIG_PALETTE
int palette_selected = 0, best_n = 0, best_l = 0, colors;
int best_m1 = 0, best_m2 = 0, palette_delta_bitdepth = 0;
@ -1335,7 +1339,7 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
uint8_t best_index[PALETTE_MAX_SIZE], best_literal[PALETTE_MAX_SIZE];
int8_t palette_color_delta[PALETTE_MAX_SIZE];
PALETTE_SCAN_ORDER best_ps = H_SCAN;
#endif
#endif // CONFIG_PALETTE
bmode_costs = cpi->y_mode_costs[A][L];
if (cpi->sf.tx_size_search_method == USE_FULL_RD)
@ -1360,11 +1364,11 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
#endif
#if CONFIG_TX_SKIP
mic->mbmi.tx_skip[0] = 0;
#endif
#endif // CONFIG_TX_SKIP
#if CONFIG_PALETTE
mic->mbmi.palette_enabled = 0;
#endif
mic->mbmi.palette_enabled[0] = 0;
#endif // CONFIG_PALETTE
super_block_yrd(cpi, x, &this_rate_tokenonly, &this_distortion,
&s, NULL, bsize, local_tx_cache, best_rd);
@ -1465,12 +1469,10 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
#endif
#if CONFIG_PALETTE
mic->mbmi.palette_enabled = 0;
mic->mbmi.palette_size = 0;
mic->mbmi.current_palette_size = cpi->common.current_palette_size;
colors = count_colors(src, src_stride, rows, cols);
if (colors > 1 && colors <= 64 && cpi->common.allow_palette_mode) {
int n, r, c, i, j, temp, max_itr = 50, k;
int n, r, c, i, j, temp, max_itr = 200, k;
int indices[4096];
int l, m1, m2, d = get_bit_depth(rows * cols);
int bits, best_bits = 0, total_bits, best_total_bits;
@ -1485,11 +1487,11 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
#if CONFIG_TX_SKIP
int this_rate_tokenonly_s, s_s;
int64_t this_distortion_s;
#endif
#endif // CONFIG_TX_SKIP
memset(data, 0, sizeof(data[0] * 4096));
memset(indices, 0, sizeof(indices[0] * 4096));
mic->mbmi.palette_enabled = 1;
mic->mbmi.palette_enabled[0] = 1;
vp9_cost_tokens(palette_size_cost,
cpi->common.fc.palette_size_prob[bsize - BLOCK_8X8],
vp9_palette_size_tree);
@ -1498,7 +1500,8 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
vp9_palette_run_length_tree);
#if CONFIG_FILTERINTRA
mic->mbmi.filterbit = 0;
#endif
#endif // CONFIG_FILTERINTRA
mic->mbmi.mode = DC_PRED;
for (r = 0; r < rows; r++) {
for (c = 0; c < cols; c++) {
val = src[r * src_stride + c];
@ -1531,7 +1534,6 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
}
}
mic->mbmi.palette_size = k;
for (i = 0; i < k; i++)
mic->mbmi.palette_colors[i] = clip_pixel(round(centroids[i]));
@ -1583,10 +1585,19 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
}
if (m1 == PALETTE_MAX_SIZE)
continue;
mic->mbmi.palette_indexed_size = m1;
mic->mbmi.palette_literal_size = m2;
mic->mbmi.palette_delta_bitdepth = best_bits;
if (m1 > 0)
for (i = 0; i < m1; i++) {
mic->mbmi.palette_colors[i] =
cpi->common.current_palette_colors
[mic->mbmi.palette_indexed_colors[i]];
}
if (m2 > 0)
memcpy(mic->mbmi.palette_colors + m1, mic->mbmi.palette_literal_colors,
m2 * sizeof(mic->mbmi.palette_colors[0]));
for (i = 0; i < k; i++) {
centroids[i] = (double) mic->mbmi.palette_colors[i];
}
calc_indices(data, centroids, indices, rows * cols, k, 1);
for (r = 0; r < rows; r++) {
for (c = 0; c < cols; c++) {
@ -1594,10 +1605,44 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
}
}
#if CONFIG_TX_SKIP
mic->mbmi.tx_skip[0] = 0;
#endif // CONFIG_TX_SKIP
super_block_yrd(cpi, x, &this_rate_tokenonly, &this_distortion,
&s, NULL, bsize, local_tx_cache, best_rd);
#if CONFIG_TX_SKIP
if (try_tx_skip) {
if (this_rate_tokenonly != INT_MAX)
this_rate_tokenonly +=
vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0], 0);
mic->mbmi.tx_skip[0] = 1;
super_block_yrd(cpi, x, &this_rate_tokenonly_s, &this_distortion_s,
&s_s, NULL, bsize, local_tx_cache, best_rd);
if (this_rate_tokenonly_s != INT_MAX)
this_rate_tokenonly_s +=
vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0], 1);
if ((this_rate_tokenonly_s != INT_MAX &&
this_rate_tokenonly == INT_MAX) ||
(RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly, this_distortion)
> RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly_s,
this_distortion_s))) {
mic->mbmi.tx_skip[0] = 1;
} else {
mic->mbmi.tx_skip[0] = 0;
}
} else {
mic->mbmi.tx_skip[0] = 0;
}
super_block_yrd(cpi, x, &this_rate_tokenonly, &this_distortion,
&s, NULL, bsize, local_tx_cache, best_rd);
#endif // CONFIG_TX_SKIP
if (this_rate_tokenonly == INT_MAX) {
continue;
}
for (ps = H_SCAN; ps < PALETTE_SCAN_ORDERS; ps++) {
int scan_order[4096];
mic->mbmi.palette_scan_order = ps;
switch (ps) {
case H_SCAN:
memcpy(map, xd->plane[0].color_index_map, rows * cols);
@ -1626,52 +1671,11 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
continue;
}
#if CONFIG_TX_SKIP
mic->mbmi.tx_skip[0] = 0;
#endif
super_block_yrd(cpi, x, &this_rate_tokenonly, &this_distortion,
&s, NULL, bsize, local_tx_cache, best_rd);
#if CONFIG_TX_SKIP
if (try_tx_skip) {
if (this_rate_tokenonly != INT_MAX)
this_rate_tokenonly +=
vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0], 0);
mic->mbmi.tx_skip[0] = 1;
super_block_yrd(cpi, x, &this_rate_tokenonly_s, &this_distortion_s,
&s_s, NULL, bsize, local_tx_cache, best_rd);
if (this_rate_tokenonly_s != INT_MAX)
this_rate_tokenonly_s +=
vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0], 1);
if ((this_rate_tokenonly_s != INT_MAX &&
this_rate_tokenonly == INT_MAX) ||
(RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly, this_distortion)
> RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly_s,
this_distortion_s))) {
mic->mbmi.tx_skip[0] = 1;
} else {
mic->mbmi.tx_skip[0] = 0;
}
} else {
mic->mbmi.tx_skip[0] = 0;
}
super_block_yrd(cpi, x, &this_rate_tokenonly, &this_distortion,
&s, NULL, bsize, local_tx_cache, best_rd);
if (this_rate_tokenonly != INT_MAX)
this_rate_tokenonly += vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0],
mic->mbmi.tx_skip[0]);
#endif
if (this_rate_tokenonly == INT_MAX) {
continue;
}
this_rate = this_rate_tokenonly +
(1 + vp9_encode_uniform_cost(MIN(k + 1, 8), m1) + PALETTE_DELTA_BIT
+ get_bit_depth(palette_max_run(bsize)) + 2 +
get_bit_depth(mic->mbmi.current_palette_size) * m1 +
mic->mbmi.palette_delta_bitdepth * m1 +
best_bits * m1 +
8 * m2 + get_bit_depth(k) * (l >> 1)) * vp9_cost_bit(128, 0) +
palette_size_cost[k - 2];
for (i = 0; i < l; i += 2) {
@ -1679,6 +1683,10 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
this_rate += palette_run_length_cost[bits > 6 ? 6 : bits - 1];
this_rate += (bits > 6 ? d : bits) * vp9_cost_bit(128, 0);
}
#if CONFIG_TX_SKIP
this_rate += vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0],
mic->mbmi.tx_skip[0]);
#endif // CONFIG_TX_SKIP
this_rd = RDCOST(x->rdmult, x->rddiv, this_rate, this_distortion);
if (this_rd < best_rd) {
mode_selected = DC_PRED;
@ -1692,9 +1700,9 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
best_l = l;
palette_selected = 1;
best_ps = ps;
best_m1 = mic->mbmi.palette_indexed_size;
best_m2 = mic->mbmi.palette_literal_size;
palette_delta_bitdepth = mic->mbmi.palette_delta_bitdepth;
best_m1 = m1;
best_m2 = m2;
palette_delta_bitdepth = best_bits;
memcpy(best_palette, mic->mbmi.palette_colors,
k * sizeof(best_palette[0]));
memcpy(best_runs, mic->mbmi.palette_runs, l * sizeof(best_runs[0]));
@ -1721,17 +1729,17 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
mic->mbmi.filterbit = fbit_selected;
else
mic->mbmi.filterbit = 0;
#endif
#endif // CONFIG_FILTERINTRA
mic->mbmi.tx_size = best_tx;
#if CONFIG_TX_SKIP
mic->mbmi.tx_skip[0] = tx_skipped;
#endif
#endif // CONFIG_TX_SKIP
#if CONFIG_PALETTE
mic->mbmi.palette_enabled = palette_selected;
