x265-1.7版本-common/pixel.cpp注释

100 阅读 0 评论 66 点赞

我是靠谱客的博主高大大白，这篇文章主要介绍x265-1.7版本-common/pixel.cpp注释，现在分享给大家，希望可以做个参考。

注：问号以及未注释部分会在x265-1.8版本内更新

复制代码

/*****************************************************************************
* Copyright (C) 2013 x265 project
*
* Authors: Steve Borho <steve@borho.org>
*
Mandar Gurav <mandar@multicorewareinc.com>
*
Mahesh Pittala <mahesh@multicorewareinc.com>
*
Min Chen <min.chen@multicorewareinc.com>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
02111, USA.
*
* This program is also available under a commercial proprietary license.
* For more information, contact us at license @ x265.com.
*****************************************************************************/
#include "common.h"
#include "primitives.h"
#include "x265.h"
#include <cstdlib> // abs()
using namespace x265;
namespace {
// place functions in anonymous namespace (file static)
/** 函数功能
： 计算SAD（8位）
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数 stride_pix1：计算块的步长
/*参数
pix2：参考块的首地址
/*参数 stride_pix2：参考块的步长
* 返回
：返回SAD值 */
template<int lx, int ly>
int sad(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
int sum = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
sum += abs(pix1[x] - pix2[x]);
pix1 += stride_pix1;
pix2 += stride_pix2;
}
return sum;
}
/** 函数功能
： 计算SAD（16位）
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数 stride_pix1：计算块的步长
/*参数
pix2：参考块的首地址
/*参数 stride_pix2：参考块的步长
* 返回
：返回SAD值 */
template<int lx, int ly>
int sad(const int16_t* pix1, intptr_t stride_pix1, const int16_t* pix2, intptr_t stride_pix2)
{
int sum = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
sum += abs(pix1[x] - pix2[x]);
pix1 += stride_pix1;
pix2 += stride_pix2;
}
return sum;
}
/** 函数功能
： 同时计算3个MV对应的3个SAD值
/*
调用范围
： ME中
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数
pix2：参考块的首地址
/*参数
pix3：参考块的首地址
/*参数
pix4：参考块的首地址
/*参数
frefstride：参考块的步长
/*参数
res：存储3个MV对应的3个SAD值
* 返回
：null */
template<int lx, int ly>
void sad_x3(const pixel* pix1, const pixel* pix2, const pixel* pix3, const pixel* pix4, intptr_t frefstride, int32_t* res)
{
res[0] = 0;
res[1] = 0;
res[2] = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
{
res[0] += abs(pix1[x] - pix2[x]);
res[1] += abs(pix1[x] - pix3[x]);
res[2] += abs(pix1[x] - pix4[x]);
}
pix1 += FENC_STRIDE;
//搜索块统一步长都为64
pix2 += frefstride;
pix3 += frefstride;
pix4 += frefstride;
}
}
/** 函数功能
： 同时计算4个MV对应的4个SAD值
/*
调用范围
： ME中
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数
pix2：参考块的首地址
/*参数
pix3：参考块的首地址
/*参数
pix4：参考块的首地址
/*参数
pix5：参考块的首地址
/*参数
frefstride：参考块的步长
/*参数
res：存储4个MV对应的4个SAD值
* 返回
：null */
template<int lx, int ly>
void sad_x4(const pixel* pix1, const pixel* pix2, const pixel* pix3, const pixel* pix4, const pixel* pix5, intptr_t frefstride, int32_t* res)
{
res[0] = 0;
res[1] = 0;
res[2] = 0;
res[3] = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
{
res[0] += abs(pix1[x] - pix2[x]);
res[1] += abs(pix1[x] - pix3[x]);
res[2] += abs(pix1[x] - pix4[x]);
res[3] += abs(pix1[x] - pix5[x]);
}
pix1 += FENC_STRIDE;//搜索块统一步长都为64
pix2 += frefstride;
pix3 += frefstride;
pix4 += frefstride;
pix5 += frefstride;
}
}
template<int lx, int ly, class T1, class T2>
int sse(const T1* pix1, intptr_t stride_pix1, const T2* pix2, intptr_t stride_pix2)
{
int sum = 0;
int tmp;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
{
tmp = pix1[x] - pix2[x];
sum += (tmp * tmp);
}
pix1 += stride_pix1;
pix2 += stride_pix2;
}
return sum;
}
#define BITS_PER_SUM (8 * sizeof(sum_t))
#define HADAMARD4(d0, d1, d2, d3, s0, s1, s2, s3) { 
sum2_t t0 = s0 + s1; 
sum2_t t1 = s0 - s1; 
sum2_t t2 = s2 + s3; 
sum2_t t3 = s2 - s3; 
d0 = t0 + t2; 
d2 = t0 - t2; 
d1 = t1 + t3; 
d3 = t1 - t3; 
}
// in: a pseudo-simd number of the form x+(y<<16)
// return: abs(x)+(abs(y)<<16)
inline sum2_t abs2(sum2_t a)
{
sum2_t s = ((a >> (BITS_PER_SUM - 1)) & (((sum2_t)1 << BITS_PER_SUM) + 1)) * ((sum_t)-1);
return (a + s) ^ s;
}
int satd_4x4(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
sum2_t tmp[4][2];
sum2_t a0, a1, a2, a3, b0, b1;
sum2_t sum = 0;
for (int i = 0; i < 4; i++, pix1 += stride_pix1, pix2 += stride_pix2)
{
a0 = pix1[0] - pix2[0];
a1 = pix1[1] - pix2[1];
b0 = (a0 + a1) + ((a0 - a1) << BITS_PER_SUM);
a2 = pix1[2] - pix2[2];
a3 = pix1[3] - pix2[3];
b1 = (a2 + a3) + ((a2 - a3) << BITS_PER_SUM);
tmp[i][0] = b0 + b1;
tmp[i][1] = b0 - b1;
}
for (int i = 0; i < 2; i++)
{
HADAMARD4(a0, a1, a2, a3, tmp[0][i], tmp[1][i], tmp[2][i], tmp[3][i]);
a0 = abs2(a0) + abs2(a1) + abs2(a2) + abs2(a3);
sum += ((sum_t)a0) + (a0 >> BITS_PER_SUM);
}
return (int)(sum >> 1);
}
static int satd_4x4(const int16_t* pix1, intptr_t stride_pix1)
{
int32_t tmp[4][4];
int32_t s01, s23, d01, d23;
int32_t satd = 0;
int d;
for (d = 0; d < 4; d++, pix1 += stride_pix1)
{
s01 = pix1[0] + pix1[1];
s23 = pix1[2] + pix1[3];
d01 = pix1[0] - pix1[1];
d23 = pix1[2] - pix1[3];
tmp[d][0] = s01 + s23;
tmp[d][1] = s01 - s23;
tmp[d][2] = d01 - d23;
tmp[d][3] = d01 + d23;
}
for (d = 0; d < 4; d++)
{
s01 = tmp[0][d] + tmp[1][d];
s23 = tmp[2][d] + tmp[3][d];
d01 = tmp[0][d] - tmp[1][d];
d23 = tmp[2][d] - tmp[3][d];
satd += abs(s01 + s23) + abs(s01 - s23) + abs(d01 - d23) + abs(d01 + d23);
}
return (int)(satd / 2);
}
// x264's SWAR version of satd 8x4, performs two 4x4 SATDs at once
int satd_8x4(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
sum2_t tmp[4][4];
sum2_t a0, a1, a2, a3;
sum2_t sum = 0;
for (int i = 0; i < 4; i++, pix1 += stride_pix1, pix2 += stride_pix2)
{
a0 = (pix1[0] - pix2[0]) + ((sum2_t)(pix1[4] - pix2[4]) << BITS_PER_SUM);
a1 = (pix1[1] - pix2[1]) + ((sum2_t)(pix1[5] - pix2[5]) << BITS_PER_SUM);
a2 = (pix1[2] - pix2[2]) + ((sum2_t)(pix1[6] - pix2[6]) << BITS_PER_SUM);
a3 = (pix1[3] - pix2[3]) + ((sum2_t)(pix1[7] - pix2[7]) << BITS_PER_SUM);
HADAMARD4(tmp[i][0], tmp[i][1], tmp[i][2], tmp[i][3], a0, a1, a2, a3);
}
for (int i = 0; i < 4; i++)
{
HADAMARD4(a0, a1, a2, a3, tmp[0][i], tmp[1][i], tmp[2][i], tmp[3][i]);
sum += abs2(a0) + abs2(a1) + abs2(a2) + abs2(a3);
}
return (((sum_t)sum) + (sum >> BITS_PER_SUM)) >> 1;
}
template<int w, int h>
// calculate satd in blocks of 4x4
int satd4(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
int satd = 0;
for (int row = 0; row < h; row += 4)
for (int col = 0; col < w; col += 4)
satd += satd_4x4(pix1 + row * stride_pix1 + col, stride_pix1,
pix2 + row * stride_pix2 + col, stride_pix2);
return satd;
}
template<int w, int h>
// calculate satd in blocks of 8x4
int satd8(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
int satd = 0;
for (int row = 0; row < h; row += 4)
for (int col = 0; col < w; col += 8)
satd += satd_8x4(pix1 + row * stride_pix1 + col, stride_pix1,
pix2 + row * stride_pix2 + col, stride_pix2);
return satd;
}
inline int _sa8d_8x8(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
sum2_t tmp[8][4];
sum2_t a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3;
sum2_t sum = 0;
for (int i = 0; i < 8; i++, pix1 += i_pix1, pix2 += i_pix2)
{
a0 = pix1[0] - pix2[0];
a1 = pix1[1] - pix2[1];
b0 = (a0 + a1) + ((a0 - a1) << BITS_PER_SUM);
a2 = pix1[2] - pix2[2];
a3 = pix1[3] - pix2[3];
b1 = (a2 + a3) + ((a2 - a3) << BITS_PER_SUM);
a4 = pix1[4] - pix2[4];
a5 = pix1[5] - pix2[5];
b2 = (a4 + a5) + ((a4 - a5) << BITS_PER_SUM);
a6 = pix1[6] - pix2[6];
a7 = pix1[7] - pix2[7];
b3 = (a6 + a7) + ((a6 - a7) << BITS_PER_SUM);
HADAMARD4(tmp[i][0], tmp[i][1], tmp[i][2], tmp[i][3], b0, b1, b2, b3);
}
for (int i = 0; i < 4; i++)
{
HADAMARD4(a0, a1, a2, a3, tmp[0][i], tmp[1][i], tmp[2][i], tmp[3][i]);
HADAMARD4(a4, a5, a6, a7, tmp[4][i], tmp[5][i], tmp[6][i], tmp[7][i]);
b0
= abs2(a0 + a4) + abs2(a0 - a4);
b0 += abs2(a1 + a5) + abs2(a1 - a5);
b0 += abs2(a2 + a6) + abs2(a2 - a6);
b0 += abs2(a3 + a7) + abs2(a3 - a7);
sum += (sum_t)b0 + (b0 >> BITS_PER_SUM);
}
return (int)sum;
}
int sa8d_8x8(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
return (int)((_sa8d_8x8(pix1, i_pix1, pix2, i_pix2) + 2) >> 2);
}
inline int _sa8d_8x8(const int16_t* pix1, intptr_t i_pix1)
{
int32_t tmp[8][8];
int32_t a0, a1, a2, a3, a4, a5, a6, a7;
int32_t sum = 0;
for (int i = 0; i < 8; i++, pix1 += i_pix1)
{
a0 = pix1[0] + pix1[1];
a1 = pix1[2] + pix1[3];
a2 = pix1[4] + pix1[5];
a3 = pix1[6] + pix1[7];
a4 = pix1[0] - pix1[1];
a5 = pix1[2] - pix1[3];
a6 = pix1[4] - pix1[5];
a7 = pix1[6] - pix1[7];
tmp[i][0] = (a0 + a1) + (a2 + a3);
tmp[i][1] = (a0 + a1) - (a2 + a3);
tmp[i][2] = (a0 - a1) + (a2 - a3);
tmp[i][3] = (a0 - a1) - (a2 - a3);
tmp[i][4] = (a4 + a5) + (a6 + a7);
tmp[i][5] = (a4 + a5) - (a6 + a7);
tmp[i][6] = (a4 - a5) + (a6 - a7);
tmp[i][7] = (a4 - a5) - (a6 - a7);
}
for (int i = 0; i < 8; i++)
{
a0 = (tmp[0][i] + tmp[1][i]) + (tmp[2][i] + tmp[3][i]);
a2 = (tmp[0][i] + tmp[1][i]) - (tmp[2][i] + tmp[3][i]);
a1 = (tmp[0][i] - tmp[1][i]) + (tmp[2][i] - tmp[3][i]);
a3 = (tmp[0][i] - tmp[1][i]) - (tmp[2][i] - tmp[3][i]);
a4 = (tmp[4][i] + tmp[5][i]) + (tmp[6][i] + tmp[7][i]);
a6 = (tmp[4][i] + tmp[5][i]) - (tmp[6][i] + tmp[7][i]);
a5 = (tmp[4][i] - tmp[5][i]) + (tmp[6][i] - tmp[7][i]);
a7 = (tmp[4][i] - tmp[5][i]) - (tmp[6][i] - tmp[7][i]);
a0 = abs(a0 + a4) + abs(a0 - a4);
a0 += abs(a1 + a5) + abs(a1 - a5);
a0 += abs(a2 + a6) + abs(a2 - a6);
a0 += abs(a3 + a7) + abs(a3 - a7);
sum += a0;
}
return (int)sum;
}
int sa8d_8x8(const int16_t* pix1, intptr_t i_pix1)
{
return (int)((_sa8d_8x8(pix1, i_pix1) + 2) >> 2);
}
int sa8d_16x16(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
int sum = _sa8d_8x8(pix1, i_pix1, pix2, i_pix2)
+ _sa8d_8x8(pix1 + 8, i_pix1, pix2 + 8, i_pix2)
+ _sa8d_8x8(pix1 + 8 * i_pix1, i_pix1, pix2 + 8 * i_pix2, i_pix2)
+ _sa8d_8x8(pix1 + 8 + 8 * i_pix1, i_pix1, pix2 + 8 + 8 * i_pix2, i_pix2);
// This matches x264 sa8d_16x16, but is slightly different from HM's behavior because
// this version only rounds once at the end
