mirror-github-bspeice/fractorium
1
0
Fork 0
mirror of https://bitbucket.org/mfeemster/fractorium.git synced 2025-02-01 18:40:12 -05:00
Code Issues Actions Packages Projects Releases Wiki Activity

--Bug fixes

Browse Source 
-Fix crash when using Lanczos2 filter with color curves due to negative numbers. Fix by passing abs() of first argument to pow() in Renderer::GammaCorrection().
 -Fix crash in hexes with SP. Rounding error caused out of bounds index.
This commit is contained in:
mfeemster 2016-02-23 21:01:02 -08:00
parent c282ee4c2c
commit 32d6982210
4 changed files with 202 additions and 218 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Source/Ember/Renderer.cpp
View File

@ -1629,7 +1629,7 @@ void Renderer<T, bucketT>::GammaCorrection(tvec4<bucketT, glm::defaultp>& bucket
for (glm::length_t rgbi = 0; rgbi < 3; rgbi++) for (glm::length_t rgbi = 0; rgbi < 3; rgbi++)
{ {
a = newRgb[rgbi] + ((1 - vibrancy) * 255 * std::pow(bucket[rgbi], g)); a = newRgb[rgbi] + ((1 - vibrancy) * 255 * std::pow(std::abs(bucket[rgbi]), g));//Must use abs(), else it it could be a negative value and return NAN.
if (NumChannels() <= 3 || !Transparency()) if (NumChannels() <= 3 || !Transparency())
{ {

9
Source/Ember/Variations03.h
View File

@ -2393,9 +2393,10 @@ public:
virtual void Precalc() override virtual void Precalc() override
{ {
m_InvN = m_Dist / m_Power; auto zp = Zeps(m_Power);
m_Inv2PiN = M_2PI / m_Power; m_InvN = m_Dist / zp;
m_Cn = m_Dist / m_Power / 2; m_Inv2PiN = M_2PI / zp;
m_Cn = m_Dist / zp / 2;
} }
protected: protected:
@ -2405,7 +2406,7 @@ protected:
m_Params.clear(); m_Params.clear();
m_Params.push_back(ParamWithName<T>(&m_Power, prefix + "phoenix_julia_power", 2)); m_Params.push_back(ParamWithName<T>(&m_Power, prefix + "phoenix_julia_power", 2));
m_Params.push_back(ParamWithName<T>(&m_Dist, prefix + "phoenix_julia_dist", 1)); m_Params.push_back(ParamWithName<T>(&m_Dist, prefix + "phoenix_julia_dist", 1));
m_Params.push_back(ParamWithName<T>(&m_XDistort, prefix + "phoenix_julia_x_distort", T(-T(0.5))));//Original omitted phoenix_ prefix. m_Params.push_back(ParamWithName<T>(&m_XDistort, prefix + "phoenix_julia_x_distort", T(-0.5)));//Original omitted phoenix_ prefix.
m_Params.push_back(ParamWithName<T>(&m_YDistort, prefix + "phoenix_julia_y_distort")); m_Params.push_back(ParamWithName<T>(&m_YDistort, prefix + "phoenix_julia_y_distort"));
m_Params.push_back(ParamWithName<T>(true, &m_Cn, prefix + "phoenix_julia_cn"));//Precalc. m_Params.push_back(ParamWithName<T>(true, &m_Cn, prefix + "phoenix_julia_cn"));//Precalc.
m_Params.push_back(ParamWithName<T>(true, &m_InvN, prefix + "phoenix_julia_invn")); m_Params.push_back(ParamWithName<T>(true, &m_InvN, prefix + "phoenix_julia_invn"));

22
Source/Ember/Variations06.h
View File

@ -36,7 +36,7 @@ public:
static const T AYoXh = T(1.7320508075688772935 / 2.0); static const T AYoXh = T(1.7320508075688772935 / 2.0);
static const T AYoYh = T(1.7320508075688772935 / 2.0); static const T AYoYh = T(1.7320508075688772935 / 2.0);
static const v2T offset[4] { { 0, 0 }, { 0, 1 }, { 1, 0 }, { 1, 1 } }; static const v2T offset[4] { { 0, 0 }, { 0, 1 }, { 1, 0 }, { 1, 1 } };
int i = 0; int i, j;
T di, dj; T di, dj;
T XCh, YCh, XCo, YCo, DXo, DYo, L, L1, L2, R, s, trgL; T XCh, YCh, XCo, YCo, DXo, DYo, L, L1, L2, R, s, trgL;
v2T u, v; v2T u, v;
@ -60,13 +60,12 @@ public:
YCh = T(Floor((AYhXo * u.x + AYhYo * u.y) / s)); YCh = T(Floor((AYhXo * u.x + AYhYo * u.y) / s));
// Get a set of 4 hex center points, based around the one above // Get a set of 4 hex center points, based around the one above
for (di = XCh; di < XCh + T(1.1); di += 1) for (i = 0, di = XCh; i < 2; di += 1, i++)//Note that in SP mode, these numbers won't advance if they are on the boundary of what can be represented with an DP number...