mic->mbmi.palette_enabled[0] = palette_selected;
if (palette_selected) {
mic->mbmi.palette_size = best_n;
mic->mbmi.palette_run_length = best_l;
mic->mbmi.palette_scan_order = best_ps;
mic->mbmi.palette_size[0] = best_n;
mic->mbmi.palette_run_length[0] = best_l;
mic->mbmi.palette_scan_order[0] = best_ps;
mic->mbmi.palette_indexed_size = best_m1;
mic->mbmi.palette_literal_size = best_m2;
mic->mbmi.palette_delta_bitdepth = palette_delta_bitdepth;
@ -1751,7 +1759,7 @@ static int64_t rd_pick_intra_sby_mode(VP9_COMP *cpi, MACROBLOCK *x,
mic->mbmi.filterbit = 0;
#endif // CONFIG_FILTERINTRA
}
#endif
#endif // CONFIG_PALETTE
return best_rd;
}
@ -1822,15 +1830,30 @@ static int64_t rd_pick_intra_sbuv_mode(VP9_COMP *cpi, MACROBLOCK *x,
int64_t this_distortion, this_sse;
#if CONFIG_FILTERINTRA
int mode_ext, fbit = 0, fbit_selected = 0;
#endif
#endif // CONFIG_FILTERINTRA
#if CONFIG_TX_SKIP
int tx_skipped = 0;
int q_idx = vp9_get_qindex(&cpi->common.seg,
xd->mi[0].src_mi->mbmi.segment_id,
cpi->common.base_qindex);
int try_tx_skip = q_idx <= TX_SKIP_Q_THRESH_INTRA;
#endif
#endif // CONFIG_TX_SKIP
#if CONFIG_PALETTE
int palette_selected = 0, best_n = 0, best_l = 0;
int rows = (4 * num_4x4_blocks_high_lookup[bsize]) >>
(xd->plane[1].subsampling_x);
int cols = (4 * num_4x4_blocks_wide_lookup[bsize]) >>
(xd->plane[1].subsampling_y);
int src_stride = x->plane[1].src.stride;
uint16_t best_runs[PALETTE_MAX_RUNS];
uint8_t best_palette[2 * PALETTE_MAX_SIZE], best_map[4096];
uint8_t *src_u = x->plane[1].src.buf;
uint8_t *src_v = x->plane[2].src.buf;
PALETTE_SCAN_ORDER best_ps;
MB_MODE_INFO *mbmi = &xd->mi[0].src_mi->mbmi;
xd->mi[0].src_mi->mbmi.palette_enabled[1] = 0;
#endif // CONFIG_PALETTE
vpx_memset(x->skip_txfm, 0, sizeof(x->skip_txfm));
#if CONFIG_FILTERINTRA
(void) max_tx_size;
@ -1865,13 +1888,20 @@ static int64_t rd_pick_intra_sbuv_mode(VP9_COMP *cpi, MACROBLOCK *x,
if (try_tx_skip)
this_rate += vp9_cost_bit(
cpi->common.fc.uv_tx_skip_prob[xd->mi[0].src_mi->mbmi.tx_skip[0]], 0);
#endif
#endif // CONFIG_TX_SKIP
#if CONFIG_FILTERINTRA
if (is_filter_allowed(mode) &&
is_filter_enabled(get_uv_tx_size(&(x->e_mbd.mi[0].mbmi), &xd->plane[1])))
this_rate += vp9_cost_bit(cpi->common.fc.filterintra_prob
[get_uv_tx_size(&(x->e_mbd.mi[0].mbmi), &xd->plane[1])][mode], fbit);
#endif
#endif // CONFIG_FILTERINTRA
#if CONFIG_PALETTE
if (frame_is_intra_only(&cpi->common) &&
xd->mi[0].src_mi->mbmi.sb_type >= BLOCK_8X8 &&
xd->plane[1].subsampling_x && xd->plane[1].subsampling_y &&
cpi->common.allow_palette_mode)
this_rate += vp9_cost_bit(128, 0);
#endif // CONFIG_PALETTE
this_rd = RDCOST(x->rdmult, x->rddiv, this_rate, this_distortion);
if (this_rd < best_rd) {
@ -1951,13 +1981,223 @@ static int64_t rd_pick_intra_sbuv_mode(VP9_COMP *cpi, MACROBLOCK *x,
}
#endif
#if CONFIG_PALETTE
if (frame_is_intra_only(&cpi->common) &&
xd->mi[0].src_mi->mbmi.sb_type >= BLOCK_8X8 &&
xd->plane[1].subsampling_x && xd->plane[1].subsampling_y &&
cpi->common.allow_palette_mode) {
int colors_u = count_colors(src_u, src_stride, rows, cols);
int colors_v = count_colors(src_v, src_stride, rows, cols);
int colors = colors_u > colors_v ? colors_u : colors_v;
if (colors > 1 && colors <= 64) {
int n, r, c, i, j, max_itr = 200;
int indices[4096];
int l, d = get_bit_depth(rows * cols);
int palette_run_length_cost[PALETTE_RUN_LENGTHS];
int palette_size_cost[PALETTE_SIZES];
double data[2048];
double centroids[2 * PALETTE_MAX_SIZE];
double lb_u = src_u[0], ub_u = src_u[0];
double lb_v = src_v[0], ub_v = src_v[0], val;
PALETTE_SCAN_ORDER ps;
uint8_t map[4096];
BLOCK_SIZE uv_bsize = get_plane_block_size(bsize, &xd->plane[1]);
#if CONFIG_TX_SKIP
int this_rate_tokenonly_s, s_s;
