liz/rawTherapee
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

SSE code for the most expensive loop. Also small cleanups

Browse Source
This commit is contained in:
Ingo Weyrich
2020-09-03 22:04:11 +02:00
parent 390d62f638
commit 33c5d6ab3e
1 changed files with 62 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

97
rtengine/rcd_demosaic.cc
View File

@@ -1,7 +1,7 @@
/*
* This file is part of RawTherapee.
*
* Copyright (c) 2017-2018 Luis Sanz Rodriguez (luis.sanz.rodriguez(at)gmail(dot)com) and Ingo Weyrich (heckflosse67@gmx.de)
* Copyright (c) 2017-2020 Luis Sanz Rodriguez (luis.sanz.rodriguez(at)gmail(dot)com) and Ingo Weyrich (heckflosse67@gmx.de)
*
* RawTherapee is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
@@ -82,8 +82,8 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
const int numTw = W / (tileSizeN) + ((W % (tileSizeN)) ? 1 : 0);
constexpr int w1 = tileSize, w2 = 2 * tileSize, w3 = 3 * tileSize, w4 = 4 * tileSize;
//Tolerance to avoid dividing by zero
static constexpr float eps = 1e-5f;
static constexpr float epssq = 1e-10f;
constexpr float eps = 1e-5f;
constexpr float epssq = 1e-10f;
#ifdef _OPENMP
#pragma omp parallel
@@ -99,16 +99,16 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
#ifdef _OPENMP
#pragma omp for schedule(dynamic, chunkSize) collapse(2) nowait
#endif
for(int tr = 0; tr < numTh; ++tr) {
for(int tc = 0; tc < numTw; ++tc) {
for (int tr = 0; tr < numTh; ++tr) {
for (int tc = 0; tc < numTw; ++tc) {
const int rowStart = tr * tileSizeN;
const int rowEnd = std::min(rowStart + tileSize, H);
if(rowStart + rcdBorder == rowEnd - rcdBorder) {
if (rowStart + rcdBorder == rowEnd - rcdBorder) {
continue;
}
const int colStart = tc * tileSizeN;
const int colEnd = std::min(colStart + tileSize, W);
if(colStart + rcdBorder == colEnd - rcdBorder) {
if (colStart + rcdBorder == colEnd - rcdBorder) {
continue;
}
@@ -125,7 +125,7 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
cfa[indx] = rgb[c0][indx] = LIM01(rawData[row][col] / 65535.f);
cfa[indx + 1] = rgb[c1][indx + 1] = LIM01(rawData[row][col + 1] / 65535.f);
}
if(col < colEnd) {
if (col < colEnd) {
cfa[indx] = rgb[c0][indx] = LIM01(rawData[row][col] / 65535.f);
}
}
@@ -160,36 +160,68 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
* STEP 3: Populate the green channel
*/
// Step 3.1: Populate the green channel at blue and red CFA positions
for (int row = 4; row < tileRows - 4; row++) {
for (int col = 4 + (fc(cfarray, row, 0) & 1), indx = row * tileSize + col; col < tilecols - 4; col += 2, indx += 2) {
for (int row = 4; row < tileRows - 4; row++) {
int col = 4 + (fc(cfarray, row, 0) & 1);
int indx = row * tileSize + col;
// Refined vertical and horizontal local discrimination
float VH_Central_Value = VH_Dir[indx];
float VH_Neighbourhood_Value = 0.25f * ((VH_Dir[indx - w1 - 1] + VH_Dir[indx - w1 + 1]) + (VH_Dir[indx + w1 - 1] + VH_Dir[indx + w1 + 1]));
