vpx/vp9/common/vp9_reconintra.c

320 lines
11 KiB
C
Raw Normal View History

2010-05-18 17:58:33 +02:00
/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
2010-05-18 17:58:33 +02:00
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
2010-05-18 17:58:33 +02:00
*/
#include <stdio.h>
#include "./vpx_config.h"
#include "vp9_rtcd.h"
#include "vp9/common/vp9_reconintra.h"
#include "vp9/common/vp9_onyxc_int.h"
2010-05-18 17:58:33 +02:00
#include "vpx_mem/vpx_mem.h"
static void d27_predictor(uint8_t *ypred_ptr, int y_stride,
int bw, int bh,
uint8_t *yabove_row, uint8_t *yleft_col) {
int r, c;
// first column
for (r = 0; r < bh - 1; ++r) {
ypred_ptr[r * y_stride] = ROUND_POWER_OF_TWO(yleft_col[r] +
yleft_col[r + 1], 1);
}
ypred_ptr[(bh - 1) * y_stride] = yleft_col[bh-1];
ypred_ptr++;
// second column
for (r = 0; r < bh - 2; ++r) {
ypred_ptr[r * y_stride] = ROUND_POWER_OF_TWO(yleft_col[r] +
yleft_col[r + 1] * 2 +
yleft_col[r + 2], 2);
}
ypred_ptr[(bh - 2) * y_stride] = ROUND_POWER_OF_TWO(yleft_col[bh - 2] +
yleft_col[bh - 1] * 3,
2);
ypred_ptr[(bh - 1) * y_stride] = yleft_col[bh-1];
ypred_ptr++;
// rest of last row
for (c = 0; c < bw - 2; ++c) {
ypred_ptr[(bh - 1) * y_stride + c] = yleft_col[bh-1];
}
for (r = bh - 2; r >= 0; --r) {
for (c = 0; c < bw - 2; ++c) {
ypred_ptr[r * y_stride + c] = ypred_ptr[(r + 1) * y_stride + c - 2];
}
}
}
static void d63_predictor(uint8_t *ypred_ptr, int y_stride,
int bw, int bh,
uint8_t *yabove_row, uint8_t *yleft_col) {
int r, c;
for (r = 0; r < bh; ++r) {
for (c = 0; c < bw; ++c) {
if (r & 1) {
ypred_ptr[c] = ROUND_POWER_OF_TWO(yabove_row[r/2 + c] +
yabove_row[r/2 + c + 1] * 2 +
yabove_row[r/2 + c + 2], 2);
} else {
ypred_ptr[c] =ROUND_POWER_OF_TWO(yabove_row[r/2 + c] +
yabove_row[r/2+ c + 1], 1);
}
}
ypred_ptr += y_stride;
}
}
static void d45_predictor(uint8_t *ypred_ptr, int y_stride,
int bw, int bh,
uint8_t *yabove_row, uint8_t *yleft_col) {
int r, c;
for (r = 0; r < bh; ++r) {
for (c = 0; c < bw; ++c) {
if (r + c + 2 < bw * 2)
ypred_ptr[c] = ROUND_POWER_OF_TWO(yabove_row[r + c] +
yabove_row[r + c + 1] * 2 +
yabove_row[r + c + 2], 2);
else
ypred_ptr[c] = yabove_row[bw * 2 - 1];
}
ypred_ptr += y_stride;
}
}
static void d117_predictor(uint8_t *ypred_ptr, int y_stride,
int bw, int bh,
uint8_t *yabove_row, uint8_t *yleft_col) {
int r, c;
// first row
for (c = 0; c < bw; c++)
ypred_ptr[c] = ROUND_POWER_OF_TWO(yabove_row[c - 1] + yabove_row[c], 1);
ypred_ptr += y_stride;
// second row
ypred_ptr[0] = ROUND_POWER_OF_TWO(yleft_col[0] +
yabove_row[-1] * 2 +
yabove_row[0], 2);
for (c = 1; c < bw; c++)
ypred_ptr[c] = ROUND_POWER_OF_TWO(yabove_row[c - 2] +
yabove_row[c - 1] * 2 +
yabove_row[c], 2);
ypred_ptr += y_stride;
// the rest of first col
ypred_ptr[0] = ROUND_POWER_OF_TWO(yabove_row[-1] +
yleft_col[0] * 2 +
yleft_col[1], 2);
for (r = 3; r < bh; ++r)
ypred_ptr[(r-2) * y_stride] = ROUND_POWER_OF_TWO(yleft_col[r - 3] +
yleft_col[r - 2] * 2 +
yleft_col[r - 1], 2);
// the rest of the block
for (r = 2; r < bh; ++r) {
for (c = 1; c < bw; c++)
ypred_ptr[c] = ypred_ptr[-2 * y_stride + c - 1];
ypred_ptr += y_stride;
}
}
static void d135_predictor(uint8_t *ypred_ptr, int y_stride,
int bw, int bh,
uint8_t *yabove_row, uint8_t *yleft_col) {
int r, c;
ypred_ptr[0] = ROUND_POWER_OF_TWO(yleft_col[0] +
yabove_row[-1] * 2 +
yabove_row[0], 2);
for (c = 1; c < bw; c++)
ypred_ptr[c] = ROUND_POWER_OF_TWO(yabove_row[c - 2] +
yabove_row[c - 1] * 2 +
yabove_row[c], 2);
ypred_ptr[y_stride] = ROUND_POWER_OF_TWO(yabove_row[-1] +
yleft_col[0] * 2 +
yleft_col[1], 2);
for (r = 2; r < bh; ++r)
ypred_ptr[r * y_stride] = ROUND_POWER_OF_TWO(yleft_col[r - 2] +
yleft_col[r - 1] * 2 +
yleft_col[r], 2);
ypred_ptr += y_stride;
for (r = 1; r < bh; ++r) {
for (c = 1; c < bw; c++)
ypred_ptr[c] = ypred_ptr[-y_stride + c - 1];
ypred_ptr += y_stride;
}
}
static void d153_predictor(uint8_t *ypred_ptr,
int y_stride,