return (sum + 2) >> 2;
}
template<int w, int h>
// Calculate sa8d in blocks of 8x8
int sa8d8(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
int cost = 0;
for (int y = 0; y < h; y += 8)
for (int x = 0; x < w; x += 8)
cost += sa8d_8x8(pix1 + i_pix1 * y + x, i_pix1, pix2 + i_pix2 * y + x, i_pix2);
return cost;
}
template<int w, int h>
// Calculate sa8d in blocks of 16x16
int sa8d16(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
int cost = 0;
for (int y = 0; y < h; y += 16)
for (int x = 0; x < w; x += 16)
cost += sa8d_16x16(pix1 + i_pix1 * y + x, i_pix1, pix2 + i_pix2 * y + x, i_pix2);
return cost;
}
template<int size>
int pixel_ssd_s_c(const int16_t* a, intptr_t dstride)
{
int sum = 0;
for (int y = 0; y < size; y++)
{
for (int x = 0; x < size; x++)
sum += a[x] * a[x];
a += dstride;
}
return sum;
}
template<int size>
void blockfill_s_c(int16_t* dst, intptr_t dstride, int16_t val)
{
for (int y = 0; y < size; y++)
for (int x = 0; x < size; x++)
dst[y * dstride + x] = val;
}
template<int size>
void cpy2Dto1D_shl(int16_t* dst, const int16_t* src, intptr_t srcStride, int shift)
{
X265_CHECK(((intptr_t)dst & 15) == 0, "dst alignment errorn");
X265_CHECK((((intptr_t)src | (srcStride * sizeof(*src))) & 15) == 0 || size == 4, "src alignment errorn");
X265_CHECK(shift >= 0, "invalid shiftn");
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = src[j] << shift;
src += srcStride;
dst += size;
}
}
template<int size>
void cpy2Dto1D_shr(int16_t* dst, const int16_t* src, intptr_t srcStride, int shift)
{
X265_CHECK(((intptr_t)dst & 15) == 0, "dst alignment errorn");
X265_CHECK((((intptr_t)src | (srcStride * sizeof(*src))) & 15) == 0 || size == 4, "src alignment errorn");
X265_CHECK(shift > 0, "invalid shiftn");
int16_t round = 1 << (shift - 1);
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = (src[j] + round) >> shift;
src += srcStride;
dst += size;
}
}
template<int size>
void cpy1Dto2D_shl(int16_t* dst, const int16_t* src, intptr_t dstStride, int shift)
{
X265_CHECK((((intptr_t)dst | (dstStride * sizeof(*dst))) & 15) == 0 || size == 4, "dst alignment errorn");
X265_CHECK(((intptr_t)src & 15) == 0, "src alignment errorn");
X265_CHECK(shift >= 0, "invalid shiftn");
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = src[j] << shift;
src += size;
dst += dstStride;
}
}
template<int size>
void cpy1Dto2D_shr(int16_t* dst, const int16_t* src, intptr_t dstStride, int shift)
{
X265_CHECK((((intptr_t)dst | (dstStride * sizeof(*dst))) & 15) == 0 || size == 4, "dst alignment errorn");
X265_CHECK(((intptr_t)src & 15) == 0, "src alignment errorn");
X265_CHECK(shift > 0, "invalid shiftn");
int16_t round = 1 << (shift - 1);
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = (src[j] + round) >> shift;
src += size;
dst += dstStride;
}
}
template<int blockSize>
void getResidual(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride)
{
for (int y = 0; y < blockSize; y++)
{
for (int x = 0; x < blockSize; x++)
residual[x] = static_cast<int16_t>(fenc[x]) - static_cast<int16_t>(pred[x]);
fenc += stride;
residual += stride;
pred += stride;
}
}
template<int blockSize>
void transpose(pixel* dst, const pixel* src, intptr_t stride)
{
for (int k = 0; k < blockSize; k++)
for (int l = 0; l < blockSize; l++)
dst[k * blockSize + l] = src[l * stride + k];
}
void weight_sp_c(const int16_t* src, pixel* dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset)
{
int x, y;
#if CHECKED_BUILD || _DEBUG
const int correction = (IF_INTERNAL_PREC - X265_DEPTH);
X265_CHECK(!((w0 << 6) > 32767), "w0 using more than 16 bits, asm output will mismatchn");
X265_CHECK(!(round > 32767), "round using more than 16 bits, asm output will mismatchn");
X265_CHECK((shift >= correction), "shift must be include factor correction, please update ASM ABIn");
#endif
for (y = 0; y <= height - 1; y++)
{
for (x = 0; x <= width - 1; )
{
// note: width can be odd
dst[x] = x265_clip(((w0 * (src[x] + IF_INTERNAL_OFFS) + round) >> shift) + offset);
x++;
}
src += srcStride;
dst += dstStride;
}
}
/** 函数功能
： P帧加权参考帧获取
/*
调用范围
： 只在MotionReference::applyWeight、LookaheadTLD::weightCostLuma、LookaheadTLD::weightsAnalyse和weightCost函数中被调用
* 参数 src
： 原始P参考帧
* 参数 dst
： 加权后P参考帧
* 参数 stride
： 步长
* 参数 width
： 宽度
* 参数 height
： 高度
* 参数 w0
： 加权系数
* 参数 round
： 四舍五入操作
* 参数 shift
： 右移位数（前面为提高精度左移位，现在恢复原有精度）
* 参数 offset
： offset信息 整帧所有像素偏移值
* 返回
： null */
void weight_pp_c(const pixel* src, pixel* dst, intptr_t stride, int width, int height, int w0, int round, int shift, int offset)
{
int x, y;
const int correction = (IF_INTERNAL_PREC - X265_DEPTH);
X265_CHECK(!(width & 15), "weightp alignment errorn");
X265_CHECK(!((w0 << 6) > 32767), "w0 using more than 16 bits, asm output will mismatchn");
X265_CHECK(!(round > 32767), "round using more than 16 bits, asm output will mismatchn");
X265_CHECK((shift >= correction), "shift must be include factor correction, please update ASM ABIn");
X265_CHECK(!(round & ((1 << correction) - 1)), "round must be include factor correction, please update ASM ABIn");
for (y = 0; y <= height - 1; y++)
{
for (x = 0; x <= width - 1; )
{
// simulating pixel to short conversion
int16_t val = src[x] << correction;
dst[x] = x265_clip(((w0 * (val) + round) >> shift) + offset);
x++;
}
src += stride;
dst += stride;
}
}
/** 函数功能
： 获取两块的平均值
/*参数
lx：块的宽度
/*参数
ly：块的高度
* 参数 dst
：平均后像素存储位置
* 参数 dstride
：步长
* 参数 src0
：块0首地址
* 参数 sstride0
：块0步长
* 参数 src1
：块1首地址
* 参数 sstride1
：块1步长
* 返回
： null */
template<int lx, int ly>
void pixelavg_pp(pixel* dst, intptr_t dstride, const pixel* src0, intptr_t sstride0, const pixel* src1, intptr_t sstride1, int)//获取两块的平均值
{
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
dst[x] = (src0[x] + src1[x] + 1) >> 1;
src0 += sstride0;
src1 += sstride1;
dst += dstride;
}
}
void scale1D_128to64(pixel *dst, const pixel *src)
{
int x;
const pixel* src1 = src;
const pixel* src2 = src + 128;
pixel* dst1 = dst;
pixel* dst2 = dst + 64/*128*/;
for (x = 0; x < 128; x += 2)
{
// Top pixel
pixel pix0 = src1[(x + 0)];
pixel pix1 = src1[(x + 1)];
// Left pixel
pixel pix2 = src2[(x + 0)];
pixel pix3 = src2[(x + 1)];
int sum1 = pix0 + pix1;
int sum2 = pix2 + pix3;
dst1[x >> 1] = (pixel)((sum1 + 1) >> 1);
dst2[x >> 1] = (pixel)((sum2 + 1) >> 1);
}
}
void scale2D_64to32(pixel* dst, const pixel* src, intptr_t stride)
{
uint32_t x, y;
for (y = 0; y < 64; y += 2)
{
for (x = 0; x < 64; x += 2)
{
pixel pix0 = src[(y + 0) * stride + (x + 0)];
pixel pix1 = src[(y + 0) * stride + (x + 1)];
pixel pix2 = src[(y + 1) * stride + (x + 0)];
pixel pix3 = src[(y + 1) * stride + (x + 1)];
int sum = pix0 + pix1 + pix2 + pix3;
dst[y / 2 * 32 + x / 2] = (pixel)((sum + 2) >> 2);
}
}
}
/** 函数功能
：将原始帧视频亮度作1/2下采样
/*
调用范围
：只在Lowres::init函数中被调用
* 参数 src0
：原始视频帧亮度首地址:origPic->m_picOrg[0]
* 参数 dst0
：lowresPlane[0]
* 参数 dsth
：lowresPlane[1]
* 参数 dstv
：lowresPlane[2]
* 参数 dstc
：lowresPlane[3]
* 参数 src_stride：原始帧亮度步长
* 参数 dst_stride：下采样视频亮度步长
* 参数 width
：下采样视频的宽度
* 参数 heigh
：下采样视频的高度
* 返回值
： null
*/
void frame_init_lowres_core(const pixel* src0, pixel* dst0, pixel* dsth, pixel* dstv, pixel* dstc,
intptr_t src_stride, intptr_t dst_stride, int width, int height)
{
/* downscale and generate 4 hpel planes for lookahead */
/* 这样做的目的是更好的通过1/2下采样视频的编码估计原始视频编码状态
将亮度分四种方法进行1/2下采样
原始点：
82
89
86
86
93
85
89
99 101 113
96
97
97 100 104
106 108 107 111 109
127 156 133 139 137
0: 在行列选择偶数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
* * + +
* * + +
- - # #
- - # #
87
94
102 104
87 = ((((82 + 85 + 1) >> 1) + ((89 + 89 + 1) >> 1) + 1) >> 1)
94 = ((((86 + 99 + 1) >> 1) + ((86 + 101 + 1) >> 1) + 1) >> 1)
102= ((((96 + 106 + 1) >> 1) + ((97 + 108 + 1) >> 1) + 1) >> 1)
104 = ((((97 + 107 + 1) >> 1) + ((100 + 111 + 1) >> 1) + 1) >> 1)
h: 在行选择偶数像素点，在列选择奇数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
= * * = + +
= * * = + +
= - - = # #
= - - = # #
91
99
103 107
91 = ((((89 + 89 + 1) >> 1) + ((86 + 99 + 1) >> 1) + 1) >> 1)
99 = ((((86 + 101 + 1) >> 1) + ((93 + 113 + 1) >> 1) + 1) >> 1)
v: 在行选择奇数像素点，在列选择偶数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
= = = =
* * + +
* * + +
= = = =
- - # #
- - # #
92 100
125 123
92 = ((((85 + 96 + 1) >> 1) + ((89 + 97 + 1) >> 1) + 1) >> 1)
在行列选择奇数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
= = = = = =
= * * = + +
= * * = + +
= = = = = =
= - - = # #
= - - = # #
96 105
126 124
96 = ((((89 + 97 + 1) >> 1) + ((99 + 97 + 1) >> 1) + 1) >> 1)
**/
for (int y = 0; y < height; y++)
{
const pixel* src1 = src0 + src_stride;
const pixel* src2 = src1 + src_stride;
for (int x = 0; x < width; x++)
{
// slower than naive bilinear, but matches asm
#define FILTER(a, b, c, d) ((((a + b + 1) >> 1) + ((c + d + 1) >> 1) + 1) >> 1)
dst0[x] = FILTER(src0[2 * x], src1[2 * x], src0[2 * x + 1], src1[2 * x + 1]);
dsth[x] = FILTER(src0[2 * x + 1], src1[2 * x + 1], src0[2 * x + 2], src1[2 * x + 2]);
dstv[x] = FILTER(src1[2 * x], src2[2 * x], src1[2 * x + 1], src2[2 * x + 1]);
dstc[x] = FILTER(src1[2 * x + 1], src2[2 * x + 1], src1[2 * x + 2], src2[2 * x + 2]);
#undef FILTER
}
src0 += src_stride * 2;
dst0 += dst_stride;
dsth += dst_stride;
dstv += dst_stride;
dstc += dst_stride;
}
}
/* structural similarity metric */
void ssim_4x4x2_core(const pixel* pix1, intptr_t stride1, const pixel* pix2, intptr_t stride2, int sums[2][4])
{
for (int z = 0; z < 2; z++)
{
uint32_t s1 = 0, s2 = 0, ss = 0, s12 = 0;
for (int y = 0; y < 4; y++)
{
for (int x = 0; x < 4; x++)
{
int a = pix1[x + y * stride1];
int b = pix2[x + y * stride2];
s1 += a;
s2 += b;
ss += a * a;
ss += b * b;
s12 += a * b;
}
}
sums[z][0] = s1;
sums[z][1] = s2;
sums[z][2] = ss;
sums[z][3] = s12;
pix1 += 4;
pix2 += 4;
}
}
float ssim_end_1(int s1, int s2, int ss, int s12)
{
/* Maximum value for 10-bit is: ss*64 = (2^10-1)^2*16*4*64 = 4286582784, which will overflow in some cases.