float VH_Disc = std::fabs(0.5f - VH_Central_Value) < std::fabs(0.5f - VH_Neighbourhood_Value) ? VH_Neighbourhood_Value : VH_Central_Value;
// Cardinal gradients
float N_Grad = eps + std::fabs(cfa[indx - w1] - cfa[indx + w1]) + std::fabs(cfa[indx] - cfa[indx - w2]) + std::fabs(cfa[indx - w1] - cfa[indx - w3]) + std::fabs(cfa[indx - w2] - cfa[indx - w4]);
float S_Grad = eps + std::fabs(cfa[indx - w1] - cfa[indx + w1]) + std::fabs(cfa[indx] - cfa[indx + w2]) + std::fabs(cfa[indx + w1] - cfa[indx + w3]) + std::fabs(cfa[indx + w2] - cfa[indx + w4]);
float W_Grad = eps + std::fabs(cfa[indx - 1] - cfa[indx + 1]) + std::fabs(cfa[indx] - cfa[indx - 2]) + std::fabs(cfa[indx - 1] - cfa[indx - 3]) + std::fabs(cfa[indx - 2] - cfa[indx - 4]);
float E_Grad = eps + std::fabs(cfa[indx - 1] - cfa[indx + 1]) + std::fabs(cfa[indx] - cfa[indx + 2]) + std::fabs(cfa[indx + 1] - cfa[indx + 3]) + std::fabs(cfa[indx + 2] - cfa[indx + 4]);
#ifdef __SSE2__
for (; col < tilecols - 7; col += 8, indx += 8) {
// Cardinal gradients
const vfloat cfai = LC2VFU(cfa[indx]);
const vfloat N_Grad = eps + (vabsf(LC2VFU(cfa[indx - w1]) - LC2VFU(cfa[indx + w1])) + vabsf(cfai - LC2VFU(cfa[indx - w2]))) + (vabsf(LC2VFU(cfa[indx - w1]) - LC2VFU(cfa[indx - w3])) + vabsf(LC2VFU(cfa[indx - w2]) - LC2VFU(cfa[indx - w4])));
const vfloat S_Grad = eps + (vabsf(LC2VFU(cfa[indx - w1]) - LC2VFU(cfa[indx + w1])) + vabsf(cfai - LC2VFU(cfa[indx + w2]))) + (vabsf(LC2VFU(cfa[indx + w1]) - LC2VFU(cfa[indx + w3])) + vabsf(LC2VFU(cfa[indx + w2]) - LC2VFU(cfa[indx + w4])));
const vfloat W_Grad = eps + (vabsf(LC2VFU(cfa[indx - 1]) - LC2VFU(cfa[indx + 1])) + vabsf(cfai - LC2VFU(cfa[indx - 2]))) + (vabsf(LC2VFU(cfa[indx - 1]) - LC2VFU(cfa[indx - 3])) + vabsf(LC2VFU(cfa[indx - 2]) - LC2VFU(cfa[indx - 4])));
const vfloat E_Grad = eps + (vabsf(LC2VFU(cfa[indx - 1]) - LC2VFU(cfa[indx + 1])) + vabsf(cfai - LC2VFU(cfa[indx + 2]))) + (vabsf(LC2VFU(cfa[indx + 1]) - LC2VFU(cfa[indx + 3])) + vabsf(LC2VFU(cfa[indx + 2]) - LC2VFU(cfa[indx + 4])));
// Cardinal pixel estimations
float N_Est = cfa[indx - w1] * (1.f + (lpf[indx>>1] - lpf[(indx - w2)>>1]) / (eps + lpf[indx>>1] + lpf[(indx - w2)>>1]));
float S_Est = cfa[indx + w1] * (1.f + (lpf[indx>>1] - lpf[(indx + w2)>>1]) / (eps + lpf[indx>>1] + lpf[(indx + w2)>>1]));
float W_Est = cfa[indx - 1] * (1.f + (lpf[indx>>1] - lpf[(indx - 2)>>1]) / (eps + lpf[indx>>1] + lpf[(indx - 2)>>1]));
float E_Est = cfa[indx + 1] * (1.f + (lpf[indx>>1] - lpf[(indx + 2)>>1]) / (eps + lpf[indx>>1] + lpf[(indx + 2)>>1]));
const vfloat lpfi = LVFU(lpf[indx>>1]);
const vfloat N_Est = LC2VFU(cfa[indx - w1]) + (LC2VFU(cfa[indx - w1]) * (lpfi - LVFU(lpf[(indx - w2)>>1])) / (eps + lpfi + LVFU(lpf[(indx - w2)>>1])));