int64_t this_distortion_s;
#endif // CONFIG_TX_SKIP
i = uv_bsize - BLOCK_4X4;
vp9_cost_tokens(palette_size_cost,
cpi->common.fc.palette_uv_size_prob[i],
vp9_palette_size_tree);
vp9_cost_tokens(palette_run_length_cost,
cpi->common.fc.palette_uv_run_length_prob[i],
vp9_palette_run_length_tree);
#if CONFIG_FILTERINTRA
mbmi->uv_filterbit = 0;
#endif // CONFIG_FILTERINTRA
#if CONFIG_TX_SKIP
mbmi->tx_skip[1] = 0;
#endif // CONFIG_TX_SKIP
mbmi->palette_enabled[1] = 1;
mbmi->uv_mode = DC_PRED;
for (r = 0; r < rows; r++) {
for (c = 0; c < cols; c++) {
data[(r * cols + c) * 2 ] = src_u[r * src_stride + c];
data[(r * cols + c) * 2 + 1] = src_v[r * src_stride + c];
val = src_u[r * src_stride + c];
if (val < lb_u)
lb_u = val;
else if (val > ub_u)
ub_u = val;
val = src_v[r * src_stride + c];
if (val < lb_v)
lb_v = val;
else if (val > ub_v)
ub_v = val;
}
}
for (n = colors > PALETTE_MAX_SIZE ? PALETTE_MAX_SIZE : colors;
n >= 2; n--) {
for (i = 0; i < n; i++) {
centroids[i * 2] = lb_u + (2 * i + 1) * (ub_u - lb_u) / n / 2;
centroids[i * 2 + 1] =
lb_v + (2 * i + 1) * (ub_v - lb_v) / n / 2;;
}
r = k_means(data, centroids, indices, rows * cols, n, 2, max_itr);
mbmi->palette_size[1] = n;
for (i = 1; i < 3; i++) {
for (j = 0; j < n; j++)
mbmi->palette_colors[i * PALETTE_MAX_SIZE + j] =
clip_pixel(round(centroids[j * 2 + i - 1]));
}
for (r = 0; r < rows; r++)
for (c = 0; c < cols; c++) {
xd->plane[1].color_index_map[r * cols + c] =
indices[r * cols + c];
}
#if CONFIG_TX_SKIP
mbmi->tx_skip[1] = 0;
#endif // CONFIG_TX_SKIP
super_block_uvrd(cpi, x, &this_rate_tokenonly,
&this_distortion, &s, &this_sse, bsize, best_rd);
#if CONFIG_TX_SKIP
if (try_tx_skip) {
if (this_rate_tokenonly != INT_MAX)
this_rate_tokenonly +=
vp9_cost_bit(cpi->common.fc.
uv_tx_skip_prob[mbmi->tx_skip[0]], 0);
mbmi->tx_skip[1] = 1;
super_block_uvrd(cpi, x, &this_rate_tokenonly_s, &this_distortion_s,
&s_s, &this_sse, bsize, best_rd);
if (this_rate_tokenonly_s != INT_MAX)
this_rate_tokenonly_s +=
vp9_cost_bit(cpi->common.fc.
uv_tx_skip_prob[mbmi->tx_skip[0]], 1);
if ((this_rate_tokenonly_s != INT_MAX &&
this_rate_tokenonly == INT_MAX) ||
(RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly, this_distortion)
> RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly_s,
this_distortion_s))) {
mbmi->tx_skip[1] = 1;
} else {
mbmi->tx_skip[1] = 0;
}
} else {
mbmi->tx_skip[1] = 0;
}
super_block_uvrd(cpi, x, &this_rate_tokenonly,
&this_distortion, &s, &this_sse, bsize, best_rd);
#endif // CONFIG_TX_SKIP
if (this_rate_tokenonly == INT_MAX) {
continue;
}
for (ps = H_SCAN; ps < PALETTE_SCAN_ORDERS; ps++) {
int scan_order[4096];
switch (ps) {
case H_SCAN:
memcpy(map, xd->plane[1].color_index_map, rows * cols);
break;
case V_SCAN:
transpose_block(xd->plane[1].color_index_map,
map, rows, cols);
break;
case SPIN_SCAN:
spin_scan_order(scan_order, rows, cols);
for (i = 0; i < rows * cols; i++)
map[i] = xd->plane[1].color_index_map[scan_order[i]];
break;
case ZZ_SCAN:
zz_scan_order(scan_order, rows, cols);
for (i = 0; i < rows * cols; i++)
map[i] = xd->plane[1].color_index_map[scan_order[i]];
break;
default:
break;
}
l = run_lengh_encoding(map, rows * cols,
mbmi->palette_runs + PALETTE_MAX_RUNS,
palette_max_run(uv_bsize));
if (!l) {
continue;
}
this_rate = this_rate_tokenonly +
(1 + get_bit_depth(palette_max_run(uv_bsize)) + 2 + 2 * 8 * n +
get_bit_depth(n) * (l >> 1)) * vp9_cost_bit(128, 0) +
palette_size_cost[n - 2];
for (i = 0; i < l; i += 2) {
int bits = get_bit_depth(
mbmi->palette_runs[PALETTE_MAX_RUNS + i + 1]);
this_rate += palette_run_length_cost[bits > 6 ? 6 : bits - 1];
this_rate += (bits > 6 ? d : bits) * vp9_cost_bit(128, 0);
}
#if CONFIG_TX_SKIP
this_rate += vp9_cost_bit(cpi->common.fc.