* s1*s1, s2*s2, and s1*s2 also obtain this value for edge cases: ((2^10-1)*16*4)^2 = 4286582784.
* Maximum value for 9-bit is: ss*64 = (2^9-1)^2*16*4*64 = 1069551616, which will not overflow. */
#define PIXEL_MAX ((1 << X265_DEPTH) - 1)
#if HIGH_BIT_DEPTH
X265_CHECK(X265_DEPTH == 10, "ssim invalid depthn");
#define type float
static const float ssim_c1 = (float)(.01 * .01 * PIXEL_MAX * PIXEL_MAX * 64);
static const float ssim_c2 = (float)(.03 * .03 * PIXEL_MAX * PIXEL_MAX * 64 * 63);
#else
X265_CHECK(X265_DEPTH == 8, "ssim invalid depthn");
#define type int
static const int ssim_c1 = (int)(.01 * .01 * PIXEL_MAX * PIXEL_MAX * 64 + .5);
static const int ssim_c2 = (int)(.03 * .03 * PIXEL_MAX * PIXEL_MAX * 64 * 63 + .5);
#endif
type fs1 = (type)s1;
type fs2 = (type)s2;
type fss = (type)ss;
type fs12 = (type)s12;
type vars = (type)(fss * 64 - fs1 * fs1 - fs2 * fs2);
type covar = (type)(fs12 * 64 - fs1 * fs2);
return (float)(2 * fs1 * fs2 + ssim_c1) * (float)(2 * covar + ssim_c2)
/ ((float)(fs1 * fs1 + fs2 * fs2 + ssim_c1) * (float)(vars + ssim_c2));
#undef type
#undef PIXEL_MAX
}
float ssim_end_4(int sum0[5][4], int sum1[5][4], int width)
{
float ssim = 0.0;
for (int i = 0; i < width; i++)
{
ssim += ssim_end_1(sum0[i][0] + sum0[i + 1][0] + sum1[i][0] + sum1[i + 1][0],
sum0[i][1] + sum0[i + 1][1] + sum1[i][1] + sum1[i + 1][1],
sum0[i][2] + sum0[i + 1][2] + sum1[i][2] + sum1[i + 1][2],
sum0[i][3] + sum0[i + 1][3] + sum1[i][3] + sum1[i + 1][3]);
}
return ssim;
}
/** 函数功能
：返回一个64位整数,低32位存储当前nxn所有元素的和，高32位存储当前nxn所有元素的平方和
/*
调用范围
：只在Lowres::init函数中被调用
* 参数 pix
：待计算的块
* 参数 i_stride：步长
* 返回值
：返回一个64位整数,低32位存储当前nxn所有元素的和，高32位存储当前nxn所有元素的平方和
*/
template<int size>
uint64_t pixel_var(const pixel* pix, intptr_t i_stride)
{
uint32_t sum = 0, sqr = 0;
for (int y = 0; y < size; y++)
{
for (int x = 0; x < size; x++)
{
sum += pix[x];
sqr += pix[x] * pix[x];
}
pix += i_stride;
}
return sum + ((uint64_t)sqr << 32);
}
#if defined(_MSC_VER)
#pragma warning(disable: 4127) // conditional expression is constant
#endif
template<int size>
int psyCost_pp(const pixel* source, intptr_t sstride, const pixel* recon, intptr_t rstride)
{
static pixel zeroBuf[8] /* = { 0 } */;
if (size)
{
int dim = 1 << (size + 2);
uint32_t totEnergy = 0;
for (int i = 0; i < dim; i += 8)
{
for (int j = 0; j < dim; j+= 8)
{
/* AC energy, measured by sa8d (AC + DC) minus SAD (DC) */
int sourceEnergy = sa8d_8x8(source + i * sstride + j, sstride, zeroBuf, 0) -
(sad<8, 8>(source + i * sstride + j, sstride, zeroBuf, 0) >> 2);
int reconEnergy =
sa8d_8x8(recon + i * rstride + j, rstride, zeroBuf, 0) -
(sad<8, 8>(recon + i * rstride + j, rstride, zeroBuf, 0) >> 2);
totEnergy += abs(sourceEnergy - reconEnergy);
}
}
return totEnergy;
}
else
{
/* 4x4 is too small for sa8d */
int sourceEnergy = satd_4x4(source, sstride, zeroBuf, 0) - (sad<4, 4>(source, sstride, zeroBuf, 0) >> 2);
int reconEnergy = satd_4x4(recon, rstride, zeroBuf, 0) - (sad<4, 4>(recon, rstride, zeroBuf, 0) >> 2);
return abs(sourceEnergy - reconEnergy);
}
}
template<int size>
int psyCost_ss(const int16_t* source, intptr_t sstride, const int16_t* recon, intptr_t rstride)
{
static int16_t zeroBuf[8] /* = { 0 } */;
if (size)
{
int dim = 1 << (size + 2);
uint32_t totEnergy = 0;
for (int i = 0; i < dim; i += 8)
{
for (int j = 0; j < dim; j+= 8)
{
/* AC energy, measured by sa8d (AC + DC) minus SAD (DC) */
int sourceEnergy = sa8d_8x8(source + i * sstride + j, sstride) -
(sad<8, 8>(source + i * sstride + j, sstride, zeroBuf, 0) >> 2);
int reconEnergy =
sa8d_8x8(recon + i * rstride + j, rstride) -
(sad<8, 8>(recon + i * rstride + j, rstride, zeroBuf, 0) >> 2);
totEnergy += abs(sourceEnergy - reconEnergy);
}
}
return totEnergy;
}
else
{
/* 4x4 is too small for sa8d */
int sourceEnergy = satd_4x4(source, sstride) - (sad<4, 4>(source, sstride, zeroBuf, 0) >> 2);
int reconEnergy = satd_4x4(recon, rstride) - (sad<4, 4>(recon, rstride, zeroBuf, 0) >> 2);
return abs(sourceEnergy - reconEnergy);
}
}
template<int bx, int by>
void blockcopy_pp_c(pixel* a, intptr_t stridea, const pixel* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = b[x];
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void blockcopy_ss_c(int16_t* a, intptr_t stridea, const int16_t* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = b[x];
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void blockcopy_sp_c(pixel* a, intptr_t stridea, const int16_t* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
{
X265_CHECK((b[x] >= 0) && (b[x] <= ((1 << X265_DEPTH) - 1)), "blockcopy pixel size failn");
a[x] = (pixel)b[x];
}
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void blockcopy_ps_c(int16_t* a, intptr_t stridea, const pixel* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = (int16_t)b[x];
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void pixel_sub_ps_c(int16_t* a, intptr_t dstride, const pixel* b0, const pixel* b1, intptr_t sstride0, intptr_t sstride1)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = (int16_t)(b0[x] - b1[x]);
b0 += sstride0;
b1 += sstride1;
a += dstride;
}
}
template<int bx, int by>
void pixel_add_ps_c(pixel* a, intptr_t dstride, const pixel* b0, const int16_t* b1, intptr_t sstride0, intptr_t sstride1)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = x265_clip(b0[x] + b1[x]);
b0 += sstride0;
b1 += sstride1;
a += dstride;
}
}
template<int bx, int by>
void addAvg(const int16_t* src0, const int16_t* src1, pixel* dst, intptr_t src0Stride, intptr_t src1Stride, intptr_t dstStride)
{
int shiftNum, offset;
shiftNum = IF_INTERNAL_PREC + 1 - X265_DEPTH;
offset = (1 << (shiftNum - 1)) + 2 * IF_INTERNAL_OFFS;
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x += 2)
{
dst[x + 0] = x265_clip((src0[x + 0] + src1[x + 0] + offset) >> shiftNum);
dst[x + 1] = x265_clip((src0[x + 1] + src1[x + 1] + offset) >> shiftNum);
}
src0 += src0Stride;
src1 += src1Stride;
dst
+= dstStride;
}
}
void planecopy_cp_c(const uint8_t* src, intptr_t srcStride, pixel* dst, intptr_t dstStride, int width, int height, int shift)
{
for (int r = 0; r < height; r++)
{
for (int c = 0; c < width; c++)
dst[c] = ((pixel)src[c]) << shift;
dst += dstStride;
src += srcStride;
}
}
void planecopy_sp_c(const uint16_t* src, intptr_t srcStride, pixel* dst, intptr_t dstStride, int width, int height, int shift, uint16_t mask)
{
for (int r = 0; r < height; r++)
{
for (int c = 0; c < width; c++)
dst[c] = (pixel)((src[c] >> shift) & mask);
dst += dstStride;
src += srcStride;
}
}
/** 函数功能
：计算每行的8x8的传播cost(累加传播cost + 加权intracost)*（intracost-最优cost）/intracost
/*
调用范围
：只在Lookahead::estimateCUPropagate函数中被调用
/*参数
dst: 存储当前行每个8x8的传播cost
/*参数 propagateIn：当前行的传播cost存储首地址
/*参数
intraCosts：当前行的intracost
/*参数
interCosts：当前行的intercost
/*参数
invQscales：当前行的AQ offsets
/*参数
fpsFactor：当前帧的帧率因子（一般为1.0）
/*参数
len：当前行的长度
* 返回
：返回SAD值 */
/* Estimate the total amount of influence on future quality that could be had if we
* were to improve the reference samples used to inter predict any given CU. */
void estimateCUPropagateCost(int* dst, const uint16_t* propagateIn, const int32_t* intraCosts, const uint16_t* interCosts,
const int32_t* invQscales, const double* fpsFactor, int len)
{
double fps = *fpsFactor / 256;
//因为invQscales存储扩大256倍的整数数据，这里除以256目的是缩放数据
for (int i = 0; i < len; i++)
{
double intraCost
= intraCosts[i] * invQscales[i]; //获取当前8x8块 加权intra cost
double propagateAmount = (double)propagateIn[i] + intraCost * fps; //累加传播cost 加上加权fps因子之后的intra cost
double propagateNum
= (double)intraCosts[i] - (interCosts[i] & ((1 << 14) - 1));//获取最优cost与intracost的差值
高14位的数字：0 表示 intra
1 表示前向搜索
2表示后向搜索
3 表示bi搜索
double propagateDenom
= (double)intraCosts[i];
//当前8x8块 intra cost
dst[i] = (int)(propagateAmount * propagateNum / propagateDenom + 0.5);//(累加传播cost + 加权intracost)*（intracost-最优cost）/intracost
}
}
}
// end anonymous namespace
namespace x265 {
// x265 private namespace
/* Extend the edges of a picture so that it may safely be used for motion
* compensation. This function assumes the picture is stored in a buffer with
* sufficient padding for the X and Y margins */
/** 函数功能
：将视频帧进行扩边，便于插值和ME搜索
/*
调用范围
：只在weightAnalyse和Lowres::init函数中被调用
* 参数 pic
：需要进行插值的视频帧数据
* 参数 stride
：视频帧步长
* 参数 width
：视频帧宽度
* 参数 height
：视频帧高度
* 参数 marginX ：两边需要扩边的宽度
* 参数 marginY ：上下需要扩边的高度
* 返回值
： null
*/
void extendPicBorder(pixel* pic, intptr_t stride, int width, int height, int marginX, int marginY)
{
/* extend left and right margins */
primitives.extendRowBorder(pic, stride, width, height, marginX); //asm 代码，快速实现一行扩边
/* copy top row to create above margin */
pixel* top = pic - marginX;
for (int y = 0; y < marginY; y++)
memcpy(top - (y + 1) * stride, top, stride * sizeof(pixel));
/* copy bottom row to create below margin */
pixel* bot = pic - marginX + (height - 1) * stride;
for (int y = 0; y < marginY; y++)
memcpy(bot + (y + 1) * stride, bot, stride * sizeof(pixel));
}
/* Initialize entries for pixel functions defined in this file */
void setupPixelPrimitives_c(EncoderPrimitives &p)
{
#define LUMA_PU(W, H) 
p.pu[LUMA_ ## W ## x ## H].copy_pp = blockcopy_pp_c<W, H>; 
p.pu[LUMA_ ## W ## x ## H].addAvg = addAvg<W, H>; 
p.pu[LUMA_ ## W ## x ## H].sad = sad<W, H>; 
p.pu[LUMA_ ## W ## x ## H].sad_x3 = sad_x3<W, H>; 
p.pu[LUMA_ ## W ## x ## H].sad_x4 = sad_x4<W, H>; 
p.pu[LUMA_ ## W ## x ## H].pixelavg_pp = pixelavg_pp<W, H>;
#define LUMA_CU(W, H) 
p.cu[BLOCK_ ## W ## x ## H].sub_ps
= pixel_sub_ps_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].add_ps
= pixel_add_ps_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].copy_sp
= blockcopy_sp_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].copy_ps
= blockcopy_ps_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].copy_ss
= blockcopy_ss_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].blockfill_s
= blockfill_s_c<W>;

p.cu[BLOCK_ ## W ## x ## H].cpy2Dto1D_shl = cpy2Dto1D_shl<W>; 
