ref: 3d7920e6ff7c8e8f032a8e685595bcd38e4a3028
dir: /src/cdef.c/
/*
* Copyright © 2018, VideoLAN and dav1d authors
* Copyright © 2018, Two Orioles, LLC
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice, this
* list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright notice,
* this list of conditions and the following disclaimer in the documentation
* and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
* WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
* DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
* ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
* SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/*
* Copyright (c) 2001-2016, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. 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 www.aomedia.org/license/patent.
*/
#include "config.h"
#include <assert.h>
#include <stdlib.h>
#include "common/intops.h"
#include "src/cdef.h"
static const int8_t cdef_directions4[8 /* dir */][2 /* pass */] = {
{ -1 * 8 + 1, -2 * 8 + 2 },
{ 0 * 8 + 1, -1 * 8 + 2 },
{ 0 * 8 + 1, 0 * 8 + 2 },
{ 0 * 8 + 1, 1 * 8 + 2 },
{ 1 * 8 + 1, 2 * 8 + 2 },
{ 1 * 8 + 0, 2 * 8 + 1 },
{ 1 * 8 + 0, 2 * 8 + 0 },
{ 1 * 8 + 0, 2 * 8 - 1 }
};
static const int8_t cdef_directions8[8 /* dir */][2 /* pass */] = {
{ -1 * 16 + 1, -2 * 16 + 2 },
{ 0 * 16 + 1, -1 * 16 + 2 },
{ 0 * 16 + 1, 0 * 16 + 2 },
{ 0 * 16 + 1, 1 * 16 + 2 },
{ 1 * 16 + 1, 2 * 16 + 2 },
{ 1 * 16 + 0, 2 * 16 + 1 },
{ 1 * 16 + 0, 2 * 16 + 0 },
{ 1 * 16 + 0, 2 * 16 - 1 }
};
static const uint8_t cdef_pri_taps[2][2] = { { 4, 2 }, { 3, 3 } };
static const uint8_t cdef_sec_taps[2][2] = { { 2, 1 }, { 2, 1 } };
static inline int constrain(const int diff, const int threshold,
const int damping)
{
if (!threshold) return 0;
const int shift = imax(0, damping - ulog2(threshold));
return apply_sign(imin(abs(diff), imax(0, threshold - (abs(diff) >> shift))),
diff);
}
/*
* <code partially copied from libaom>
*/
#define CDEF_VERY_LARGE (30000)
static void fill(uint16_t *tmp, const ptrdiff_t stride,
const int w, const int h)
{
for (int y = 0; y < h; y++) {
for (int x = 0; x < w; x++)
tmp[x] = CDEF_VERY_LARGE;
tmp += stride;
}
}
/* Smooth in the direction detected. */
static void cdef_filter_block_c(pixel *const dst, const ptrdiff_t dst_stride,
/*const*/ pixel *const top[2],
const int w, const int h, const int pri_strength,
const int sec_strength, const int dir,
const int damping, const enum CdefEdgeFlags edges)
{
const ptrdiff_t tmp_stride = 16 >> (w == 4);
assert((w == 4 || w == 8) && (h == 4 || h == 8));
uint16_t tmp[192]; // 16*12 is the maximum value of tmp_stride * (h + 4)
uint16_t *tmp2 = tmp + 2 * tmp_stride + 2;
const uint8_t *const pri_taps = cdef_pri_taps[(pri_strength >> (BITDEPTH - 8)) & 1];
const uint8_t *const sec_taps = cdef_sec_taps[(pri_strength >> (BITDEPTH - 8)) & 1];
const int8_t (*cdef_directions)[2];
assert(w == 4 || w == 8);
cdef_directions = w == 4 ? cdef_directions4 : cdef_directions8;
// fill extended input buffer
int x_start = -2, x_end = w + 2, y_start = -2, y_end = h + 2;
if (!(edges & HAVE_TOP)) {
fill(tmp, tmp_stride, w + 4, 2);
y_start = 0;
}
if (!(edges & HAVE_BOTTOM)) {
fill(tmp + (h + 2) * tmp_stride, tmp_stride, w + 4, 2);
y_end -= 2;
}
if (!(edges & HAVE_LEFT)) {
fill(tmp + (2 + y_start) * tmp_stride, tmp_stride, 2, y_end - y_start);
x_start = 0;
}
if (!(edges & HAVE_RIGHT)) {
fill(tmp + (2 + y_start) * tmp_stride + w + 2, tmp_stride,
2, y_end - y_start);
x_end -= 2;
}
for (int y = y_start; y < 0; y++)
for (int x = x_start; x < x_end; x++)
tmp2[y * tmp_stride + x] = top[y & 1][x];
for (int y = 0; y < y_end; y++)
for (int x = x_start; x < x_end; x++)
tmp2[y * tmp_stride + x] = dst[y * PXSTRIDE(dst_stride) + x];
// run actual filter
for (int y = 0; y < h; y++) {
for (int x = 0; x < w; x++) {
int sum = 0;
const int px = dst[y * PXSTRIDE(dst_stride) + x];
int max = px, min = px;
for (int k = 0; k < 2; k++) {
const int8_t off1 = cdef_directions[dir][k];
const int p0 = tmp2[y * tmp_stride + x + off1];
const int p1 = tmp2[y * tmp_stride + x - off1];
sum += pri_taps[k] * constrain(p0 - px, pri_strength, damping);
sum += pri_taps[k] * constrain(p1 - px, pri_strength, damping);
if (p0 != CDEF_VERY_LARGE) max = imax(p0, max);
if (p1 != CDEF_VERY_LARGE) max = imax(p1, max);