int bw, int bh,
uint8_t *yabove_row,
uint8_t *yleft_col) {
int r, c;
ypred_ptr[0] = ROUND_POWER_OF_TWO(yabove_row[-1] + yleft_col[0], 1);
for (r = 1; r < bh; r++)
ypred_ptr[r * y_stride] =
ROUND_POWER_OF_TWO(yleft_col[r - 1] + yleft_col[r], 1);
ypred_ptr++;
ypred_ptr[0] = ROUND_POWER_OF_TWO(yleft_col[0] +
yabove_row[-1] * 2 +
yabove_row[0], 2);
ypred_ptr[y_stride] = ROUND_POWER_OF_TWO(yabove_row[-1] +
yleft_col[0] * 2 +
yleft_col[1], 2);
for (r = 2; r < bh; r++)
ypred_ptr[r * y_stride] = ROUND_POWER_OF_TWO(yleft_col[r - 2] +
yleft_col[r - 1] * 2 +
yleft_col[r], 2);
ypred_ptr++;
for (c = 0; c < bw - 2; c++)
ypred_ptr[c] = ROUND_POWER_OF_TWO(yabove_row[c - 1] +
yabove_row[c] * 2 +
yabove_row[c + 1], 2);
ypred_ptr += y_stride;
for (r = 1; r < bh; ++r) {
for (c = 0; c < bw - 2; c++)
ypred_ptr[c] = ypred_ptr[-y_stride + c - 2];
ypred_ptr += y_stride;
}
}
void vp9_build_intra_predictors(uint8_t *src, int src_stride,
uint8_t *ypred_ptr,
int y_stride, int mode,
int bw, int bh,
int up_available, int left_available,
int right_available) {
int r, c, i;
uint8_t yleft_col[64], yabove_data[129], ytop_left;
[WIP] Add column-based tiling. This patch adds column-based tiling. The idea is to make each tile independently decodable (after reading the common frame header) and also independendly encodable (minus within-frame cost adjustments in the RD loop) to speed-up hardware & software en/decoders if they used multi-threading. Column-based tiling has the added advantage (over other tiling methods) that it minimizes realtime use-case latency, since all threads can start encoding data as soon as the first SB-row worth of data is available to the encoder. There is some test code that does random tile ordering in the decoder, to confirm that each tile is indeed independently decodable from other tiles in the same frame. At tile edges, all contexts assume default values (i.e. 0, 0 motion vector, no coefficients, DC intra4x4 mode), and motion vector search and ordering do not cross tiles in the same frame. t log Tile independence is not maintained between frames ATM, i.e. tile 0 of frame 1 is free to use motion vectors that point into any tile of frame 0. We support 1 (i.e. no tiling), 2 or 4 column-tiles. The loopfilter crosses tile boundaries. I discussed this briefly with Aki and he says that's OK. An in-loop loopfilter would need to do some sync between tile threads, but that shouldn't be a big issue. Resuls: with tiling disabled, we go up slightly because of improved edge use in the intra4x4 prediction. With 2 tiles, we lose about ~1% on derf, ~0.35% on HD and ~0.55% on STD/HD. With 4 tiles, we lose another ~1.5% on derf ~0.77% on HD and ~0.85% on STD/HD. Most of this loss is concentrated in the low-bitrate end of clips, and most of it is because of the loss of edges at tile boundaries and the resulting loss of intra predictors. TODO: - more tiles (perhaps allow row-based tiling also, and max. 8 tiles)? - maybe optionally (for EC purposes), motion vectors themselves should not cross tile edges, or we should emulate such borders as if they were off-frame, to limit error propagation to within one tile only. This doesn't have to be the default behaviour but could be an optional bitstream flag. Change-Id: I5951c3a0742a767b20bc9fb5af685d9892c2c96f
2013-02-01 18:35:28 +01:00
uint8_t *yabove_row = yabove_data + 1;
// 127 127 127 .. 127 127 127 127 127 127
// 129 A B .. Y Z
// 129 C D .. W X
// 129 E F .. U V
// 129 G H .. S T T T T T
// ..
assert(bw == bh);