p.cu[BLOCK_ ## W ## x ## H].cpy2Dto1D_shr = cpy2Dto1D_shr<W>; 
p.cu[BLOCK_ ## W ## x ## H].cpy1Dto2D_shl = cpy1Dto2D_shl<W>; 
p.cu[BLOCK_ ## W ## x ## H].cpy1Dto2D_shr = cpy1Dto2D_shr<W>; 
p.cu[BLOCK_ ## W ## x ## H].psy_cost_pp
= psyCost_pp<BLOCK_ ## W ## x ## H>; 
p.cu[BLOCK_ ## W ## x ## H].psy_cost_ss
= psyCost_ss<BLOCK_ ## W ## x ## H>; 
p.cu[BLOCK_ ## W ## x ## H].transpose
= transpose<W>; 
p.cu[BLOCK_ ## W ## x ## H].ssd_s
= pixel_ssd_s_c<W>; 
p.cu[BLOCK_ ## W ## x ## H].var
= pixel_var<W>; 
p.cu[BLOCK_ ## W ## x ## H].calcresidual
= getResidual<W>; 
p.cu[BLOCK_ ## W ## x ## H].sse_pp
= sse<W, H, pixel, pixel>; 
p.cu[BLOCK_ ## W ## x ## H].sse_ss
= sse<W, H, int16_t, int16_t>;
LUMA_PU(4, 4);
LUMA_PU(8, 8);
LUMA_PU(16, 16);
LUMA_PU(32, 32);
LUMA_PU(64, 64);
LUMA_PU(4, 8);
LUMA_PU(8, 4);
LUMA_PU(16,
8);
LUMA_PU(8, 16);
LUMA_PU(16, 12);
LUMA_PU(12, 16);
LUMA_PU(16,
4);
LUMA_PU(4, 16);
LUMA_PU(32, 16);
LUMA_PU(16, 32);
LUMA_PU(32, 24);
LUMA_PU(24, 32);
LUMA_PU(32,
8);
LUMA_PU(8, 32);
LUMA_PU(64, 32);
LUMA_PU(32, 64);
LUMA_PU(64, 48);
LUMA_PU(48, 64);
LUMA_PU(64, 16);
LUMA_PU(16, 64);
p.pu[LUMA_4x4].satd
= satd_4x4;
p.pu[LUMA_8x8].satd
= satd8<8, 8>;
p.pu[LUMA_8x4].satd
= satd_8x4;
p.pu[LUMA_4x8].satd
= satd4<4, 8>;
p.pu[LUMA_16x16].satd = satd8<16, 16>;
p.pu[LUMA_16x8].satd
= satd8<16, 8>;
p.pu[LUMA_8x16].satd
= satd8<8, 16>;
p.pu[LUMA_16x12].satd = satd8<16, 12>;
p.pu[LUMA_12x16].satd = satd4<12, 16>;
p.pu[LUMA_16x4].satd
= satd8<16, 4>;
p.pu[LUMA_4x16].satd
= satd4<4, 16>;
p.pu[LUMA_32x32].satd = satd8<32, 32>;
p.pu[LUMA_32x16].satd = satd8<32, 16>;
p.pu[LUMA_16x32].satd = satd8<16, 32>;
p.pu[LUMA_32x24].satd = satd8<32, 24>;
p.pu[LUMA_24x32].satd = satd8<24, 32>;
p.pu[LUMA_32x8].satd
= satd8<32, 8>;
p.pu[LUMA_8x32].satd
= satd8<8, 32>;
p.pu[LUMA_64x64].satd = satd8<64, 64>;
p.pu[LUMA_64x32].satd = satd8<64, 32>;
p.pu[LUMA_32x64].satd = satd8<32, 64>;
p.pu[LUMA_64x48].satd = satd8<64, 48>;
p.pu[LUMA_48x64].satd = satd8<48, 64>;
p.pu[LUMA_64x16].satd = satd8<64, 16>;
p.pu[LUMA_16x64].satd = satd8<16, 64>;
LUMA_CU(4, 4);
LUMA_CU(8, 8);
LUMA_CU(16, 16);
LUMA_CU(32, 32);
LUMA_CU(64, 64);
p.cu[BLOCK_4x4].sa8d
= satd_4x4;
p.cu[BLOCK_8x8].sa8d
= sa8d_8x8;
p.cu[BLOCK_16x16].sa8d = sa8d_16x16;
p.cu[BLOCK_32x32].sa8d = sa8d16<32, 32>;
p.cu[BLOCK_64x64].sa8d = sa8d16<64, 64>;
#define CHROMA_PU_420(W, H) 
p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].addAvg
= addAvg<W, H>;

p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].copy_pp = blockcopy_pp_c<W, H>; 
CHROMA_PU_420(2, 2);
CHROMA_PU_420(2, 4);
CHROMA_PU_420(4, 4);
CHROMA_PU_420(8, 8);
CHROMA_PU_420(16, 16);
CHROMA_PU_420(32, 32);
CHROMA_PU_420(4, 2);
CHROMA_PU_420(8, 4);
CHROMA_PU_420(4, 8);
CHROMA_PU_420(8, 6);
CHROMA_PU_420(6, 8);
CHROMA_PU_420(8, 2);
CHROMA_PU_420(2, 8);
CHROMA_PU_420(16, 8);
CHROMA_PU_420(8,
16);
CHROMA_PU_420(16, 12);
CHROMA_PU_420(12, 16);
CHROMA_PU_420(16, 4);
CHROMA_PU_420(4,
16);
CHROMA_PU_420(32, 16);
CHROMA_PU_420(16, 32);
CHROMA_PU_420(32, 24);
CHROMA_PU_420(24, 32);
CHROMA_PU_420(32, 8);
CHROMA_PU_420(8,
32);
p.chroma[X265_CSP_I420].pu[CHROMA_420_2x2].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x4].satd
= satd_4x4;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x8].satd
= satd8<8, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x16].satd = satd8<16, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].satd = satd8<32, 32>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_2x4].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].satd
= satd_8x4;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x8].satd
= satd4<4, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x8].satd
= satd8<16, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x16].satd
= satd8<8, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].satd = satd8<32, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x32].satd = satd8<16, 32>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x6].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_2x8].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x12].satd = satd4<16, 12>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].satd = satd4<12, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x4].satd
= satd4<16, 4>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x16].satd
= satd4<4, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].satd = satd8<32, 24>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].satd = satd8<24, 32>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x8].satd
= satd8<32, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x32].satd
= satd8<8, 32>;
#define CHROMA_CU_420(W, H) 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].sse_pp
= sse<W, H, pixel, pixel>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].copy_sp = blockcopy_sp_c<W, H>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].copy_ps = blockcopy_ps_c<W, H>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].copy_ss = blockcopy_ss_c<W, H>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].sub_ps = pixel_sub_ps_c<W, H>;

p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].add_ps = pixel_add_ps_c<W, H>;
CHROMA_CU_420(2, 2)
CHROMA_CU_420(4, 4)
CHROMA_CU_420(8, 8)
CHROMA_CU_420(16, 16)
CHROMA_CU_420(32, 32)
p.chroma[X265_CSP_I420].cu[BLOCK_8x8].sa8d
= p.chroma[X265_CSP_I420].pu[CHROMA_420_4x4].satd;
p.chroma[X265_CSP_I420].cu[BLOCK_16x16].sa8d = sa8d8<8, 8>;
p.chroma[X265_CSP_I420].cu[BLOCK_32x32].sa8d = sa8d16<16, 16>;
p.chroma[X265_CSP_I420].cu[BLOCK_64x64].sa8d = sa8d16<32, 32>;
#define CHROMA_PU_422(W, H) 
p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].addAvg
= addAvg<W, H>;

p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].copy_pp = blockcopy_pp_c<W, H>; 
CHROMA_PU_422(2, 4);
CHROMA_PU_422(4, 8);
CHROMA_PU_422(8, 16);
CHROMA_PU_422(16, 32);
CHROMA_PU_422(32, 64);
CHROMA_PU_422(4, 4);
CHROMA_PU_422(2, 8);
CHROMA_PU_422(8, 8);
CHROMA_PU_422(4, 16);
CHROMA_PU_422(8, 12);
CHROMA_PU_422(6, 16);
CHROMA_PU_422(8, 4);
CHROMA_PU_422(2, 16);
CHROMA_PU_422(16, 16);
CHROMA_PU_422(8, 32);
CHROMA_PU_422(16, 24);
CHROMA_PU_422(12, 32);
CHROMA_PU_422(16, 8);
CHROMA_PU_422(4,
32);
CHROMA_PU_422(32, 32);
CHROMA_PU_422(16, 64);
CHROMA_PU_422(32, 48);
CHROMA_PU_422(24, 64);
CHROMA_PU_422(32, 16);
CHROMA_PU_422(8,
64);
p.chroma[X265_CSP_I422].pu[CHROMA_422_2x4].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].satd
= satd4<4, 8>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x16].satd
= satd8<8, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].satd = satd8<16, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].satd = satd8<32, 64>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x4].satd
= satd_4x4;
p.chroma[X265_CSP_I422].pu[CHROMA_422_2x8].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].satd
= satd8<8, 8>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x16].satd
= satd4<4, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].satd = satd8<16, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].satd
= satd8<8, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].satd = satd8<32, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x64].satd = satd8<16, 64>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x12].satd
= satd4<8, 12>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].satd
= satd4<8, 4>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_2x16].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x24].satd = satd8<16, 24>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].satd = satd4<12, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x8].satd
= satd8<16, 8>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x32].satd
= satd4<4, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].satd = satd8<32, 48>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].satd = satd8<24, 64>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].satd = satd8<32, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].satd
= satd8<8, 64>;
#define CHROMA_CU_422(W, H) 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].sse_pp
= sse<W, H, pixel, pixel>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].copy_sp = blockcopy_sp_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].copy_ps = blockcopy_ps_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].copy_ss = blockcopy_ss_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].sub_ps = pixel_sub_ps_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].add_ps = pixel_add_ps_c<W, H>;
CHROMA_CU_422(2, 4)
CHROMA_CU_422(4, 8)
CHROMA_CU_422(8, 16)
CHROMA_CU_422(16, 32)
CHROMA_CU_422(32, 64)
p.chroma[X265_CSP_I422].cu[BLOCK_8x8].sa8d
= p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].satd;
p.chroma[X265_CSP_I422].cu[BLOCK_16x16].sa8d = sa8d8<8, 16>;
p.chroma[X265_CSP_I422].cu[BLOCK_32x32].sa8d = sa8d16<16, 32>;
p.chroma[X265_CSP_I422].cu[BLOCK_64x64].sa8d = sa8d16<32, 64>;
p.weight_pp = weight_pp_c;
p.weight_sp = weight_sp_c;
p.scale1D_128to64 = scale1D_128to64;
p.scale2D_64to32 = scale2D_64to32;
p.frameInitLowres = frame_init_lowres_core;
p.ssim_4x4x2_core = ssim_4x4x2_core;
p.ssim_end_4 = ssim_end_4;
p.planecopy_cp = planecopy_cp_c;
p.planecopy_sp = planecopy_sp_c;
p.propagateCost = estimateCUPropagateCost;
}
}

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
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
/*****************************************************************************
* Copyright (C) 2013 x265 project
*
* Authors: Steve Borho <steve@borho.org>
*
Mandar Gurav <mandar@multicorewareinc.com>
*
Mahesh Pittala <mahesh@multicorewareinc.com>
*
Min Chen <min.chen@multicorewareinc.com>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
02111, USA.
*
* This program is also available under a commercial proprietary license.
* For more information, contact us at license @ x265.com.