min = imin(p0, min);
min = imin(p1, min);
const int8_t off2 = cdef_directions[(dir + 2) & 7][k];
const int s0 = tmp2[y * tmp_stride + x + off2];
const int s1 = tmp2[y * tmp_stride + x - off2];
const int8_t off3 = cdef_directions[(dir + 6) & 7][k];
const int s2 = tmp2[y * tmp_stride + x + off3];
const int s3 = tmp2[y * tmp_stride + x - off3];
if (s0 != CDEF_VERY_LARGE) max = imax(s0, max);
if (s1 != CDEF_VERY_LARGE) max = imax(s1, max);
if (s2 != CDEF_VERY_LARGE) max = imax(s2, max);
if (s3 != CDEF_VERY_LARGE) max = imax(s3, max);
min = imin(s0, min);
min = imin(s1, min);
min = imin(s2, min);
min = imin(s3, min);
sum += sec_taps[k] * constrain(s0 - px, sec_strength, damping);
sum += sec_taps[k] * constrain(s1 - px, sec_strength, damping);
sum += sec_taps[k] * constrain(s2 - px, sec_strength, damping);
sum += sec_taps[k] * constrain(s3 - px, sec_strength, damping);
}
dst[y * PXSTRIDE(dst_stride) + x] =
iclip(px + ((8 + sum - (sum < 0)) >> 4), min, max);
}
}
}
/*
* </code partially copied from libaom>
*/
#define cdef_fn(w, h) \
static void cdef_filter_block_##w##x##h##_c(pixel *const dst, \
const ptrdiff_t stride, \
/*const*/ pixel *const top[2], \
const int pri_strength, \
const int sec_strength, \
const int dir, \
const int damping, \
const enum CdefEdgeFlags edges) \
{ \
cdef_filter_block_c(dst, stride, top, w, h, pri_strength, sec_strength, \
dir, damping, edges); \
}
cdef_fn(4, 4);
cdef_fn(4, 8);
cdef_fn(8, 8);
/*
* <code copied from libaom>
*/
/* Detect direction. 0 means 45-degree up-right, 2 is horizontal, and so on.
The search minimizes the weighted variance along all the lines in a
particular direction, i.e. the squared error between the input and a
"predicted" block where each pixel is replaced by the average along a line
in a particular direction. Since each direction have the same sum(x^2) term,
that term is never computed. See Section 2, step 2, of:
http://jmvalin.ca/notes/intra_paint.pdf */
static const uint16_t div_table[] = {
0, 840, 420, 280, 210, 168, 140, 120, 105
};
static int cdef_find_dir_c(const pixel *img, const ptrdiff_t stride,
unsigned *const var)
{
int i;
int32_t cost[8] = { 0 };
int partial[8][15] = { { 0 } };
int32_t best_cost = 0;
int best_dir = 0;
/* Instead of dividing by n between 2 and 8, we multiply by 3*5*7*8/n.
The output is then 840 times larger, but we don't care for finding
the max. */
for (i = 0; i < 8; i++) {
int j;
for (j = 0; j < 8; j++) {
int x;
/* We subtract 128 here to reduce the maximum range of the squared
partial sums. */
x = (img[i * PXSTRIDE(stride) + j] >> (BITDEPTH - 8)) - 128;
partial[0][i + j] += x;
partial[1][i + j / 2] += x;
partial[2][i] += x;
partial[3][3 + i - j / 2] += x;
partial[4][7 + i - j] += x;
partial[5][3 - i / 2 + j] += x;
partial[6][j] += x;
partial[7][i / 2 + j] += x;
}
}
for (i = 0; i < 8; i++) {
cost[2] += partial[2][i] * partial[2][i];
cost[6] += partial[6][i] * partial[6][i];
}
cost[2] *= div_table[8];
cost[6] *= div_table[8];
for (i = 0; i < 7; i++) {
cost[0] += (partial[0][i] * partial[0][i] +
partial[0][14 - i] * partial[0][14 - i]) *
div_table[i + 1];
cost[4] += (partial[4][i] * partial[4][i] +
partial[4][14 - i] * partial[4][14 - i]) *
div_table[i + 1];
}
cost[0] += partial[0][7] * partial[0][7] * div_table[8];
cost[4] += partial[4][7] * partial[4][7] * div_table[8];
for (i = 1; i < 8; i += 2) {
int j;
for (j = 0; j < 4 + 1; j++) {
cost[i] += partial[i][3 + j] * partial[i][3 + j];
}
cost[i] *= div_table[8];
for (j = 0; j < 4 - 1; j++) {
cost[i] += (partial[i][j] * partial[i][j] +
partial[i][10 - j] * partial[i][10 - j]) *
div_table[2 * j + 2];
}
}
for (i = 0; i < 8; i++) {
if (cost[i] > best_cost) {
best_cost = cost[i];
best_dir = i;
}
}
/* Difference between the optimal variance and the variance along the
orthogonal direction. Again, the sum(x^2) terms cancel out. */
*var = best_cost - cost[(best_dir + 4) & 7];
/* We'd normally divide by 840, but dividing by 1024 is close enough
for what we're going to do with this. */
*var >>= 10;
return best_dir;
}
/*
* </code copied from libaom>
*/
void bitfn(dav1d_cdef_dsp_init)(Dav1dCdefDSPContext *const c) {
c->dir = cdef_find_dir_c;
c->fb[0] = cdef_filter_block_8x8_c;
c->fb[1] = cdef_filter_block_4x8_c;
c->fb[2] = cdef_filter_block_4x4_c;
}