[WIP] Add column-based tiling. This patch adds column-based tiling. The idea is to make each tile independently decodable (after reading the common frame header) and also independendly encodable (minus within-frame cost adjustments in the RD loop) to speed-up hardware & software en/decoders if they used multi-threading. Column-based tiling has the added advantage (over other tiling methods) that it minimizes realtime use-case latency, since all threads can start encoding data as soon as the first SB-row worth of data is available to the encoder. There is some test code that does random tile ordering in the decoder, to confirm that each tile is indeed independently decodable from other tiles in the same frame. At tile edges, all contexts assume default values (i.e. 0, 0 motion vector, no coefficients, DC intra4x4 mode), and motion vector search and ordering do not cross tiles in the same frame. t log Tile independence is not maintained between frames ATM, i.e. tile 0 of frame 1 is free to use motion vectors that point into any tile of frame 0. We support 1 (i.e. no tiling), 2 or 4 column-tiles. The loopfilter crosses tile boundaries. I discussed this briefly with Aki and he says that's OK. An in-loop loopfilter would need to do some sync between tile threads, but that shouldn't be a big issue. Resuls: with tiling disabled, we go up slightly because of improved edge use in the intra4x4 prediction. With 2 tiles, we lose about ~1% on derf, ~0.35% on HD and ~0.55% on STD/HD. With 4 tiles, we lose another ~1.5% on derf ~0.77% on HD and ~0.85% on STD/HD. Most of this loss is concentrated in the low-bitrate end of clips, and most of it is because of the loss of edges at tile boundaries and the resulting loss of intra predictors. TODO: - more tiles (perhaps allow row-based tiling also, and max. 8 tiles)? - maybe optionally (for EC purposes), motion vectors themselves should not cross tile edges, or we should emulate such borders as if they were off-frame, to limit error propagation to within one tile only. This doesn't have to be the default behaviour but could be an optional bitstream flag. Change-Id: I5951c3a0742a767b20bc9fb5af685d9892c2c96f
2013-02-01 18:35:28 +01:00
if (left_available) {
for (i = 0; i < bh; i++)
[WIP] Add column-based tiling. This patch adds column-based tiling. The idea is to make each tile independently decodable (after reading the common frame header) and also independendly encodable (minus within-frame cost adjustments in the RD loop) to speed-up hardware & software en/decoders if they used multi-threading. Column-based tiling has the added advantage (over other tiling methods) that it minimizes realtime use-case latency, since all threads can start encoding data as soon as the first SB-row worth of data is available to the encoder. There is some test code that does random tile ordering in the decoder, to confirm that each tile is indeed independently decodable from other tiles in the same frame. At tile edges, all contexts assume default values (i.e. 0, 0 motion vector, no coefficients, DC intra4x4 mode), and motion vector search and ordering do not cross tiles in the same frame. t log Tile independence is not maintained between frames ATM, i.e. tile 0 of frame 1 is free to use motion vectors that point into any tile of frame 0. We support 1 (i.e. no tiling), 2 or 4 column-tiles. The loopfilter crosses tile boundaries. I discussed this briefly with Aki and he says that's OK. An in-loop loopfilter would need to do some sync between tile threads, but that shouldn't be a big issue. Resuls: with tiling disabled, we go up slightly because of improved edge use in the intra4x4 prediction. With 2 tiles, we lose about ~1% on derf, ~0.35% on HD and ~0.55% on STD/HD. With 4 tiles, we lose another ~1.5% on derf ~0.77% on HD and ~0.85% on STD/HD. Most of this loss is concentrated in the low-bitrate end of clips, and most of it is because of the loss of edges at tile boundaries and the resulting loss of intra predictors. TODO: - more tiles (perhaps allow row-based tiling also, and max. 8 tiles)? - maybe optionally (for EC purposes), motion vectors themselves should not cross tile edges, or we should emulate such borders as if they were off-frame, to limit error propagation to within one tile only. This doesn't have to be the default behaviour but could be an optional bitstream flag. Change-Id: I5951c3a0742a767b20bc9fb5af685d9892c2c96f