*****************************************************************************/
#include "common.h"
#include "primitives.h"
#include "x265.h"
#include <cstdlib> // abs()
using namespace x265;
namespace {
// place functions in anonymous namespace (file static)
/** 函数功能
： 计算SAD（8位）
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数 stride_pix1：计算块的步长
/*参数
pix2：参考块的首地址
/*参数 stride_pix2：参考块的步长
* 返回
：返回SAD值 */
template<int lx, int ly>
int sad(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
int sum = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
sum += abs(pix1[x] - pix2[x]);
pix1 += stride_pix1;
pix2 += stride_pix2;
}
return sum;
}
/** 函数功能
： 计算SAD（16位）
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数 stride_pix1：计算块的步长
/*参数
pix2：参考块的首地址
/*参数 stride_pix2：参考块的步长
* 返回
：返回SAD值 */
template<int lx, int ly>
int sad(const int16_t* pix1, intptr_t stride_pix1, const int16_t* pix2, intptr_t stride_pix2)
{
int sum = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
sum += abs(pix1[x] - pix2[x]);
pix1 += stride_pix1;
pix2 += stride_pix2;
}
return sum;
}
/** 函数功能
： 同时计算3个MV对应的3个SAD值
/*
调用范围
： ME中
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数
pix2：参考块的首地址
/*参数
pix3：参考块的首地址
/*参数
pix4：参考块的首地址
/*参数
frefstride：参考块的步长
/*参数
res：存储3个MV对应的3个SAD值
* 返回
：null */
template<int lx, int ly>
void sad_x3(const pixel* pix1, const pixel* pix2, const pixel* pix3, const pixel* pix4, intptr_t frefstride, int32_t* res)
{
res[0] = 0;
res[1] = 0;
res[2] = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
{
res[0] += abs(pix1[x] - pix2[x]);
res[1] += abs(pix1[x] - pix3[x]);
res[2] += abs(pix1[x] - pix4[x]);
}
pix1 += FENC_STRIDE;
//搜索块统一步长都为64
pix2 += frefstride;
pix3 += frefstride;
pix4 += frefstride;
}
}
/** 函数功能
： 同时计算4个MV对应的4个SAD值
/*
调用范围
： ME中
/*参数
lx：块的宽度
/*参数
ly：块的高度
/*参数
pix1：计算块的首地址
/*参数
pix2：参考块的首地址
/*参数
pix3：参考块的首地址
/*参数
pix4：参考块的首地址
/*参数
pix5：参考块的首地址
/*参数
frefstride：参考块的步长
/*参数
res：存储4个MV对应的4个SAD值
* 返回
：null */
template<int lx, int ly>
void sad_x4(const pixel* pix1, const pixel* pix2, const pixel* pix3, const pixel* pix4, const pixel* pix5, intptr_t frefstride, int32_t* res)
{
res[0] = 0;
res[1] = 0;
res[2] = 0;
res[3] = 0;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
{
res[0] += abs(pix1[x] - pix2[x]);
res[1] += abs(pix1[x] - pix3[x]);
res[2] += abs(pix1[x] - pix4[x]);
res[3] += abs(pix1[x] - pix5[x]);
}
pix1 += FENC_STRIDE;//搜索块统一步长都为64
pix2 += frefstride;
pix3 += frefstride;
pix4 += frefstride;
pix5 += frefstride;
}
}
template<int lx, int ly, class T1, class T2>
int sse(const T1* pix1, intptr_t stride_pix1, const T2* pix2, intptr_t stride_pix2)
{
int sum = 0;
int tmp;
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
{
tmp = pix1[x] - pix2[x];
sum += (tmp * tmp);
}
pix1 += stride_pix1;
pix2 += stride_pix2;
}
return sum;
}
#define BITS_PER_SUM (8 * sizeof(sum_t))
#define HADAMARD4(d0, d1, d2, d3, s0, s1, s2, s3) { 
sum2_t t0 = s0 + s1; 
sum2_t t1 = s0 - s1; 
sum2_t t2 = s2 + s3; 
sum2_t t3 = s2 - s3; 
d0 = t0 + t2; 
d2 = t0 - t2; 
d1 = t1 + t3; 
d3 = t1 - t3; 
}
// in: a pseudo-simd number of the form x+(y<<16)
// return: abs(x)+(abs(y)<<16)
inline sum2_t abs2(sum2_t a)
{
sum2_t s = ((a >> (BITS_PER_SUM - 1)) & (((sum2_t)1 << BITS_PER_SUM) + 1)) * ((sum_t)-1);
return (a + s) ^ s;
}
int satd_4x4(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
sum2_t tmp[4][2];
sum2_t a0, a1, a2, a3, b0, b1;
sum2_t sum = 0;
for (int i = 0; i < 4; i++, pix1 += stride_pix1, pix2 += stride_pix2)
{
a0 = pix1[0] - pix2[0];
a1 = pix1[1] - pix2[1];
b0 = (a0 + a1) + ((a0 - a1) << BITS_PER_SUM);
a2 = pix1[2] - pix2[2];
a3 = pix1[3] - pix2[3];
b1 = (a2 + a3) + ((a2 - a3) << BITS_PER_SUM);
tmp[i][0] = b0 + b1;
tmp[i][1] = b0 - b1;
}
for (int i = 0; i < 2; i++)
{
HADAMARD4(a0, a1, a2, a3, tmp[0][i], tmp[1][i], tmp[2][i], tmp[3][i]);
a0 = abs2(a0) + abs2(a1) + abs2(a2) + abs2(a3);
sum += ((sum_t)a0) + (a0 >> BITS_PER_SUM);
}
return (int)(sum >> 1);
}
static int satd_4x4(const int16_t* pix1, intptr_t stride_pix1)
{
int32_t tmp[4][4];
int32_t s01, s23, d01, d23;
int32_t satd = 0;
int d;
for (d = 0; d < 4; d++, pix1 += stride_pix1)
{
s01 = pix1[0] + pix1[1];
s23 = pix1[2] + pix1[3];
d01 = pix1[0] - pix1[1];
d23 = pix1[2] - pix1[3];
tmp[d][0] = s01 + s23;
tmp[d][1] = s01 - s23;
tmp[d][2] = d01 - d23;
tmp[d][3] = d01 + d23;
}
for (d = 0; d < 4; d++)
{
s01 = tmp[0][d] + tmp[1][d];
s23 = tmp[2][d] + tmp[3][d];
d01 = tmp[0][d] - tmp[1][d];
d23 = tmp[2][d] - tmp[3][d];
satd += abs(s01 + s23) + abs(s01 - s23) + abs(d01 - d23) + abs(d01 + d23);
}
return (int)(satd / 2);
}
// x264's SWAR version of satd 8x4, performs two 4x4 SATDs at once
int satd_8x4(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
sum2_t tmp[4][4];
sum2_t a0, a1, a2, a3;
sum2_t sum = 0;
for (int i = 0; i < 4; i++, pix1 += stride_pix1, pix2 += stride_pix2)
{
a0 = (pix1[0] - pix2[0]) + ((sum2_t)(pix1[4] - pix2[4]) << BITS_PER_SUM);
a1 = (pix1[1] - pix2[1]) + ((sum2_t)(pix1[5] - pix2[5]) << BITS_PER_SUM);
a2 = (pix1[2] - pix2[2]) + ((sum2_t)(pix1[6] - pix2[6]) << BITS_PER_SUM);
a3 = (pix1[3] - pix2[3]) + ((sum2_t)(pix1[7] - pix2[7]) << BITS_PER_SUM);
HADAMARD4(tmp[i][0], tmp[i][1], tmp[i][2], tmp[i][3], a0, a1, a2, a3);
}
for (int i = 0; i < 4; i++)
{
HADAMARD4(a0, a1, a2, a3, tmp[0][i], tmp[1][i], tmp[2][i], tmp[3][i]);
sum += abs2(a0) + abs2(a1) + abs2(a2) + abs2(a3);
}
return (((sum_t)sum) + (sum >> BITS_PER_SUM)) >> 1;
}
template<int w, int h>
// calculate satd in blocks of 4x4
int satd4(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
int satd = 0;
for (int row = 0; row < h; row += 4)
for (int col = 0; col < w; col += 4)
satd += satd_4x4(pix1 + row * stride_pix1 + col, stride_pix1,
pix2 + row * stride_pix2 + col, stride_pix2);
return satd;
}
template<int w, int h>
// calculate satd in blocks of 8x4
int satd8(const pixel* pix1, intptr_t stride_pix1, const pixel* pix2, intptr_t stride_pix2)
{
int satd = 0;
for (int row = 0; row < h; row += 4)
for (int col = 0; col < w; col += 8)
satd += satd_8x4(pix1 + row * stride_pix1 + col, stride_pix1,
pix2 + row * stride_pix2 + col, stride_pix2);
return satd;
}
inline int _sa8d_8x8(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
sum2_t tmp[8][4];
sum2_t a0, a1, a2, a3, a4, a5, a6, a7, b0, b1, b2, b3;
sum2_t sum = 0;
for (int i = 0; i < 8; i++, pix1 += i_pix1, pix2 += i_pix2)
{
a0 = pix1[0] - pix2[0];
a1 = pix1[1] - pix2[1];
b0 = (a0 + a1) + ((a0 - a1) << BITS_PER_SUM);
a2 = pix1[2] - pix2[2];
a3 = pix1[3] - pix2[3];
b1 = (a2 + a3) + ((a2 - a3) << BITS_PER_SUM);
a4 = pix1[4] - pix2[4];
a5 = pix1[5] - pix2[5];
b2 = (a4 + a5) + ((a4 - a5) << BITS_PER_SUM);
a6 = pix1[6] - pix2[6];
a7 = pix1[7] - pix2[7];
b3 = (a6 + a7) + ((a6 - a7) << BITS_PER_SUM);
HADAMARD4(tmp[i][0], tmp[i][1], tmp[i][2], tmp[i][3], b0, b1, b2, b3);
}
for (int i = 0; i < 4; i++)
{
HADAMARD4(a0, a1, a2, a3, tmp[0][i], tmp[1][i], tmp[2][i], tmp[3][i]);
HADAMARD4(a4, a5, a6, a7, tmp[4][i], tmp[5][i], tmp[6][i], tmp[7][i]);
b0
= abs2(a0 + a4) + abs2(a0 - a4);
b0 += abs2(a1 + a5) + abs2(a1 - a5);
b0 += abs2(a2 + a6) + abs2(a2 - a6);
b0 += abs2(a3 + a7) + abs2(a3 - a7);
sum += (sum_t)b0 + (b0 >> BITS_PER_SUM);
}
return (int)sum;
}
int sa8d_8x8(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
return (int)((_sa8d_8x8(pix1, i_pix1, pix2, i_pix2) + 2) >> 2);
}
inline int _sa8d_8x8(const int16_t* pix1, intptr_t i_pix1)
{
int32_t tmp[8][8];
int32_t a0, a1, a2, a3, a4, a5, a6, a7;
int32_t sum = 0;
for (int i = 0; i < 8; i++, pix1 += i_pix1)
{
a0 = pix1[0] + pix1[1];
a1 = pix1[2] + pix1[3];
a2 = pix1[4] + pix1[5];
a3 = pix1[6] + pix1[7];
a4 = pix1[0] - pix1[1];
a5 = pix1[2] - pix1[3];
a6 = pix1[4] - pix1[5];
a7 = pix1[6] - pix1[7];
tmp[i][0] = (a0 + a1) + (a2 + a3);
tmp[i][1] = (a0 + a1) - (a2 + a3);
tmp[i][2] = (a0 - a1) + (a2 - a3);
tmp[i][3] = (a0 - a1) - (a2 - a3);
tmp[i][4] = (a4 + a5) + (a6 + a7);
tmp[i][5] = (a4 + a5) - (a6 + a7);
tmp[i][6] = (a4 - a5) + (a6 - a7);
tmp[i][7] = (a4 - a5) - (a6 - a7);
}
for (int i = 0; i < 8; i++)
{
a0 = (tmp[0][i] + tmp[1][i]) + (tmp[2][i] + tmp[3][i]);
a2 = (tmp[0][i] + tmp[1][i]) - (tmp[2][i] + tmp[3][i]);
a1 = (tmp[0][i] - tmp[1][i]) + (tmp[2][i] - tmp[3][i]);
a3 = (tmp[0][i] - tmp[1][i]) - (tmp[2][i] - tmp[3][i]);
a4 = (tmp[4][i] + tmp[5][i]) + (tmp[6][i] + tmp[7][i]);
a6 = (tmp[4][i] + tmp[5][i]) - (tmp[6][i] + tmp[7][i]);
a5 = (tmp[4][i] - tmp[5][i]) + (tmp[6][i] - tmp[7][i]);
a7 = (tmp[4][i] - tmp[5][i]) - (tmp[6][i] - tmp[7][i]);
a0 = abs(a0 + a4) + abs(a0 - a4);
a0 += abs(a1 + a5) + abs(a1 - a5);
a0 += abs(a2 + a6) + abs(a2 - a6);
a0 += abs(a3 + a7) + abs(a3 - a7);
sum += a0;
}
return (int)sum;
}
int sa8d_8x8(const int16_t* pix1, intptr_t i_pix1)
{
return (int)((_sa8d_8x8(pix1, i_pix1) + 2) >> 2);
}
int sa8d_16x16(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
int sum = _sa8d_8x8(pix1, i_pix1, pix2, i_pix2)
+ _sa8d_8x8(pix1 + 8, i_pix1, pix2 + 8, i_pix2)
+ _sa8d_8x8(pix1 + 8 * i_pix1, i_pix1, pix2 + 8 * i_pix2, i_pix2)
+ _sa8d_8x8(pix1 + 8 + 8 * i_pix1, i_pix1, pix2 + 8 + 8 * i_pix2, i_pix2);
// This matches x264 sa8d_16x16, but is slightly different from HM's behavior because
// this version only rounds once at the end
return (sum + 2) >> 2;
}
template<int w, int h>
// Calculate sa8d in blocks of 8x8
int sa8d8(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
int cost = 0;
for (int y = 0; y < h; y += 8)
for (int x = 0; x < w; x += 8)
cost += sa8d_8x8(pix1 + i_pix1 * y + x, i_pix1, pix2 + i_pix2 * y + x, i_pix2);
return cost;
}
template<int w, int h>