uv_tx_skip_prob[mbmi->tx_skip[0]],
mbmi->tx_skip[1]);
#endif // CONFIG_TX_SKIP
this_rd = RDCOST(x->rdmult, x->rddiv, this_rate, this_distortion);
if (this_rd < best_rd) {
mode_selected = DC_PRED;
best_rd = this_rd;
*rate = this_rate;
*rate_tokenonly = this_rate_tokenonly;
*distortion = this_distortion;
*skippable = s;
best_n = n;
best_l = l;
palette_selected = 1;
best_ps = ps;
memcpy(best_palette, mbmi->palette_colors + PALETTE_MAX_SIZE,
2 * PALETTE_MAX_SIZE * sizeof(best_palette[0]));
memcpy(best_runs, mbmi->palette_runs + PALETTE_MAX_RUNS,
l * sizeof(best_runs[0]));
memcpy(best_map, xd->plane[1].color_index_map,
rows * cols * sizeof(best_map[0]));
#if CONFIG_TX_SKIP
tx_skipped = mbmi->tx_skip[1];
#endif // CONFIG_TX_SKIP
#if CONFIG_FILTERINTRA
fbit_selected = 0;
#endif // CONFIG_FILTERINTRA
}
}
}
}
}
#endif // CONFIG_PALETTE
xd->mi[0].src_mi->mbmi.uv_mode = mode_selected;
#if CONFIG_FILTERINTRA
xd->mi[0].mbmi.uv_filterbit = fbit_selected;
#endif
#endif // CONFIG_FILTERINTRA
#if CONFIG_TX_SKIP
xd->mi[0].src_mi->mbmi.tx_skip[1] = tx_skipped;
#endif
#endif // CONFIG_TX_SKIP
#if CONFIG_PALETTE
mbmi->palette_enabled[1] = palette_selected;
if (palette_selected) {
mbmi->palette_size[1] = best_n;
mbmi->palette_run_length[1] = best_l;
mbmi->palette_scan_order[1] = best_ps;
memcpy(mbmi->palette_colors + PALETTE_MAX_SIZE, best_palette,
2 * PALETTE_MAX_SIZE * sizeof(best_palette[0]));
memcpy(mbmi->palette_runs + PALETTE_MAX_RUNS, best_runs,
best_l * sizeof(best_runs[0]));
memcpy(xd->plane[1].color_index_map, best_map,
rows * cols * sizeof(best_map[0]));
}
#endif // CONFIG_PALETTE
return best_rd;
}
@ -4562,14 +4802,6 @@ void vp9_rd_pick_intra_mode_sb(VP9_COMP *cpi, MACROBLOCK *x,
rd_cost->rate = INT_MAX;
return;
}
#if CONFIG_PALETTE
if (xd->mi[0].src_mi->mbmi.palette_enabled) {
palette_color_insertion(ctx->palette_colors_buf,
&ctx->palette_buf_size,
ctx->palette_count_buf,
&(xd->mi[0].src_mi->mbmi));
}
#endif
} else {
y_skip = 0;
if (rd_pick_intra_sub_8x8_y_mode(cpi, x, &rate_y, &rate_y_tokenonly,
@ -4606,6 +4838,302 @@ void vp9_rd_pick_intra_mode_sb(VP9_COMP *cpi, MACROBLOCK *x,
ctx->mic = *xd->mi[0].src_mi;
rd_cost->rdcost = RDCOST(x->rdmult, x->rddiv, rd_cost->rate, rd_cost->dist);
#if CONFIG_PALETTE
if (bsize >= BLOCK_8X8 && !pd[1].subsampling_x && !pd[1].subsampling_y) {
uint8_t *src_y = x->plane[0].src.buf;
uint8_t *src_u = x->plane[1].src.buf;
uint8_t *src_v = x->plane[2].src.buf;
int rows = 4 * num_4x4_blocks_high_lookup[bsize];
int cols = 4 * num_4x4_blocks_wide_lookup[bsize];
int src_stride_y = x->plane[0].src.stride;
int src_stride_uv = x->plane[1].src.stride;
int colors = count_colors(src_y, src_stride_y, rows, cols);
if (colors >= 2 && colors <= 64 && cm->allow_palette_mode) {
uint8_t color_index_map_copy[4096], map[4096];
uint16_t best_runs[PALETTE_MAX_RUNS];
uint8_t best_palette[PALETTE_MAX_SIZE * 3], best_map[4096];
uint8_t best_index[PALETTE_MAX_SIZE], best_literal[PALETTE_MAX_SIZE];
int8_t palette_color_delta[PALETTE_MAX_SIZE];
int64_t local_tx_cache[TX_MODES], sse;
int m1, m2, n, l, best_bits, best_n = 0, best_l = 0;
int r, c, i, j;
int d = get_bit_depth(rows * cols), max_itr = 200;
int palette_run_length_cost[PALETTE_RUN_LENGTHS];
int palette_size_cost[PALETTE_SIZES], indices[4096];
int best_m1 = 0, best_m2 = 0, palette_delta_bitdepth = 0;
double data[3 * 4096], centroids[3 * PALETTE_MAX_SIZE];
double lb = src_y[0], ub = src_y[0];