2013-02-01 18:35:28 +01:00
yleft_col[i] = src[i * src_stride - 1];
} else {
vpx_memset(yleft_col, 129, bh);
[WIP] Add column-based tiling. This patch adds column-based tiling. The idea is to make each tile independently decodable (after reading the common frame header) and also independendly encodable (minus within-frame cost adjustments in the RD loop) to speed-up hardware & software en/decoders if they used multi-threading. Column-based tiling has the added advantage (over other tiling methods) that it minimizes realtime use-case latency, since all threads can start encoding data as soon as the first SB-row worth of data is available to the encoder. There is some test code that does random tile ordering in the decoder, to confirm that each tile is indeed independently decodable from other tiles in the same frame. At tile edges, all contexts assume default values (i.e. 0, 0 motion vector, no coefficients, DC intra4x4 mode), and motion vector search and ordering do not cross tiles in the same frame. t log Tile independence is not maintained between frames ATM, i.e. tile 0 of frame 1 is free to use motion vectors that point into any tile of frame 0. We support 1 (i.e. no tiling), 2 or 4 column-tiles. The loopfilter crosses tile boundaries. I discussed this briefly with Aki and he says that's OK. An in-loop loopfilter would need to do some sync between tile threads, but that shouldn't be a big issue. Resuls: with tiling disabled, we go up slightly because of improved edge use in the intra4x4 prediction. With 2 tiles, we lose about ~1% on derf, ~0.35% on HD and ~0.55% on STD/HD. With 4 tiles, we lose another ~1.5% on derf ~0.77% on HD and ~0.85% on STD/HD. Most of this loss is concentrated in the low-bitrate end of clips, and most of it is because of the loss of edges at tile boundaries and the resulting loss of intra predictors. TODO: - more tiles (perhaps allow row-based tiling also, and max. 8 tiles)? - maybe optionally (for EC purposes), motion vectors themselves should not cross tile edges, or we should emulate such borders as if they were off-frame, to limit error propagation to within one tile only. This doesn't have to be the default behaviour but could be an optional bitstream flag. Change-Id: I5951c3a0742a767b20bc9fb5af685d9892c2c96f
2013-02-01 18:35:28 +01:00
}
if (up_available) {
uint8_t *yabove_ptr = src - src_stride;
vpx_memcpy(yabove_row, yabove_ptr, bw);
if (bw == 4 && right_available)
vpx_memcpy(yabove_row + bw, yabove_ptr + bw, bw);
else
vpx_memset(yabove_row + bw, yabove_row[bw -1], bw);
ytop_left = left_available ? yabove_ptr[-1] : 129;
[WIP] Add column-based tiling. This patch adds column-based tiling. The idea is to make each tile independently decodable (after reading the common frame header) and also independendly encodable (minus within-frame cost adjustments in the RD loop) to speed-up hardware & software en/decoders if they used multi-threading. Column-based tiling has the added advantage (over other tiling methods) that it minimizes realtime use-case latency, since all threads can start encoding data as soon as the first SB-row worth of data is available to the encoder. There is some test code that does random tile ordering in the decoder, to confirm that each tile is indeed independently decodable from other tiles in the same frame. At tile edges, all contexts assume default values (i.e. 0, 0 motion vector, no coefficients, DC intra4x4 mode), and motion vector search and ordering do not cross tiles in the same frame. t log Tile independence is not maintained between frames ATM, i.e. tile 0 of frame 1 is free to use motion vectors that point into any tile of frame 0. We support 1 (i.e. no tiling), 2 or 4 column-tiles. The loopfilter crosses tile boundaries. I discussed this briefly with Aki and he says that's OK. An in-loop loopfilter would need to do some sync between tile threads, but that shouldn't be a big issue. Resuls: with tiling disabled, we go up slightly because of improved edge use in the intra4x4 prediction. With 2 tiles, we lose about ~1% on derf, ~0.35% on HD and ~0.55% on STD/HD. With 4 tiles, we lose another ~1.5% on derf ~0.77% on HD and ~0.85% on STD/HD. Most of this loss is concentrated in the low-bitrate end of clips, and most of it is because of the loss of edges at tile boundaries and the resulting loss of intra predictors. TODO: - more tiles (perhaps allow row-based tiling also, and max. 8 tiles)? - maybe optionally (for EC purposes), motion vectors themselves should not cross tile edges, or we should emulate such borders as if they were off-frame, to limit error propagation to within one tile only. This doesn't have to be the default behaviour but could be an optional bitstream flag. Change-Id: I5951c3a0742a767b20bc9fb5af685d9892c2c96f
2013-02-01 18:35:28 +01:00
} else {
vpx_memset(yabove_row, 127, bw * 2);
[WIP] Add column-based tiling. This patch adds column-based tiling. The idea is to make each tile independently decodable (after reading the common frame header) and also independendly encodable (minus within-frame cost adjustments in the RD loop) to speed-up hardware & software en/decoders if they used multi-threading. Column-based tiling has the added advantage (over other tiling methods) that it minimizes realtime use-case latency, since all threads can start encoding data as soon as the first SB-row worth of data is available to the encoder. There is some test code that does random tile ordering in the decoder, to confirm that each tile is indeed independently decodable from other tiles in the same frame. At tile edges, all contexts assume default values (i.e. 0, 0 motion vector, no coefficients, DC intra4x4 mode), and motion vector search and ordering do not cross tiles in the same frame. t log Tile independence is not maintained between frames ATM, i.e. tile 0 of frame 1 is free to use motion vectors that point into any tile of frame 0. We support 1 (i.e. no tiling), 2 or 4 column-tiles. The loopfilter crosses tile boundaries. I discussed this briefly with Aki and he says that's OK. An in-loop loopfilter would need to do some sync between tile threads, but that shouldn't be a big issue. Resuls: with tiling disabled, we go up slightly because of improved edge use in the intra4x4 prediction. With 2 tiles, we lose about ~1% on derf, ~0.35% on HD and ~0.55% on STD/HD. With 4 tiles, we lose another ~1.5% on derf ~0.77% on HD and ~0.85% on STD/HD. Most of this loss is concentrated in the low-bitrate end of clips, and most of it is because of the loss of edges at tile boundaries and the resulting loss of intra predictors. TODO: - more tiles (perhaps allow row-based tiling also, and max. 8 tiles)? - maybe optionally (for EC purposes), motion vectors themselves should not cross tile edges, or we should emulate such borders as if they were off-frame, to limit error propagation to within one tile only. This doesn't have to be the default behaviour but could be an optional bitstream flag. Change-Id: I5951c3a0742a767b20bc9fb5af685d9892c2c96f
2013-02-01 18:35:28 +01:00
ytop_left = 127;
}