{ {
for (dj = YCh; dj < YCh + T(1.1); dj += 1) for (j = 0, dj = YCh; j < 2; dj += 1, j++)//...which is why the check uses i and j.
{ {
P[i].x = (AXoXh * di + AXoYh * dj) * s; P[(i * 2) + j].x = (AXoXh * di + AXoYh * dj) * s;
P[i].y = (AYoXh * di + AYoYh * dj) * s; P[(i * 2) + j].y = (AYoXh * di + AYoYh * dj) * s;
i++;
} }
} }
@ -151,7 +150,7 @@ public:
string rotsin = "parVars[" + ToUpper(m_Params[i++].Name()) + index; string rotsin = "parVars[" + ToUpper(m_Params[i++].Name()) + index;
string rotcos = "parVars[" + ToUpper(m_Params[i++].Name()) + index; string rotcos = "parVars[" + ToUpper(m_Params[i++].Name()) + index;
ss << "\t{\n" ss << "\t{\n"
<< "\t\tint i = 0;\n" << "\t\tint i, j;\n"
<< "\t\treal_t di, dj;\n" << "\t\treal_t di, dj;\n"
<< "\t\treal_t XCh, YCh, XCo, YCo, DXo, DYo, L, L1, L2, R, s, trgL, Vx, Vy;\n" << "\t\treal_t XCh, YCh, XCo, YCo, DXo, DYo, L, L1, L2, R, s, trgL, Vx, Vy;\n"
<< "\t\treal2 U;\n" << "\t\treal2 U;\n"
@ -173,13 +172,12 @@ public:
<< "\t\tXCh = floor((AXhXo * U.x + AXhYo * U.y) / s);\n" << "\t\tXCh = floor((AXhXo * U.x + AXhYo * U.y) / s);\n"
<< "\t\tYCh = floor((AYhXo * U.x + AYhYo * U.y) / s);\n" << "\t\tYCh = floor((AYhXo * U.x + AYhYo * U.y) / s);\n"
<< "\n" << "\n"
<< "\t\tfor (di = XCh; di < XCh + 1.1; di += 1)\n" << "\t\tfor (i = 0, di = XCh; i < 2; di += 1, i++)\n"
<< "\t\t{\n" << "\t\t{\n"
<< "\t\t for (dj = YCh; dj < YCh + 1.1; dj += 1)\n" << "\t\t for (j = 0, dj = YCh; j < 2; dj += 1, j++)\n"
<< "\t\t {\n" << "\t\t {\n"
<< "\t\t P[i].x = (AXoXh * di + AXoYh * dj) * s;\n" << "\t\t P[(i * 2) + j].x = (AXoXh * di + AXoYh * dj) * s;\n"
<< "\t\t P[i].y = (AYoXh * di + AYoYh * dj) * s;\n" << "\t\t P[(i * 2) + j].y = (AYoXh * di + AYoYh * dj) * s;\n"
<< "\t\t i++;\n"
<< "\t\t }\n" << "\t\t }\n"
<< "\t\t}\n" << "\t\t}\n"
<< "\n" << "\n"

387
Source/EmberCL/FinalAccumOpenCLKernelCreator.cpp
View File

@ -12,22 +12,17 @@ FinalAccumOpenCLKernelCreator::FinalAccumOpenCLKernelCreator(bool doublePrecisio
m_DoublePrecision = doublePrecision; m_DoublePrecision = doublePrecision;
m_GammaCorrectionWithAlphaCalcEntryPoint = "GammaCorrectionWithAlphaCalcKernel"; m_GammaCorrectionWithAlphaCalcEntryPoint = "GammaCorrectionWithAlphaCalcKernel";
m_GammaCorrectionWithoutAlphaCalcEntryPoint = "GammaCorrectionWithoutAlphaCalcKernel"; m_GammaCorrectionWithoutAlphaCalcEntryPoint = "GammaCorrectionWithoutAlphaCalcKernel";
m_GammaCorrectionWithAlphaCalcKernel = CreateGammaCorrectionKernelString(true); m_GammaCorrectionWithAlphaCalcKernel = CreateGammaCorrectionKernelString(true);
m_GammaCorrectionWithoutAlphaCalcKernel = CreateGammaCorrectionKernelString(false); m_GammaCorrectionWithoutAlphaCalcKernel = CreateGammaCorrectionKernelString(false);
m_FinalAccumEarlyClipEntryPoint = "FinalAccumEarlyClipKernel"; m_FinalAccumEarlyClipEntryPoint = "FinalAccumEarlyClipKernel";
m_FinalAccumEarlyClipWithAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumEarlyClipWithAlphaCalcWithAlphaAccumKernel"; m_FinalAccumEarlyClipWithAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumEarlyClipWithAlphaCalcWithAlphaAccumKernel";
m_FinalAccumEarlyClipWithoutAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumEarlyClipWithoutAlphaCalcWithAlphaAccumKernel"; m_FinalAccumEarlyClipWithoutAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumEarlyClipWithoutAlphaCalcWithAlphaAccumKernel";
m_FinalAccumEarlyClipKernel = CreateFinalAccumKernelString(true, false, false); m_FinalAccumEarlyClipKernel = CreateFinalAccumKernelString(true, false, false);
m_FinalAccumEarlyClipWithAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(true, true, true); m_FinalAccumEarlyClipWithAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(true, true, true);
m_FinalAccumEarlyClipWithoutAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(true, false, true); m_FinalAccumEarlyClipWithoutAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(true, false, true);
m_FinalAccumLateClipEntryPoint = "FinalAccumLateClipKernel"; m_FinalAccumLateClipEntryPoint = "FinalAccumLateClipKernel";
m_FinalAccumLateClipWithAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumLateClipWithAlphaCalcWithAlphaAccumKernel"; m_FinalAccumLateClipWithAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumLateClipWithAlphaCalcWithAlphaAccumKernel";
m_FinalAccumLateClipWithoutAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumLateClipWithoutAlphaCalcWithAlphaAccumKernel"; m_FinalAccumLateClipWithoutAlphaCalcWithAlphaAccumEntryPoint = "FinalAccumLateClipWithoutAlphaCalcWithAlphaAccumKernel";
m_FinalAccumLateClipKernel = CreateFinalAccumKernelString(false, false, false); m_FinalAccumLateClipKernel = CreateFinalAccumKernelString(false, false, false);
m_FinalAccumLateClipWithAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(false, true, true); m_FinalAccumLateClipWithAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(false, true, true);
m_FinalAccumLateClipWithoutAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(false, false, true); m_FinalAccumLateClipWithoutAlphaCalcWithAlphaAccumKernel = CreateFinalAccumKernelString(false, false, true);
@ -183,15 +178,14 @@ string FinalAccumOpenCLKernelCreator::CreateFinalAccumKernelString(bool earlyCli
{ {
ostringstream os; ostringstream os;
string channels = alphaAccum ? "4" : "3"; string channels = alphaAccum ? "4" : "3";
os << os <<
ConstantDefinesString(m_DoublePrecision) << ConstantDefinesString(m_DoublePrecision) <<
UnionCLStructString << UnionCLStructString <<
RgbToHsvFunctionString << RgbToHsvFunctionString <<
HsvToRgbFunctionString << HsvToRgbFunctionString <<
CalcAlphaFunctionString << CalcAlphaFunctionString <<
CurveAdjustFunctionString << CurveAdjustFunctionString <<
SpatialFilterCLStructString; SpatialFilterCLStructString;
if (earlyClip) if (earlyClip)
{ {
@ -207,8 +201,8 @@ string FinalAccumOpenCLKernelCreator::CreateFinalAccumKernelString(bool earlyCli
else else
{ {
os << os <<
CreateCalcNewRgbFunctionString(false) << CreateCalcNewRgbFunctionString(false) <<
CreateGammaCorrectionFunctionString(false, alphaCalc, alphaAccum, true); CreateGammaCorrectionFunctionString(false, alphaCalc, alphaAccum, true);
if (!alphaCalc && !alphaAccum)//Rgb output, the most common case. if (!alphaCalc && !alphaAccum)//Rgb output, the most common case.