// Calculate sa8d in blocks of 16x16
int sa8d16(const pixel* pix1, intptr_t i_pix1, const pixel* pix2, intptr_t i_pix2)
{
int cost = 0;
for (int y = 0; y < h; y += 16)
for (int x = 0; x < w; x += 16)
cost += sa8d_16x16(pix1 + i_pix1 * y + x, i_pix1, pix2 + i_pix2 * y + x, i_pix2);
return cost;
}
template<int size>
int pixel_ssd_s_c(const int16_t* a, intptr_t dstride)
{
int sum = 0;
for (int y = 0; y < size; y++)
{
for (int x = 0; x < size; x++)
sum += a[x] * a[x];
a += dstride;
}
return sum;
}
template<int size>
void blockfill_s_c(int16_t* dst, intptr_t dstride, int16_t val)
{
for (int y = 0; y < size; y++)
for (int x = 0; x < size; x++)
dst[y * dstride + x] = val;
}
template<int size>
void cpy2Dto1D_shl(int16_t* dst, const int16_t* src, intptr_t srcStride, int shift)
{
X265_CHECK(((intptr_t)dst & 15) == 0, "dst alignment errorn");
X265_CHECK((((intptr_t)src | (srcStride * sizeof(*src))) & 15) == 0 || size == 4, "src alignment errorn");
X265_CHECK(shift >= 0, "invalid shiftn");
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = src[j] << shift;
src += srcStride;
dst += size;
}
}
template<int size>
void cpy2Dto1D_shr(int16_t* dst, const int16_t* src, intptr_t srcStride, int shift)
{
X265_CHECK(((intptr_t)dst & 15) == 0, "dst alignment errorn");
X265_CHECK((((intptr_t)src | (srcStride * sizeof(*src))) & 15) == 0 || size == 4, "src alignment errorn");
X265_CHECK(shift > 0, "invalid shiftn");
int16_t round = 1 << (shift - 1);
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = (src[j] + round) >> shift;
src += srcStride;
dst += size;
}
}
template<int size>
void cpy1Dto2D_shl(int16_t* dst, const int16_t* src, intptr_t dstStride, int shift)
{
X265_CHECK((((intptr_t)dst | (dstStride * sizeof(*dst))) & 15) == 0 || size == 4, "dst alignment errorn");
X265_CHECK(((intptr_t)src & 15) == 0, "src alignment errorn");
X265_CHECK(shift >= 0, "invalid shiftn");
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = src[j] << shift;
src += size;
dst += dstStride;
}
}
template<int size>
void cpy1Dto2D_shr(int16_t* dst, const int16_t* src, intptr_t dstStride, int shift)
{
X265_CHECK((((intptr_t)dst | (dstStride * sizeof(*dst))) & 15) == 0 || size == 4, "dst alignment errorn");
X265_CHECK(((intptr_t)src & 15) == 0, "src alignment errorn");
X265_CHECK(shift > 0, "invalid shiftn");
int16_t round = 1 << (shift - 1);
for (int i = 0; i < size; i++)
{
for (int j = 0; j < size; j++)
dst[j] = (src[j] + round) >> shift;
src += size;
dst += dstStride;
}
}
template<int blockSize>
void getResidual(const pixel* fenc, const pixel* pred, int16_t* residual, intptr_t stride)
{
for (int y = 0; y < blockSize; y++)
{
for (int x = 0; x < blockSize; x++)
residual[x] = static_cast<int16_t>(fenc[x]) - static_cast<int16_t>(pred[x]);
fenc += stride;
residual += stride;
pred += stride;
}
}
template<int blockSize>
void transpose(pixel* dst, const pixel* src, intptr_t stride)
{
for (int k = 0; k < blockSize; k++)
for (int l = 0; l < blockSize; l++)
dst[k * blockSize + l] = src[l * stride + k];
}
void weight_sp_c(const int16_t* src, pixel* dst, intptr_t srcStride, intptr_t dstStride, int width, int height, int w0, int round, int shift, int offset)
{
int x, y;
#if CHECKED_BUILD || _DEBUG
const int correction = (IF_INTERNAL_PREC - X265_DEPTH);
X265_CHECK(!((w0 << 6) > 32767), "w0 using more than 16 bits, asm output will mismatchn");
X265_CHECK(!(round > 32767), "round using more than 16 bits, asm output will mismatchn");
X265_CHECK((shift >= correction), "shift must be include factor correction, please update ASM ABIn");
#endif
for (y = 0; y <= height - 1; y++)
{
for (x = 0; x <= width - 1; )
{
// note: width can be odd
dst[x] = x265_clip(((w0 * (src[x] + IF_INTERNAL_OFFS) + round) >> shift) + offset);
x++;
}
src += srcStride;
dst += dstStride;
}
}
/** 函数功能
： P帧加权参考帧获取
/*
调用范围
： 只在MotionReference::applyWeight、LookaheadTLD::weightCostLuma、LookaheadTLD::weightsAnalyse和weightCost函数中被调用
* 参数 src
： 原始P参考帧
* 参数 dst
： 加权后P参考帧
* 参数 stride
： 步长
* 参数 width
： 宽度
* 参数 height
： 高度
* 参数 w0
： 加权系数
* 参数 round
： 四舍五入操作
* 参数 shift
： 右移位数（前面为提高精度左移位，现在恢复原有精度）
* 参数 offset
： offset信息 整帧所有像素偏移值
* 返回
： null */
void weight_pp_c(const pixel* src, pixel* dst, intptr_t stride, int width, int height, int w0, int round, int shift, int offset)
{
int x, y;
const int correction = (IF_INTERNAL_PREC - X265_DEPTH);
X265_CHECK(!(width & 15), "weightp alignment errorn");
X265_CHECK(!((w0 << 6) > 32767), "w0 using more than 16 bits, asm output will mismatchn");
X265_CHECK(!(round > 32767), "round using more than 16 bits, asm output will mismatchn");
X265_CHECK((shift >= correction), "shift must be include factor correction, please update ASM ABIn");
X265_CHECK(!(round & ((1 << correction) - 1)), "round must be include factor correction, please update ASM ABIn");
for (y = 0; y <= height - 1; y++)
{
for (x = 0; x <= width - 1; )
{
// simulating pixel to short conversion
int16_t val = src[x] << correction;
dst[x] = x265_clip(((w0 * (val) + round) >> shift) + offset);
x++;
}
src += stride;
dst += stride;
}
}
/** 函数功能
： 获取两块的平均值
/*参数
lx：块的宽度
/*参数
ly：块的高度
* 参数 dst
：平均后像素存储位置
* 参数 dstride
：步长
* 参数 src0
：块0首地址
* 参数 sstride0
：块0步长
* 参数 src1
：块1首地址
* 参数 sstride1
：块1步长
* 返回
： null */
template<int lx, int ly>
void pixelavg_pp(pixel* dst, intptr_t dstride, const pixel* src0, intptr_t sstride0, const pixel* src1, intptr_t sstride1, int)//获取两块的平均值
{
for (int y = 0; y < ly; y++)
{
for (int x = 0; x < lx; x++)
dst[x] = (src0[x] + src1[x] + 1) >> 1;
src0 += sstride0;
src1 += sstride1;
dst += dstride;
}
}
void scale1D_128to64(pixel *dst, const pixel *src)
{
int x;
const pixel* src1 = src;
const pixel* src2 = src + 128;
pixel* dst1 = dst;
pixel* dst2 = dst + 64/*128*/;
for (x = 0; x < 128; x += 2)
{
// Top pixel
pixel pix0 = src1[(x + 0)];
pixel pix1 = src1[(x + 1)];
// Left pixel
pixel pix2 = src2[(x + 0)];
pixel pix3 = src2[(x + 1)];
int sum1 = pix0 + pix1;
int sum2 = pix2 + pix3;
dst1[x >> 1] = (pixel)((sum1 + 1) >> 1);
dst2[x >> 1] = (pixel)((sum2 + 1) >> 1);
}
}
void scale2D_64to32(pixel* dst, const pixel* src, intptr_t stride)
{
uint32_t x, y;
for (y = 0; y < 64; y += 2)
{
for (x = 0; x < 64; x += 2)
{
pixel pix0 = src[(y + 0) * stride + (x + 0)];
pixel pix1 = src[(y + 0) * stride + (x + 1)];
pixel pix2 = src[(y + 1) * stride + (x + 0)];
pixel pix3 = src[(y + 1) * stride + (x + 1)];
int sum = pix0 + pix1 + pix2 + pix3;
dst[y / 2 * 32 + x / 2] = (pixel)((sum + 2) >> 2);
}
}
}
/** 函数功能
：将原始帧视频亮度作1/2下采样
/*
调用范围
：只在Lowres::init函数中被调用
* 参数 src0
：原始视频帧亮度首地址:origPic->m_picOrg[0]
* 参数 dst0
：lowresPlane[0]
* 参数 dsth
：lowresPlane[1]
* 参数 dstv
：lowresPlane[2]
* 参数 dstc
：lowresPlane[3]
* 参数 src_stride：原始帧亮度步长
* 参数 dst_stride：下采样视频亮度步长
* 参数 width
：下采样视频的宽度
* 参数 heigh
：下采样视频的高度
* 返回值
： null
*/
void frame_init_lowres_core(const pixel* src0, pixel* dst0, pixel* dsth, pixel* dstv, pixel* dstc,
intptr_t src_stride, intptr_t dst_stride, int width, int height)
{
/* downscale and generate 4 hpel planes for lookahead */
/* 这样做的目的是更好的通过1/2下采样视频的编码估计原始视频编码状态
将亮度分四种方法进行1/2下采样
原始点：
82
89
86
86
93
85
89
99 101 113
96
97
97 100 104
106 108 107 111 109
127 156 133 139 137
0: 在行列选择偶数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
* * + +
* * + +
- - # #
- - # #
87
94
102 104
87 = ((((82 + 85 + 1) >> 1) + ((89 + 89 + 1) >> 1) + 1) >> 1)
94 = ((((86 + 99 + 1) >> 1) + ((86 + 101 + 1) >> 1) + 1) >> 1)
102= ((((96 + 106 + 1) >> 1) + ((97 + 108 + 1) >> 1) + 1) >> 1)
104 = ((((97 + 107 + 1) >> 1) + ((100 + 111 + 1) >> 1) + 1) >> 1)
h: 在行选择偶数像素点，在列选择奇数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
= * * = + +
= * * = + +
= - - = # #
= - - = # #
91
99
103 107
91 = ((((89 + 89 + 1) >> 1) + ((86 + 99 + 1) >> 1) + 1) >> 1)
99 = ((((86 + 101 + 1) >> 1) + ((93 + 113 + 1) >> 1) + 1) >> 1)
v: 在行选择奇数像素点，在列选择偶数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
= = = =
* * + +
* * + +
= = = =
- - # #
- - # #
92 100
125 123
92 = ((((85 + 96 + 1) >> 1) + ((89 + 97 + 1) >> 1) + 1) >> 1)
在行列选择奇数像素点为基准并选择相邻的右边、下边、右下机本身4个点作平均
= = = = = =
= * * = + +
= * * = + +
= = = = = =
= - - = # #
= - - = # #
96 105
126 124
96 = ((((89 + 97 + 1) >> 1) + ((99 + 97 + 1) >> 1) + 1) >> 1)
**/
for (int y = 0; y < height; y++)
{
const pixel* src1 = src0 + src_stride;
const pixel* src2 = src1 + src_stride;
for (int x = 0; x < width; x++)
{
// slower than naive bilinear, but matches asm
#define FILTER(a, b, c, d) ((((a + b + 1) >> 1) + ((c + d + 1) >> 1) + 1) >> 1)
dst0[x] = FILTER(src0[2 * x], src1[2 * x], src0[2 * x + 1], src1[2 * x + 1]);
dsth[x] = FILTER(src0[2 * x + 1], src1[2 * x + 1], src0[2 * x + 2], src1[2 * x + 2]);
dstv[x] = FILTER(src1[2 * x], src2[2 * x], src1[2 * x + 1], src2[2 * x + 1]);
dstc[x] = FILTER(src1[2 * x + 1], src2[2 * x + 1], src1[2 * x + 2], src2[2 * x + 2]);
#undef FILTER
}
src0 += src_stride * 2;
dst0 += dst_stride;
dsth += dst_stride;
dstv += dst_stride;
dstc += dst_stride;
}
}
/* structural similarity metric */
void ssim_4x4x2_core(const pixel* pix1, intptr_t stride1, const pixel* pix2, intptr_t stride2, int sums[2][4])
{
for (int z = 0; z < 2; z++)
{
uint32_t s1 = 0, s2 = 0, ss = 0, s12 = 0;
for (int y = 0; y < 4; y++)
{
for (int x = 0; x < 4; x++)
{
int a = pix1[x + y * stride1];
int b = pix2[x + y * stride2];
s1 += a;
s2 += b;
ss += a * a;
ss += b * b;
s12 += a * b;
}
}
sums[z][0] = s1;
sums[z][1] = s2;
sums[z][2] = ss;
sums[z][3] = s12;
pix1 += 4;
pix2 += 4;
}
}
float ssim_end_1(int s1, int s2, int ss, int s12)
{
/* Maximum value for 10-bit is: ss*64 = (2^10-1)^2*16*4*64 = 4286582784, which will overflow in some cases.