[WIP] Add column-based tiling. This patch adds column-based tiling. The idea is to make each tile independently decodable (after reading the common frame header) and also independendly encodable (minus within-frame cost adjustments in the RD loop) to speed-up hardware & software en/decoders if they used multi-threading. Column-based tiling has the added advantage (over other tiling methods) that it minimizes realtime use-case latency, since all threads can start encoding data as soon as the first SB-row worth of data is available to the encoder. There is some test code that does random tile ordering in the decoder, to confirm that each tile is indeed independently decodable from other tiles in the same frame. At tile edges, all contexts assume default values (i.e. 0, 0 motion vector, no coefficients, DC intra4x4 mode), and motion vector search and ordering do not cross tiles in the same frame. t log Tile independence is not maintained between frames ATM, i.e. tile 0 of frame 1 is free to use motion vectors that point into any tile of frame 0. We support 1 (i.e. no tiling), 2 or 4 column-tiles. The loopfilter crosses tile boundaries. I discussed this briefly with Aki and he says that's OK. An in-loop loopfilter would need to do some sync between tile threads, but that shouldn't be a big issue. Resuls: with tiling disabled, we go up slightly because of improved edge use in the intra4x4 prediction. With 2 tiles, we lose about ~1% on derf, ~0.35% on HD and ~0.55% on STD/HD. With 4 tiles, we lose another ~1.5% on derf ~0.77% on HD and ~0.85% on STD/HD. Most of this loss is concentrated in the low-bitrate end of clips, and most of it is because of the loss of edges at tile boundaries and the resulting loss of intra predictors. TODO: - more tiles (perhaps allow row-based tiling also, and max. 8 tiles)? - maybe optionally (for EC purposes), motion vectors themselves should not cross tile edges, or we should emulate such borders as if they were off-frame, to limit error propagation to within one tile only. This doesn't have to be the default behaviour but could be an optional bitstream flag. Change-Id: I5951c3a0742a767b20bc9fb5af685d9892c2c96f
2013-02-01 18:35:28 +01:00
yabove_row[-1] = ytop_left;
switch (mode) {
case DC_PRED: {
int i;
int expected_dc = 128;
int average = 0;
int count = 0;
if (up_available || left_available) {
if (up_available) {
for (i = 0; i < bw; i++)
average += yabove_row[i];
count += bw;
2010-05-18 17:58:33 +02:00
}
if (left_available) {
for (i = 0; i < bh; i++)
average += yleft_col[i];
count += bh;
2010-05-18 17:58:33 +02:00
}
expected_dc = (average + (count >> 1)) / count;
}
for (r = 0; r < bh; r++) {
vpx_memset(ypred_ptr, expected_dc, bw);
ypred_ptr += y_stride;
}
2010-05-18 17:58:33 +02:00
}
break;
case V_PRED:
for (r = 0; r < bh; r++) {
vpx_memcpy(ypred_ptr, yabove_row, bw);
ypred_ptr += y_stride;
}
break;
case H_PRED:
for (r = 0; r < bh; r++) {
vpx_memset(ypred_ptr, yleft_col[r], bw);
ypred_ptr += y_stride;
}
break;
case TM_PRED:
for (r = 0; r < bh; r++) {
for (c = 0; c < bw; c++)
ypred_ptr[c] = clip_pixel(yleft_col[r] + yabove_row[c] - ytop_left);
ypred_ptr += y_stride;
}
break;
case D45_PRED:
d45_predictor(ypred_ptr, y_stride, bw, bh, yabove_row, yleft_col);
break;
case D135_PRED:
d135_predictor(ypred_ptr, y_stride, bw, bh, yabove_row, yleft_col);
break;
case D117_PRED:
d117_predictor(ypred_ptr, y_stride, bw, bh, yabove_row, yleft_col);
break;
case D153_PRED:
d153_predictor(ypred_ptr, y_stride, bw, bh, yabove_row, yleft_col);
break;
case D27_PRED:
d27_predictor(ypred_ptr, y_stride, bw, bh, yabove_row, yleft_col);
break;
case D63_PRED:
d63_predictor(ypred_ptr, y_stride, bw, bh, yabove_row, yleft_col);
break;
default:
break;
}
2010-05-18 17:58:33 +02:00
}
void vp9_predict_intra_block(MACROBLOCKD *xd,
int block_idx,
int bwl_in,
TX_SIZE tx_size,
int mode,
uint8_t *predictor, int pre_stride) {
const int bwl = bwl_in - tx_size;
const int wmask = (1 << bwl) - 1;
const int have_top = (block_idx >> bwl) || xd->up_available;
const int have_left = (block_idx & wmask) || xd->left_available;
const int have_right = ((block_idx & wmask) != wmask);
const int txfm_block_size = 4 << tx_size;
assert(bwl >= 0);
vp9_build_intra_predictors(predictor, pre_stride,
predictor, pre_stride,
mode,
txfm_block_size,
txfm_block_size,
have_top, have_left,
have_right);
}