os << "__kernel void " << m_FinalAccumLateClipEntryPoint << "(\n"; os << "__kernel void " << m_FinalAccumLateClipEntryPoint << "(\n";
@ -221,53 +215,53 @@ string FinalAccumOpenCLKernelCreator::CreateFinalAccumKernelString(bool earlyCli
} }
os << os <<
" const __global real4reals_bucket* accumulator,\n" " const __global real4reals_bucket* accumulator,\n"
" __write_only image2d_t pixels,\n" " __write_only image2d_t pixels,\n"
" __constant SpatialFilterCL* spatialFilter,\n" " __constant SpatialFilterCL* spatialFilter,\n"
" __constant real_bucket_t* filterCoefs,\n" " __constant real_bucket_t* filterCoefs,\n"
" __constant real4reals_bucket* csa,\n" " __constant real4reals_bucket* csa,\n"
" const uint doCurves,\n" " const uint doCurves,\n"
" const real_bucket_t alphaBase,\n" " const real_bucket_t alphaBase,\n"
" const real_bucket_t alphaScale\n" " const real_bucket_t alphaScale\n"
"\t)\n" "\t)\n"
"{\n" "{\n"
"\n" "\n"
" if ((GLOBAL_ID_Y >= spatialFilter->m_FinalRasH) || (GLOBAL_ID_X >= spatialFilter->m_FinalRasW))\n" " if ((GLOBAL_ID_Y >= spatialFilter->m_FinalRasH) || (GLOBAL_ID_X >= spatialFilter->m_FinalRasW))\n"
" return;\n" " return;\n"
"\n" "\n"
" uint accumX = spatialFilter->m_DensityFilterOffset + (GLOBAL_ID_X * spatialFilter->m_Supersample);\n" " uint accumX = spatialFilter->m_DensityFilterOffset + (GLOBAL_ID_X * spatialFilter->m_Supersample);\n"
" uint accumY = spatialFilter->m_DensityFilterOffset + (GLOBAL_ID_Y * spatialFilter->m_Supersample);\n" " uint accumY = spatialFilter->m_DensityFilterOffset + (GLOBAL_ID_Y * spatialFilter->m_Supersample);\n"
" int2 finalCoord;\n" " int2 finalCoord;\n"
" finalCoord.x = GLOBAL_ID_X;\n" " finalCoord.x = GLOBAL_ID_X;\n"
" finalCoord.y = (int)((spatialFilter->m_YAxisUp == 1) ? ((spatialFilter->m_FinalRasH - GLOBAL_ID_Y) - 1) : GLOBAL_ID_Y);\n" " finalCoord.y = (int)((spatialFilter->m_YAxisUp == 1) ? ((spatialFilter->m_FinalRasH - GLOBAL_ID_Y) - 1) : GLOBAL_ID_Y);\n"
" float4floats finalColor;\n" " float4floats finalColor;\n"
" int ii, jj;\n" " int ii, jj;\n"
" uint filterKRowIndex;\n" " uint filterKRowIndex;\n"
" const __global real4reals_bucket* accumBucket;\n" " const __global real4reals_bucket* accumBucket;\n"
" real4reals_bucket newBucket;\n" " real4reals_bucket newBucket;\n"
" newBucket.m_Real4 = 0;\n" " newBucket.m_Real4 = 0;\n"
"\n" "\n"
" for (jj = 0; jj < spatialFilter->m_FilterWidth; jj++)\n" " for (jj = 0; jj < spatialFilter->m_FilterWidth; jj++)\n"
" {\n" " {\n"
" filterKRowIndex = jj * spatialFilter->m_FilterWidth;\n" " filterKRowIndex = jj * spatialFilter->m_FilterWidth;\n"
"\n" "\n"
" for (ii = 0; ii < spatialFilter->m_FilterWidth; ii++)\n" " for (ii = 0; ii < spatialFilter->m_FilterWidth; ii++)\n"
" {\n" " {\n"
" real_bucket_t k = filterCoefs[ii + filterKRowIndex];\n" " real_bucket_t k = filterCoefs[ii + filterKRowIndex];\n"
"\n" "\n"
" accumBucket = accumulator + (accumX + ii) + ((accumY + jj) * spatialFilter->m_SuperRasW);\n" " accumBucket = accumulator + (accumX + ii) + ((accumY + jj) * spatialFilter->m_SuperRasW);\n"
" newBucket.m_Real4 += (k * accumBucket->m_Real4);\n" " newBucket.m_Real4 += (k * accumBucket->m_Real4);\n"
" }\n" " }\n"
" }\n" " }\n"
"\n"; "\n";
//Not supporting 2 bytes per channel on the GPU. If the user wants it, run on the CPU. //Not supporting 2 bytes per channel on the GPU. If the user wants it, run on the CPU.