MB_MODE_INFO *mbmi = &xd->mi[0].src_mi->mbmi;
MB_MODE_INFO mbmi_copy;
RD_COST palette_rd, palette_best_rd;
PALETTE_SCAN_ORDER ps, best_ps = H_SCAN;
#if CONFIG_TX_SKIP
int q_idx = vp9_get_qindex(&cpi->common.seg, mbmi->segment_id,
cpi->common.base_qindex);
int try_tx_skip = q_idx <= TX_SKIP_Q_THRESH_INTRA;
int this_rate_tokenonly_s, s_s;
int tx_skipped = 0, tx_skipped_uv = 0;
int64_t this_distortion_s;
#endif // CONFIG_TX_SKIP
palette_best_rd.rate = INT_MAX;
palette_best_rd.dist = INT64_MAX;
palette_best_rd.rdcost = INT64_MAX;
mbmi_copy = *mbmi;
memcpy(color_index_map_copy, xd->plane[0].color_index_map,
rows * cols * sizeof(color_index_map_copy[0]));
memset(data, 0, sizeof(data[0]) * 3 * 4096);
memset(map, 0, sizeof(map[0]) * 4096);
memset(centroids, 0, sizeof(centroids[0]) * 3 * PALETTE_MAX_SIZE);
vp9_cost_tokens(palette_size_cost,
cpi->common.fc.palette_size_prob[bsize - BLOCK_8X8],
vp9_palette_size_tree);
vp9_cost_tokens(palette_run_length_cost,
cpi->common.fc.palette_run_length_prob[bsize - BLOCK_8X8],
vp9_palette_run_length_tree);
for (i = 0; i < 2; i++)
mbmi->palette_enabled[i] = 1;
#if CONFIG_FILTERINTRA
mbmi->filterbit = 0;
mbmi->uv_filterbit = 0;
#endif // CONFIG_FILTERINTRA
for (r = 0; r < rows; r++) {
for (c = 0; c < cols; c++) {
data[(r * cols + c) * 3] = src_y[r * src_stride_y + c];
data[(r * cols + c) * 3 + 1] = src_u[r * src_stride_uv + c];
data[(r * cols + c) * 3 + 2] = src_v[r * src_stride_uv + c];
}
}
for (n = colors > PALETTE_MAX_SIZE ? PALETTE_MAX_SIZE : colors;
n >= 2; n--) {
for (i = 0; i < n; i++) {
centroids[i * 3] = lb + (2 * i + 1) * (ub - lb) / n / 2;
centroids[i * 3 + 1] = 128;
centroids[i * 3 + 2] = 128;
}
r = k_means(data, centroids, indices, rows * cols, n, 3, max_itr);
for (i = 0; i < 3; i++) {
for (j = 0; j < n; j++)
mbmi->palette_colors[i * PALETTE_MAX_SIZE + j] =
clip_pixel(round(centroids[j * 3 + i]));
}
for (r = 0; r < rows; r++)
for (c = 0; c < cols; c++)
xd->plane[0].color_index_map[r * cols + c] =
indices[r * cols + c];
m1 = 0;
m2 = n;
best_bits = 0;
memcpy(mbmi->palette_literal_colors, mbmi->palette_colors,
m2 * sizeof(mbmi->palette_literal_colors[0]));
#if CONFIG_TX_SKIP
mbmi->tx_skip[0] = 0;
mbmi->tx_skip[1] = 0;
#endif // CONFIG_TX_SKIP
super_block_yrd(cpi, x, &rate_y_tokenonly, &dist_y, &y_skip,
&sse, bsize, local_tx_cache, best_rd);
#if CONFIG_TX_SKIP
if (try_tx_skip) {
if (rate_y_tokenonly != INT_MAX)
rate_y_tokenonly +=
vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0], 0);
mbmi->tx_skip[0] = 1;
super_block_yrd(cpi, x, &this_rate_tokenonly_s, &this_distortion_s,
&s_s, NULL, bsize, local_tx_cache, best_rd);
if (this_rate_tokenonly_s != INT_MAX)
this_rate_tokenonly_s +=
vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0], 1);
if ((this_rate_tokenonly_s != INT_MAX &&
rate_y_tokenonly == INT_MAX) ||
(RDCOST(x->rdmult, x->rddiv, rate_y_tokenonly, dist_y)
> RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly_s,
this_distortion_s))) {
mbmi->tx_skip[0] = 1;
} else {
mbmi->tx_skip[0] = 0;
}
} else {
mbmi->tx_skip[0] = 0;
}
super_block_yrd(cpi, x, &rate_y_tokenonly, &dist_y, &y_skip,
&sse, bsize, local_tx_cache, best_rd);
#endif // CONFIG_TX_SKIP
if (rate_y_tokenonly == INT_MAX)
continue;
super_block_uvrd(cpi, x, &rate_uv_tokenonly, &dist_uv, &uv_skip, &sse,
bsize, best_rd);
#if CONFIG_TX_SKIP
if (try_tx_skip) {
if (rate_uv_tokenonly != INT_MAX)
rate_uv_tokenonly +=
vp9_cost_bit(cpi->common.fc.