const vfloat S_Est = LC2VFU(cfa[indx + w1]) + (LC2VFU(cfa[indx + w1]) * (lpfi - LVFU(lpf[(indx + w2)>>1])) / (eps + lpfi + LVFU(lpf[(indx + w2)>>1])));
const vfloat W_Est = LC2VFU(cfa[indx - 1]) + (LC2VFU(cfa[indx - 1]) * (lpfi - LVFU(lpf[(indx - 2)>>1])) / (eps + lpfi + LVFU(lpf[(indx - 2)>>1])));
const vfloat E_Est = LC2VFU(cfa[indx + 1]) + (LC2VFU(cfa[indx + 1]) * (lpfi - LVFU(lpf[(indx + 2)>>1])) / (eps + lpfi + LVFU(lpf[(indx + 2)>>1])));
// Vertical and horizontal estimations
float V_Est = (S_Grad * N_Est + N_Grad * S_Est) / (N_Grad + S_Grad);
float H_Est = (W_Grad * E_Est + E_Grad * W_Est) / (E_Grad + W_Grad);
const vfloat V_Est = (S_Grad * N_Est + N_Grad * S_Est) / (N_Grad + S_Grad);
const vfloat H_Est = (W_Grad * E_Est + E_Grad * W_Est) / (E_Grad + W_Grad);
// G@B and G@R interpolation
rgb[1][indx] = VH_Disc * H_Est + (1.f - VH_Disc) * V_Est;
// Refined vertical and horizontal local discrimination
const vfloat VH_Central_Value = LC2VFU(VH_Dir[indx]);
const vfloat VH_Neighbourhood_Value = 0.25f * ((LC2VFU(VH_Dir[indx - w1 - 1]) + LC2VFU(VH_Dir[indx - w1 + 1])) + (LC2VFU(VH_Dir[indx + w1 - 1]) + LC2VFU(VH_Dir[indx + w1 + 1])));
const vfloat VH_Disc = vabsf(0.5f - VH_Central_Value) < vabsf(0.5f - VH_Neighbourhood_Value) ? VH_Neighbourhood_Value : VH_Central_Value;
STC2VFU(rgb[1][indx], vintpf(VH_Disc, H_Est, V_Est));
}
#endif
for (; col < tilecols - 4; col += 2, indx += 2) {
// Cardinal gradients
const float cfai = cfa[indx];
const float N_Grad = eps + (std::fabs(cfa[indx - w1] - cfa[indx + w1]) + std::fabs(cfai - cfa[indx - w2])) + (std::fabs(cfa[indx - w1] - cfa[indx - w3]) + std::fabs(cfa[indx - w2] - cfa[indx - w4]));
const float S_Grad = eps + (std::fabs(cfa[indx - w1] - cfa[indx + w1]) + std::fabs(cfai - cfa[indx + w2])) + (std::fabs(cfa[indx + w1] - cfa[indx + w3]) + std::fabs(cfa[indx + w2] - cfa[indx + w4]));
const float W_Grad = eps + (std::fabs(cfa[indx - 1] - cfa[indx + 1]) + std::fabs(cfai - cfa[indx - 2])) + (std::fabs(cfa[indx - 1] - cfa[indx - 3]) + std::fabs(cfa[indx - 2] - cfa[indx - 4]));
const float E_Grad = eps + (std::fabs(cfa[indx - 1] - cfa[indx + 1]) + std::fabs(cfai - cfa[indx + 2])) + (std::fabs(cfa[indx + 1] - cfa[indx + 3]) + std::fabs(cfa[indx + 2] - cfa[indx + 4]));
// Cardinal pixel estimations
const float lpfi = lpf[indx>>1];
const float N_Est = cfa[indx - w1] * (1.f + (lpfi - lpf[(indx - w2)>>1]) / (eps + lpfi + lpf[(indx - w2)>>1]));
const float S_Est = cfa[indx + w1] * (1.f + (lpfi - lpf[(indx + w2)>>1]) / (eps + lpfi + lpf[(indx + w2)>>1]));
const float W_Est = cfa[indx - 1] * (1.f + (lpfi - lpf[(indx - 2)>>1]) / (eps + lpfi + lpf[(indx - 2)>>1]));
const float E_Est = cfa[indx + 1] * (1.f + (lpfi - lpf[(indx + 2)>>1]) / (eps + lpfi + lpf[(indx + 2)>>1]));
// Vertical and horizontal estimations
const float V_Est = (S_Grad * N_Est + N_Grad * S_Est) / (N_Grad + S_Grad);
const float H_Est = (W_Grad * E_Est + E_Grad * W_Est) / (E_Grad + W_Grad);