if (earlyClip)//If early clip, simply assign values directly to the temp float4 since they've been gamma corrected already, then write it straight to the output image below. if (earlyClip)//If early clip, simply assign values directly to the temp float4 since they've been gamma corrected already, then write it straight to the output image below.
{ {
os << os <<
" finalColor.m_Float4.x = (float)newBucket.m_Real4.x;\n"//CPU side clamps, skip here because write_imagef() does the clamping for us. " finalColor.m_Float4.x = (float)newBucket.m_Real4.x;\n"//CPU side clamps, skip here because write_imagef() does the clamping for us.
" finalColor.m_Float4.y = (float)newBucket.m_Real4.y;\n" " finalColor.m_Float4.y = (float)newBucket.m_Real4.y;\n"
" finalColor.m_Float4.z = (float)newBucket.m_Real4.z;\n"; " finalColor.m_Float4.z = (float)newBucket.m_Real4.z;\n";
if (alphaAccum) if (alphaAccum)
{ {
@ -283,37 +277,36 @@ string FinalAccumOpenCLKernelCreator::CreateFinalAccumKernelString(bool earlyCli
if (m_DoublePrecision) if (m_DoublePrecision)
{ {
os << os <<
" real4reals_bucket realFinal;\n" " real4reals_bucket realFinal;\n"
"\n" "\n"
" GammaCorrectionFloats(&newBucket, &(spatialFilter->m_Background[0]), spatialFilter->m_Gamma, spatialFilter->m_LinRange, spatialFilter->m_Vibrancy, spatialFilter->m_HighlightPower, alphaBase, alphaScale, &(realFinal.m_Reals[0]));\n" " GammaCorrectionFloats(&newBucket, &(spatialFilter->m_Background[0]), spatialFilter->m_Gamma, spatialFilter->m_LinRange, spatialFilter->m_Vibrancy, spatialFilter->m_HighlightPower, alphaBase, alphaScale, &(realFinal.m_Reals[0]));\n"
" finalColor.m_Float4.x = (float)realFinal.m_Real4.x;\n" " finalColor.m_Float4.x = (float)realFinal.m_Real4.x;\n"
" finalColor.m_Float4.y = (float)realFinal.m_Real4.y;\n" " finalColor.m_Float4.y = (float)realFinal.m_Real4.y;\n"
" finalColor.m_Float4.z = (float)realFinal.m_Real4.z;\n" " finalColor.m_Float4.z = (float)realFinal.m_Real4.z;\n"
" finalColor.m_Float4.w = (float)realFinal.m_Real4.w;\n" " finalColor.m_Float4.w = (float)realFinal.m_Real4.w;\n"
; ;
} }
else else
{ {
os << os <<
" GammaCorrectionFloats(&newBucket, &(spatialFilter->m_Background[0]), spatialFilter->m_Gamma, spatialFilter->m_LinRange, spatialFilter->m_Vibrancy, spatialFilter->m_HighlightPower, alphaBase, alphaScale, &(finalColor.m_Floats[0]));\n"; " GammaCorrectionFloats(&newBucket, &(spatialFilter->m_Background[0]), spatialFilter->m_Gamma, spatialFilter->m_LinRange, spatialFilter->m_Vibrancy, spatialFilter->m_HighlightPower, alphaBase, alphaScale, &(finalColor.m_Floats[0]));\n";
} }
} }
os << os <<
"\n" "\n"
" if (doCurves)\n" " if (doCurves)\n"
" {\n" " {\n"
" CurveAdjust(csa, &(finalColor.m_Floats[0]), 1);\n" " CurveAdjust(csa, &(finalColor.m_Floats[0]), 1);\n"
" CurveAdjust(csa, &(finalColor.m_Floats[1]), 2);\n" " CurveAdjust(csa, &(finalColor.m_Floats[1]), 2);\n"
" CurveAdjust(csa, &(finalColor.m_Floats[2]), 3);\n" " CurveAdjust(csa, &(finalColor.m_Floats[2]), 3);\n"
" }\n" " }\n"
"\n" "\n"
" finalColor.m_Float4 /= 255.0f;\n" " finalColor.m_Float4 /= 255.0f;\n"
" write_imagef(pixels, finalCoord, finalColor.m_Float4);\n"//Use write_imagef instead of write_imageui because only the former works when sharing with an OpenGL texture. " write_imagef(pixels, finalCoord, finalColor.m_Float4);\n"//Use write_imagef instead of write_imageui because only the former works when sharing with an OpenGL texture.