uv_tx_skip_prob[mbmi->tx_skip[0]], 0);
mbmi->tx_skip[1] = 1;
super_block_uvrd(cpi, x, &this_rate_tokenonly_s, &this_distortion_s,
&s_s, &sse, bsize, best_rd);
if (this_rate_tokenonly_s != INT_MAX)
this_rate_tokenonly_s +=
vp9_cost_bit(cpi->common.fc.
uv_tx_skip_prob[mbmi->tx_skip[0]], 1);
if ((this_rate_tokenonly_s != INT_MAX &&
rate_uv_tokenonly == INT_MAX) ||
(RDCOST(x->rdmult, x->rddiv, rate_uv_tokenonly, dist_uv)
> RDCOST(x->rdmult, x->rddiv, this_rate_tokenonly_s,
this_distortion_s))) {
mbmi->tx_skip[1] = 1;
} else {
mbmi->tx_skip[1] = 0;
}
} else {
mbmi->tx_skip[1] = 0;
}
super_block_uvrd(cpi, x, &rate_uv_tokenonly, &dist_uv, &uv_skip, &sse,
bsize, best_rd);
#endif // CONFIG_TX_SKIP
if (rate_uv_tokenonly == INT_MAX)
continue;
for (ps = H_SCAN; ps < PALETTE_SCAN_ORDERS; ps++) {
int scan_order[4096];
switch (ps) {
case H_SCAN:
memcpy(map, xd->plane[0].color_index_map, rows * cols);
break;
case V_SCAN:
transpose_block(xd->plane[0].color_index_map,
map, rows, cols);
break;
case SPIN_SCAN:
spin_scan_order(scan_order, rows, cols);
for (i = 0; i < rows * cols; i++)
map[i] = xd->plane[0].color_index_map[scan_order[i]];
break;
case ZZ_SCAN:
zz_scan_order(scan_order, rows, cols);
for (i = 0; i < rows * cols; i++)
map[i] = xd->plane[0].color_index_map[scan_order[i]];
break;
default:
break;
}
l = run_lengh_encoding(map, rows * cols, mbmi->palette_runs,
palette_max_run(bsize));
if (!l) {
continue;
}
rate_y = rate_y_tokenonly +
(1 + PALETTE_DELTA_BIT + get_bit_depth(palette_max_run(bsize)) +
2 + n * m2 + get_bit_depth(n) * (l >> 1)) *
vp9_cost_bit(128, 0) + palette_size_cost[n - 2];
for (i = 0; i < l; i += 2) {
int bits = get_bit_depth(mbmi->palette_runs[i + 1]);
rate_y += palette_run_length_cost[bits > 6 ? 6 : bits - 1];
rate_y += (bits > 6 ? d : bits) * vp9_cost_bit(128, 0);
}
rate_uv = rate_uv_tokenonly +
(1 + 8 * 2 * n) * vp9_cost_bit(128, 0);
#if CONFIG_TX_SKIP
rate_y += vp9_cost_bit(cpi->common.fc.y_tx_skip_prob[0],
mbmi->tx_skip[0]);
rate_uv +=
vp9_cost_bit(cpi->common.fc.uv_tx_skip_prob[mbmi->tx_skip[0]],
mbmi->tx_skip[1]);
#endif // CONFIG_TX_SKIP
if (y_skip && uv_skip) {
palette_rd.rate = rate_y + rate_uv - rate_y_tokenonly -
rate_uv_tokenonly + vp9_cost_bit(vp9_get_skip_prob(cm, xd), 1);
palette_rd.dist = dist_y + dist_uv;
} else {
palette_rd.rate = rate_y + rate_uv +
vp9_cost_bit(vp9_get_skip_prob(cm, xd), 0);
palette_rd.dist = dist_y + dist_uv;
}
palette_rd.rdcost =
RDCOST(x->rdmult, x->rddiv, palette_rd.rate, palette_rd.dist);
if (palette_rd.rdcost < palette_best_rd.rdcost) {
palette_best_rd = palette_rd;
best_n = n;
best_l = l;
best_ps = ps;
best_m1 = m1;
best_m2 = m2;
palette_delta_bitdepth = best_bits;
#if CONFIG_TX_SKIP
tx_skipped = mbmi->tx_skip[0];
tx_skipped_uv = mbmi->tx_skip[1];
#endif // CONFIG_TX_SKIP
memcpy(best_palette, mbmi->palette_colors,
PALETTE_MAX_SIZE * 3 * sizeof(best_palette[0]));
memcpy(best_runs, mbmi->palette_runs, l * sizeof(best_runs[0]));
memcpy(best_map, xd->plane[0].color_index_map,