// G@B and G@R interpolation
// Refined vertical and horizontal local discrimination
const float VH_Central_Value = VH_Dir[indx];
const float VH_Neighbourhood_Value = 0.25f * ((VH_Dir[indx - w1 - 1] + VH_Dir[indx - w1 + 1]) + (VH_Dir[indx + w1 - 1] + VH_Dir[indx + w1 + 1]));
const float VH_Disc = std::fabs(0.5f - VH_Central_Value) < std::fabs(0.5f - VH_Neighbourhood_Value) ? VH_Neighbourhood_Value : VH_Central_Value;
rgb[1][indx] = VH_Disc * H_Est + (1.f - VH_Disc) * V_Est;
}
}
/**
* STEP 4: Populate the red and blue channels
*/
@@ -203,7 +235,6 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
float Q_Stat = max(epssq, - 18.f * cfai * (cfa[indx + w1 - 1] + cfa[indx - w1 + 1] + 2.f * (cfa[indx + w2 - 2] + cfa[indx - w2 + 2]) - cfa[indx + w3 - 3] - cfa[indx - w3 + 3]) - 2.f * cfai * (cfa[indx + w4 - 4] + cfa[indx - w4 + 4] - 19.f * cfai) - cfa[indx + w1 - 1] * (70.f * cfa[indx - w1 + 1] - 12.f * cfa[indx + w2 - 2] + 24.f * cfa[indx - w2 + 2] - 38.f * cfa[indx + w3 - 3] + 16.f * cfa[indx - w3 + 3] + 12.f * cfa[indx + w4 - 4] - 6.f * cfa[indx - w4 + 4] + 46.f * cfa[indx + w1 - 1]) + cfa[indx - w1 + 1] * (24.f * cfa[indx + w2 - 2] - 12.f * cfa[indx - w2 + 2] + 16.f * cfa[indx + w3 - 3] - 38.f * cfa[indx - w3 + 3] - 6.f * cfa[indx + w4 - 4] + 12.f * cfa[indx - w4 + 4] + 46.f * cfa[indx - w1 + 1]) + cfa[indx + w2 - 2] * (14.f * cfa[indx - w2 + 2] - 12.f * cfa[indx - w3 + 3] - 2.f * (cfa[indx + w4 - 4] - cfa[indx - w4 + 4]) + 11.f * cfa[indx + w2 - 2]) - cfa[indx - w2 + 2] * (12.f * cfa[indx + w3 - 3] + 2.f * (cfa[indx + w4 - 4] - cfa[indx - w4 + 4]) + 11.f * cfa[indx - w2 + 2]) + cfa[indx + w3 - 3] * (2.f * cfa[indx - w3 + 3] - 6.f * cfa[indx + w4 - 4] + 10.f * cfa[indx + w3 - 3]) - cfa[indx - w3 + 3] * (6.f * cfa[indx - w4 + 4] + 10.f * cfa[indx - w3 + 3]) + cfa[indx + w4 - 4] * cfa[indx + w4 - 4] + cfa[indx - w4 + 4] * cfa[indx - w4 + 4]);
PQ_Dir[indx] = P_Stat / (P_Stat + Q_Stat);
}
}
@@ -235,7 +266,6 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
// R@B and B@R interpolation
rgb[c][indx] = rgb[1][indx] + (1.f - PQ_Disc) * P_Est + PQ_Disc * Q_Est;
}
}
@@ -279,7 +309,6 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
// R@G and B@G interpolation
rgb[c][indx] = rgb1 + (1.f - VH_Disc) * V_Est + VH_Disc * H_Est;
}
}
}
@@ -293,9 +322,9 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
}
}
if(plistener) {
if (plistener) {
progresscounter++;
if(progresscounter % 32 == 0) {
if (progresscounter % 32 == 0) {
#ifdef _OPENMP
#pragma omp critical (rcdprogress)
#endif
@@ -306,7 +335,6 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
}
}
}
}
}
@@ -321,7 +349,6 @@ void RawImageSource::rcd_demosaic(size_t chunkSize, bool measure)
if (plistener) {
plistener->setProgress(1);
}
// -------------------------------------------------------------------------
}
} /* namespace */