* s1*s1, s2*s2, and s1*s2 also obtain this value for edge cases: ((2^10-1)*16*4)^2 = 4286582784.
* Maximum value for 9-bit is: ss*64 = (2^9-1)^2*16*4*64 = 1069551616, which will not overflow. */
#define PIXEL_MAX ((1 << X265_DEPTH) - 1)
#if HIGH_BIT_DEPTH
X265_CHECK(X265_DEPTH == 10, "ssim invalid depthn");
#define type float
static const float ssim_c1 = (float)(.01 * .01 * PIXEL_MAX * PIXEL_MAX * 64);
static const float ssim_c2 = (float)(.03 * .03 * PIXEL_MAX * PIXEL_MAX * 64 * 63);
#else
X265_CHECK(X265_DEPTH == 8, "ssim invalid depthn");
#define type int
static const int ssim_c1 = (int)(.01 * .01 * PIXEL_MAX * PIXEL_MAX * 64 + .5);
static const int ssim_c2 = (int)(.03 * .03 * PIXEL_MAX * PIXEL_MAX * 64 * 63 + .5);
#endif
type fs1 = (type)s1;
type fs2 = (type)s2;
type fss = (type)ss;
type fs12 = (type)s12;
type vars = (type)(fss * 64 - fs1 * fs1 - fs2 * fs2);
type covar = (type)(fs12 * 64 - fs1 * fs2);
return (float)(2 * fs1 * fs2 + ssim_c1) * (float)(2 * covar + ssim_c2)
/ ((float)(fs1 * fs1 + fs2 * fs2 + ssim_c1) * (float)(vars + ssim_c2));
#undef type
#undef PIXEL_MAX
}
float ssim_end_4(int sum0[5][4], int sum1[5][4], int width)
{
float ssim = 0.0;
for (int i = 0; i < width; i++)
{
ssim += ssim_end_1(sum0[i][0] + sum0[i + 1][0] + sum1[i][0] + sum1[i + 1][0],
sum0[i][1] + sum0[i + 1][1] + sum1[i][1] + sum1[i + 1][1],
sum0[i][2] + sum0[i + 1][2] + sum1[i][2] + sum1[i + 1][2],
sum0[i][3] + sum0[i + 1][3] + sum1[i][3] + sum1[i + 1][3]);
}
return ssim;
}
/** 函数功能
：返回一个64位整数,低32位存储当前nxn所有元素的和，高32位存储当前nxn所有元素的平方和
/*
调用范围
：只在Lowres::init函数中被调用
* 参数 pix
：待计算的块
* 参数 i_stride：步长
* 返回值
：返回一个64位整数,低32位存储当前nxn所有元素的和，高32位存储当前nxn所有元素的平方和
*/
template<int size>
uint64_t pixel_var(const pixel* pix, intptr_t i_stride)
{
uint32_t sum = 0, sqr = 0;
for (int y = 0; y < size; y++)
{
for (int x = 0; x < size; x++)
{
sum += pix[x];
sqr += pix[x] * pix[x];
}
pix += i_stride;
}
return sum + ((uint64_t)sqr << 32);
}
#if defined(_MSC_VER)
#pragma warning(disable: 4127) // conditional expression is constant
#endif
template<int size>
int psyCost_pp(const pixel* source, intptr_t sstride, const pixel* recon, intptr_t rstride)
{
static pixel zeroBuf[8] /* = { 0 } */;
if (size)
{
int dim = 1 << (size + 2);
uint32_t totEnergy = 0;
for (int i = 0; i < dim; i += 8)
{
for (int j = 0; j < dim; j+= 8)
{
/* AC energy, measured by sa8d (AC + DC) minus SAD (DC) */
int sourceEnergy = sa8d_8x8(source + i * sstride + j, sstride, zeroBuf, 0) -
(sad<8, 8>(source + i * sstride + j, sstride, zeroBuf, 0) >> 2);
int reconEnergy =
sa8d_8x8(recon + i * rstride + j, rstride, zeroBuf, 0) -
(sad<8, 8>(recon + i * rstride + j, rstride, zeroBuf, 0) >> 2);
totEnergy += abs(sourceEnergy - reconEnergy);
}
}
return totEnergy;
}
else
{
/* 4x4 is too small for sa8d */
int sourceEnergy = satd_4x4(source, sstride, zeroBuf, 0) - (sad<4, 4>(source, sstride, zeroBuf, 0) >> 2);
int reconEnergy = satd_4x4(recon, rstride, zeroBuf, 0) - (sad<4, 4>(recon, rstride, zeroBuf, 0) >> 2);
return abs(sourceEnergy - reconEnergy);
}
}
template<int size>
int psyCost_ss(const int16_t* source, intptr_t sstride, const int16_t* recon, intptr_t rstride)
{
static int16_t zeroBuf[8] /* = { 0 } */;
if (size)
{
int dim = 1 << (size + 2);
uint32_t totEnergy = 0;
for (int i = 0; i < dim; i += 8)
{
for (int j = 0; j < dim; j+= 8)
{
/* AC energy, measured by sa8d (AC + DC) minus SAD (DC) */
int sourceEnergy = sa8d_8x8(source + i * sstride + j, sstride) -
(sad<8, 8>(source + i * sstride + j, sstride, zeroBuf, 0) >> 2);
int reconEnergy =
sa8d_8x8(recon + i * rstride + j, rstride) -
(sad<8, 8>(recon + i * rstride + j, rstride, zeroBuf, 0) >> 2);
totEnergy += abs(sourceEnergy - reconEnergy);
}
}
return totEnergy;
}
else
{
/* 4x4 is too small for sa8d */
int sourceEnergy = satd_4x4(source, sstride) - (sad<4, 4>(source, sstride, zeroBuf, 0) >> 2);
int reconEnergy = satd_4x4(recon, rstride) - (sad<4, 4>(recon, rstride, zeroBuf, 0) >> 2);
return abs(sourceEnergy - reconEnergy);
}
}
template<int bx, int by>
void blockcopy_pp_c(pixel* a, intptr_t stridea, const pixel* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = b[x];
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void blockcopy_ss_c(int16_t* a, intptr_t stridea, const int16_t* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = b[x];
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void blockcopy_sp_c(pixel* a, intptr_t stridea, const int16_t* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
{
X265_CHECK((b[x] >= 0) && (b[x] <= ((1 << X265_DEPTH) - 1)), "blockcopy pixel size failn");
a[x] = (pixel)b[x];
}
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void blockcopy_ps_c(int16_t* a, intptr_t stridea, const pixel* b, intptr_t strideb)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = (int16_t)b[x];
a += stridea;
b += strideb;
}
}
template<int bx, int by>
void pixel_sub_ps_c(int16_t* a, intptr_t dstride, const pixel* b0, const pixel* b1, intptr_t sstride0, intptr_t sstride1)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = (int16_t)(b0[x] - b1[x]);
b0 += sstride0;
b1 += sstride1;
a += dstride;
}
}
template<int bx, int by>
void pixel_add_ps_c(pixel* a, intptr_t dstride, const pixel* b0, const int16_t* b1, intptr_t sstride0, intptr_t sstride1)
{
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x++)
a[x] = x265_clip(b0[x] + b1[x]);
b0 += sstride0;
b1 += sstride1;
a += dstride;
}
}
template<int bx, int by>
void addAvg(const int16_t* src0, const int16_t* src1, pixel* dst, intptr_t src0Stride, intptr_t src1Stride, intptr_t dstStride)
{
int shiftNum, offset;
shiftNum = IF_INTERNAL_PREC + 1 - X265_DEPTH;
offset = (1 << (shiftNum - 1)) + 2 * IF_INTERNAL_OFFS;
for (int y = 0; y < by; y++)
{
for (int x = 0; x < bx; x += 2)
{
dst[x + 0] = x265_clip((src0[x + 0] + src1[x + 0] + offset) >> shiftNum);
dst[x + 1] = x265_clip((src0[x + 1] + src1[x + 1] + offset) >> shiftNum);
}
src0 += src0Stride;
src1 += src1Stride;
dst
+= dstStride;
}
}
void planecopy_cp_c(const uint8_t* src, intptr_t srcStride, pixel* dst, intptr_t dstStride, int width, int height, int shift)
{
for (int r = 0; r < height; r++)
{
for (int c = 0; c < width; c++)
dst[c] = ((pixel)src[c]) << shift;
dst += dstStride;
src += srcStride;
}
}
void planecopy_sp_c(const uint16_t* src, intptr_t srcStride, pixel* dst, intptr_t dstStride, int width, int height, int shift, uint16_t mask)
{
for (int r = 0; r < height; r++)
{
for (int c = 0; c < width; c++)
dst[c] = (pixel)((src[c] >> shift) & mask);
dst += dstStride;
src += srcStride;
}
}
/** 函数功能
：计算每行的8x8的传播cost(累加传播cost + 加权intracost)*（intracost-最优cost）/intracost
/*
调用范围
：只在Lookahead::estimateCUPropagate函数中被调用
/*参数
dst: 存储当前行每个8x8的传播cost
/*参数 propagateIn：当前行的传播cost存储首地址
/*参数
intraCosts：当前行的intracost
/*参数
interCosts：当前行的intercost
/*参数
invQscales：当前行的AQ offsets
/*参数
fpsFactor：当前帧的帧率因子（一般为1.0）
/*参数
len：当前行的长度
* 返回
：返回SAD值 */
/* Estimate the total amount of influence on future quality that could be had if we
* were to improve the reference samples used to inter predict any given CU. */
void estimateCUPropagateCost(int* dst, const uint16_t* propagateIn, const int32_t* intraCosts, const uint16_t* interCosts,
const int32_t* invQscales, const double* fpsFactor, int len)
{
double fps = *fpsFactor / 256;
//因为invQscales存储扩大256倍的整数数据，这里除以256目的是缩放数据
for (int i = 0; i < len; i++)
{
double intraCost
= intraCosts[i] * invQscales[i]; //获取当前8x8块 加权intra cost
double propagateAmount = (double)propagateIn[i] + intraCost * fps; //累加传播cost 加上加权fps因子之后的intra cost
double propagateNum
= (double)intraCosts[i] - (interCosts[i] & ((1 << 14) - 1));//获取最优cost与intracost的差值
高14位的数字：0 表示 intra
1 表示前向搜索
2表示后向搜索
3 表示bi搜索
double propagateDenom
= (double)intraCosts[i];
//当前8x8块 intra cost
dst[i] = (int)(propagateAmount * propagateNum / propagateDenom + 0.5);//(累加传播cost + 加权intracost)*（intracost-最优cost）/intracost
}
}
}
// end anonymous namespace
namespace x265 {
// x265 private namespace
/* Extend the edges of a picture so that it may safely be used for motion
* compensation. This function assumes the picture is stored in a buffer with
* sufficient padding for the X and Y margins */
/** 函数功能
：将视频帧进行扩边，便于插值和ME搜索
/*
调用范围
：只在weightAnalyse和Lowres::init函数中被调用
* 参数 pic
：需要进行插值的视频帧数据
* 参数 stride
：视频帧步长
* 参数 width
：视频帧宽度
* 参数 height
：视频帧高度
* 参数 marginX ：两边需要扩边的宽度
* 参数 marginY ：上下需要扩边的高度
* 返回值
： null
*/
void extendPicBorder(pixel* pic, intptr_t stride, int width, int height, int marginX, int marginY)
{
/* extend left and right margins */
primitives.extendRowBorder(pic, stride, width, height, marginX); //asm 代码，快速实现一行扩边
/* copy top row to create above margin */
pixel* top = pic - marginX;
for (int y = 0; y < marginY; y++)
memcpy(top - (y + 1) * stride, top, stride * sizeof(pixel));
/* copy bottom row to create below margin */
pixel* bot = pic - marginX + (height - 1) * stride;
for (int y = 0; y < marginY; y++)
memcpy(bot + (y + 1) * stride, bot, stride * sizeof(pixel));
}
/* Initialize entries for pixel functions defined in this file */
void setupPixelPrimitives_c(EncoderPrimitives &p)
{
#define LUMA_PU(W, H) 
p.pu[LUMA_ ## W ## x ## H].copy_pp = blockcopy_pp_c<W, H>; 
p.pu[LUMA_ ## W ## x ## H].addAvg = addAvg<W, H>; 
p.pu[LUMA_ ## W ## x ## H].sad = sad<W, H>; 
p.pu[LUMA_ ## W ## x ## H].sad_x3 = sad_x3<W, H>; 
p.pu[LUMA_ ## W ## x ## H].sad_x4 = sad_x4<W, H>; 
p.pu[LUMA_ ## W ## x ## H].pixelavg_pp = pixelavg_pp<W, H>;
#define LUMA_CU(W, H) 
p.cu[BLOCK_ ## W ## x ## H].sub_ps
= pixel_sub_ps_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].add_ps
= pixel_add_ps_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].copy_sp