rows * cols * sizeof(best_map[0]));
memcpy(best_index, mbmi->palette_indexed_colors,
best_m1 * sizeof(best_index[0]));
memcpy(palette_color_delta, mbmi->palette_color_delta,
best_m1 * sizeof(palette_color_delta[0]));
memcpy(best_literal, mbmi->palette_literal_colors,
best_m2 * sizeof(best_literal[0]));
}
}
}
if (palette_best_rd.rdcost < rd_cost->rdcost) {
*rd_cost = palette_best_rd;
mbmi->mode = DC_PRED;
mbmi->uv_mode = DC_PRED;
for (i = 0; i < 2; i++)
mbmi->palette_enabled[i] = 1;
mbmi->palette_size[0] = best_n;
mbmi->palette_size[1] = best_n;
mbmi->palette_run_length[0] = best_l;
mbmi->palette_scan_order[0] = best_ps;
mbmi->palette_indexed_size = best_m1;
mbmi->palette_literal_size = best_m2;
mbmi->palette_delta_bitdepth = palette_delta_bitdepth;
memcpy(mbmi->palette_colors, best_palette,
PALETTE_MAX_SIZE * 3 * sizeof(best_palette[0]));
memcpy(mbmi->palette_runs, best_runs, best_l * sizeof(best_runs[0]));
memcpy(xd->plane[0].color_index_map, best_map,
rows * cols * sizeof(best_map[0]));
memcpy(mbmi->palette_indexed_colors, best_index,
best_m1 * sizeof(best_index[0]));
memcpy(mbmi->palette_color_delta, palette_color_delta,
best_m1 * sizeof(palette_color_delta[0]));
memcpy(mbmi->palette_literal_colors, best_literal,
best_m2 * sizeof(best_literal[0]));
#if CONFIG_FILTERINTRA
mbmi->filterbit = 0;
mbmi->uv_filterbit = 0;
#endif // CONFIG_FILTERINTRA
#if CONFIG_TX_SKIP
mbmi->tx_skip[0] = tx_skipped;
mbmi->tx_skip[1] = tx_skipped_uv;
#endif // CONFIG_TX_SKIP
} else {
*mbmi = mbmi_copy;
memcpy(xd->plane[0].color_index_map, color_index_map_copy,
rows * cols * sizeof(color_index_map_copy[0]));
}
ctx->mic = *xd->mi[0].src_mi;
}
}
if (xd->mi[0].src_mi->mbmi.palette_enabled[0]) {
palette_color_insertion(ctx->palette_colors_buf,
&ctx->palette_buf_size,
ctx->palette_count_buf,
&(xd->mi[0].src_mi->mbmi));
}
#endif // CONFIG_PALETTE
}
static void update_rd_thresh_fact(VP9_COMP *cpi, int bsize,
@ -4849,6 +5377,12 @@ void vp9_rd_pick_inter_mode_sb(VP9_COMP *cpi, MACROBLOCK *x,
midx = end_pos;
}
#if CONFIG_PALETTE
for (i = 0; i < 2; ++i) {
mbmi->palette_enabled[i] = 0;
}
#endif // CONFIG_PALETTE
for (midx = 0; midx < MAX_MODES; ++midx) {
int mode_index = mode_map[midx];
int mode_excluded = 0;
@ -5090,9 +5624,6 @@ void vp9_rd_pick_inter_mode_sb(VP9_COMP *cpi, MACROBLOCK *x,
#if CONFIG_COPY_MODE
mbmi->copy_mode = NOREF;
#endif // CONFIG_COPY_MODE
#if CONFIG_PALETTE
mbmi->palette_enabled = 0;
#endif // CONFIG_PALETTE
// Evaluate all sub-pel filters irrespective of whether we can use
// them for this frame.
mbmi->interp_filter = cm->interp_filter == SWITCHABLE ? EIGHTTAP
@ -5992,7 +6523,11 @@ void vp9_rd_pick_inter_mode_sub8x8(VP9_COMP *cpi, MACROBLOCK *x,
#if CONFIG_SUPERTX
*returnrate_nocoef = INT_MAX;
#endif
#if CONFIG_PALETTE
for (i = 0; i < 2; ++i) {
mbmi->palette_enabled[i] = 0;
}
#endif // CONFIG_PALETTE
for (ref_frame = LAST_FRAME; ref_frame <= ALTREF_FRAME; ref_frame++) {
if (cpi->ref_frame_flags & flag_list[ref_frame]) {
setup_buffer_inter(cpi, x, tile,