= blockcopy_sp_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].copy_ps
= blockcopy_ps_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].copy_ss
= blockcopy_ss_c<W, H>; 
p.cu[BLOCK_ ## W ## x ## H].blockfill_s
= blockfill_s_c<W>;

p.cu[BLOCK_ ## W ## x ## H].cpy2Dto1D_shl = cpy2Dto1D_shl<W>; 
p.cu[BLOCK_ ## W ## x ## H].cpy2Dto1D_shr = cpy2Dto1D_shr<W>; 
p.cu[BLOCK_ ## W ## x ## H].cpy1Dto2D_shl = cpy1Dto2D_shl<W>; 
p.cu[BLOCK_ ## W ## x ## H].cpy1Dto2D_shr = cpy1Dto2D_shr<W>; 
p.cu[BLOCK_ ## W ## x ## H].psy_cost_pp
= psyCost_pp<BLOCK_ ## W ## x ## H>; 
p.cu[BLOCK_ ## W ## x ## H].psy_cost_ss
= psyCost_ss<BLOCK_ ## W ## x ## H>; 
p.cu[BLOCK_ ## W ## x ## H].transpose
= transpose<W>; 
p.cu[BLOCK_ ## W ## x ## H].ssd_s
= pixel_ssd_s_c<W>; 
p.cu[BLOCK_ ## W ## x ## H].var
= pixel_var<W>; 
p.cu[BLOCK_ ## W ## x ## H].calcresidual
= getResidual<W>; 
p.cu[BLOCK_ ## W ## x ## H].sse_pp
= sse<W, H, pixel, pixel>; 
p.cu[BLOCK_ ## W ## x ## H].sse_ss
= sse<W, H, int16_t, int16_t>;
LUMA_PU(4, 4);
LUMA_PU(8, 8);
LUMA_PU(16, 16);
LUMA_PU(32, 32);
LUMA_PU(64, 64);
LUMA_PU(4, 8);
LUMA_PU(8, 4);
LUMA_PU(16,
8);
LUMA_PU(8, 16);
LUMA_PU(16, 12);
LUMA_PU(12, 16);
LUMA_PU(16,
4);
LUMA_PU(4, 16);
LUMA_PU(32, 16);
LUMA_PU(16, 32);
LUMA_PU(32, 24);
LUMA_PU(24, 32);
LUMA_PU(32,
8);
LUMA_PU(8, 32);
LUMA_PU(64, 32);
LUMA_PU(32, 64);
LUMA_PU(64, 48);
LUMA_PU(48, 64);
LUMA_PU(64, 16);
LUMA_PU(16, 64);
p.pu[LUMA_4x4].satd
= satd_4x4;
p.pu[LUMA_8x8].satd
= satd8<8, 8>;
p.pu[LUMA_8x4].satd
= satd_8x4;
p.pu[LUMA_4x8].satd
= satd4<4, 8>;
p.pu[LUMA_16x16].satd = satd8<16, 16>;
p.pu[LUMA_16x8].satd
= satd8<16, 8>;
p.pu[LUMA_8x16].satd
= satd8<8, 16>;
p.pu[LUMA_16x12].satd = satd8<16, 12>;
p.pu[LUMA_12x16].satd = satd4<12, 16>;
p.pu[LUMA_16x4].satd
= satd8<16, 4>;
p.pu[LUMA_4x16].satd
= satd4<4, 16>;
p.pu[LUMA_32x32].satd = satd8<32, 32>;
p.pu[LUMA_32x16].satd = satd8<32, 16>;
p.pu[LUMA_16x32].satd = satd8<16, 32>;
p.pu[LUMA_32x24].satd = satd8<32, 24>;
p.pu[LUMA_24x32].satd = satd8<24, 32>;
p.pu[LUMA_32x8].satd
= satd8<32, 8>;
p.pu[LUMA_8x32].satd
= satd8<8, 32>;
p.pu[LUMA_64x64].satd = satd8<64, 64>;
p.pu[LUMA_64x32].satd = satd8<64, 32>;
p.pu[LUMA_32x64].satd = satd8<32, 64>;
p.pu[LUMA_64x48].satd = satd8<64, 48>;
p.pu[LUMA_48x64].satd = satd8<48, 64>;
p.pu[LUMA_64x16].satd = satd8<64, 16>;
p.pu[LUMA_16x64].satd = satd8<16, 64>;
LUMA_CU(4, 4);
LUMA_CU(8, 8);
LUMA_CU(16, 16);
LUMA_CU(32, 32);
LUMA_CU(64, 64);
p.cu[BLOCK_4x4].sa8d
= satd_4x4;
p.cu[BLOCK_8x8].sa8d
= sa8d_8x8;
p.cu[BLOCK_16x16].sa8d = sa8d_16x16;
p.cu[BLOCK_32x32].sa8d = sa8d16<32, 32>;
p.cu[BLOCK_64x64].sa8d = sa8d16<64, 64>;
#define CHROMA_PU_420(W, H) 
p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].addAvg
= addAvg<W, H>;

p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].copy_pp = blockcopy_pp_c<W, H>; 
CHROMA_PU_420(2, 2);
CHROMA_PU_420(2, 4);
CHROMA_PU_420(4, 4);
CHROMA_PU_420(8, 8);
CHROMA_PU_420(16, 16);
CHROMA_PU_420(32, 32);
CHROMA_PU_420(4, 2);
CHROMA_PU_420(8, 4);
CHROMA_PU_420(4, 8);
CHROMA_PU_420(8, 6);
CHROMA_PU_420(6, 8);
CHROMA_PU_420(8, 2);
CHROMA_PU_420(2, 8);
CHROMA_PU_420(16, 8);
CHROMA_PU_420(8,
16);
CHROMA_PU_420(16, 12);
CHROMA_PU_420(12, 16);
CHROMA_PU_420(16, 4);
CHROMA_PU_420(4,
16);
CHROMA_PU_420(32, 16);
CHROMA_PU_420(16, 32);
CHROMA_PU_420(32, 24);
CHROMA_PU_420(24, 32);
CHROMA_PU_420(32, 8);
CHROMA_PU_420(8,
32);
p.chroma[X265_CSP_I420].pu[CHROMA_420_2x2].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x4].satd
= satd_4x4;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x8].satd
= satd8<8, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x16].satd = satd8<16, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].satd = satd8<32, 32>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_2x4].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].satd
= satd_8x4;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x8].satd
= satd4<4, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x8].satd
= satd8<16, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x16].satd
= satd8<8, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].satd = satd8<32, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x32].satd = satd8<16, 32>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x6].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_2x8].satd
= NULL;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x12].satd = satd4<16, 12>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].satd = satd4<12, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_16x4].satd
= satd4<16, 4>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_4x16].satd
= satd4<4, 16>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].satd = satd8<32, 24>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].satd = satd8<24, 32>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_32x8].satd
= satd8<32, 8>;
p.chroma[X265_CSP_I420].pu[CHROMA_420_8x32].satd
= satd8<8, 32>;
#define CHROMA_CU_420(W, H) 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].sse_pp
= sse<W, H, pixel, pixel>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].copy_sp = blockcopy_sp_c<W, H>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].copy_ps = blockcopy_ps_c<W, H>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].copy_ss = blockcopy_ss_c<W, H>; 
p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].sub_ps = pixel_sub_ps_c<W, H>;

p.chroma[X265_CSP_I420].cu[BLOCK_420_ ## W ## x ## H].add_ps = pixel_add_ps_c<W, H>;
CHROMA_CU_420(2, 2)
CHROMA_CU_420(4, 4)
CHROMA_CU_420(8, 8)
CHROMA_CU_420(16, 16)
CHROMA_CU_420(32, 32)
p.chroma[X265_CSP_I420].cu[BLOCK_8x8].sa8d
= p.chroma[X265_CSP_I420].pu[CHROMA_420_4x4].satd;
p.chroma[X265_CSP_I420].cu[BLOCK_16x16].sa8d = sa8d8<8, 8>;
p.chroma[X265_CSP_I420].cu[BLOCK_32x32].sa8d = sa8d16<16, 16>;
p.chroma[X265_CSP_I420].cu[BLOCK_64x64].sa8d = sa8d16<32, 32>;
#define CHROMA_PU_422(W, H) 
p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].addAvg
= addAvg<W, H>;

p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].copy_pp = blockcopy_pp_c<W, H>; 
CHROMA_PU_422(2, 4);
CHROMA_PU_422(4, 8);
CHROMA_PU_422(8, 16);
CHROMA_PU_422(16, 32);
CHROMA_PU_422(32, 64);
CHROMA_PU_422(4, 4);
CHROMA_PU_422(2, 8);
CHROMA_PU_422(8, 8);
CHROMA_PU_422(4, 16);
CHROMA_PU_422(8, 12);
CHROMA_PU_422(6, 16);
CHROMA_PU_422(8, 4);
CHROMA_PU_422(2, 16);
CHROMA_PU_422(16, 16);
CHROMA_PU_422(8, 32);
CHROMA_PU_422(16, 24);
CHROMA_PU_422(12, 32);
CHROMA_PU_422(16, 8);
CHROMA_PU_422(4,
32);
CHROMA_PU_422(32, 32);
CHROMA_PU_422(16, 64);
CHROMA_PU_422(32, 48);
CHROMA_PU_422(24, 64);
CHROMA_PU_422(32, 16);
CHROMA_PU_422(8,
64);
p.chroma[X265_CSP_I422].pu[CHROMA_422_2x4].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].satd
= satd4<4, 8>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x16].satd
= satd8<8, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].satd = satd8<16, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].satd = satd8<32, 64>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x4].satd
= satd_4x4;
p.chroma[X265_CSP_I422].pu[CHROMA_422_2x8].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].satd
= satd8<8, 8>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x16].satd
= satd4<4, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].satd = satd8<16, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].satd
= satd8<8, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].satd = satd8<32, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x64].satd = satd8<16, 64>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x12].satd
= satd4<8, 12>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].satd
= satd4<8, 4>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_2x16].satd
= NULL;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x24].satd = satd8<16, 24>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].satd = satd4<12, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_16x8].satd
= satd8<16, 8>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_4x32].satd
= satd4<4, 32>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].satd = satd8<32, 48>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].satd = satd8<24, 64>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].satd = satd8<32, 16>;
p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].satd
= satd8<8, 64>;
#define CHROMA_CU_422(W, H) 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].sse_pp
= sse<W, H, pixel, pixel>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].copy_sp = blockcopy_sp_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].copy_ps = blockcopy_ps_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].copy_ss = blockcopy_ss_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].sub_ps = pixel_sub_ps_c<W, H>; 
p.chroma[X265_CSP_I422].cu[BLOCK_422_ ## W ## x ## H].add_ps = pixel_add_ps_c<W, H>;
CHROMA_CU_422(2, 4)
CHROMA_CU_422(4, 8)
CHROMA_CU_422(8, 16)
CHROMA_CU_422(16, 32)
CHROMA_CU_422(32, 64)
p.chroma[X265_CSP_I422].cu[BLOCK_8x8].sa8d
= p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].satd;
p.chroma[X265_CSP_I422].cu[BLOCK_16x16].sa8d = sa8d8<8, 16>;
p.chroma[X265_CSP_I422].cu[BLOCK_32x32].sa8d = sa8d16<16, 32>;
p.chroma[X265_CSP_I422].cu[BLOCK_64x64].sa8d = sa8d16<32, 64>;
p.weight_pp = weight_pp_c;
p.weight_sp = weight_sp_c;
p.scale1D_128to64 = scale1D_128to64;
p.scale2D_64to32 = scale2D_64to32;
p.frameInitLowres = frame_init_lowres_core;
p.ssim_4x4x2_core = ssim_4x4x2_core;
p.ssim_end_4 = ssim_end_4;
p.planecopy_cp = planecopy_cp_c;
p.planecopy_sp = planecopy_sp_c;
p.propagateCost = estimateCUPropagateCost;
}
}