" barrier(CLK_GLOBAL_MEM_FENCE);\n"//Required, or else page tearing will occur during interactive rendering. " barrier(CLK_GLOBAL_MEM_FENCE);\n"//Required, or else page tearing will occur during interactive rendering.
"}\n" "}\n"
; ;
return os.str(); return os.str();
} }
@ -332,54 +325,52 @@ string FinalAccumOpenCLKernelCreator::CreateGammaCorrectionFunctionString(bool g
string dataType; string dataType;
string unionMember; string unionMember;
dataType = "real_bucket_t"; dataType = "real_bucket_t";
//Use real_t for all cases, early clip and final accum. //Use real_t for all cases, early clip and final accum.
os << "void GammaCorrectionFloats(" << (globalBucket ? "__global " : "") << "real4reals_bucket* bucket, __constant real_bucket_t* background, real_bucket_t g, real_bucket_t linRange, real_bucket_t vibrancy, real_bucket_t highlightPower, real_bucket_t alphaBase, real_bucket_t alphaScale, " << (finalOut ? "" : "__global") << " real_bucket_t* correctedChannels)\n"; os << "void GammaCorrectionFloats(" << (globalBucket ? "__global " : "") << "real4reals_bucket* bucket, __constant real_bucket_t* background, real_bucket_t g, real_bucket_t linRange, real_bucket_t vibrancy, real_bucket_t highlightPower, real_bucket_t alphaBase, real_bucket_t alphaScale, " << (finalOut ? "" : "__global") << " real_bucket_t* correctedChannels)\n";
os os
<< "{\n" << "{\n"
<< " real_bucket_t alpha, ls, tmp, a;\n" << " real_bucket_t alpha, ls, tmp, a;\n"
<< " real4reals_bucket newRgb;\n" << " real4reals_bucket newRgb;\n"
<< "\n" << "\n"
<< " if (bucket->m_Reals[3] <= 0)\n" << " if (bucket->m_Reals[3] <= 0)\n"
<< " {\n" << " {\n"
<< " alpha = 0;\n" << " alpha = 0;\n"
<< " ls = 0;\n" << " ls = 0;\n"
<< " }\n" << " }\n"
<< " else\n" << " else\n"
<< " {\n" << " {\n"
<< " tmp = bucket->m_Reals[3];\n" << " tmp = bucket->m_Reals[3];\n"
<< " alpha = CalcAlpha(tmp, g, linRange);\n" << " alpha = CalcAlpha(tmp, g, linRange);\n"
<< " ls = vibrancy * 256.0 * alpha / tmp;\n" << " ls = vibrancy * 256.0 * alpha / tmp;\n"
<< " alpha = clamp(alpha, (real_bucket_t)0.0, (real_bucket_t)1.0);\n" << " alpha = clamp(alpha, (real_bucket_t)0.0, (real_bucket_t)1.0);\n"
<< " }\n" << " }\n"
<< "\n" << "\n"
<< " CalcNewRgb(bucket, ls, highlightPower, &newRgb);\n" << " CalcNewRgb(bucket, ls, highlightPower, &newRgb);\n"
<< "\n" << "\n"
<< " for (uint rgbi = 0; rgbi < 3; rgbi++)\n" << " for (uint rgbi = 0; rgbi < 3; rgbi++)\n"
<< " {\n" << " {\n"
<< " a = newRgb.m_Reals[rgbi] + ((1.0 - vibrancy) * 256.0 * pow(bucket->m_Reals[rgbi], g));\n" << " a = newRgb.m_Reals[rgbi] + ((1.0 - vibrancy) * 256.0 * pow(fabs(bucket->m_Reals[rgbi]), g));\n"
<< "\n"; << "\n";
if (!alphaCalc) if (!alphaCalc)
{ {
os << os <<
" a += ((1.0 - alpha) * background[rgbi]);\n"; " a += ((1.0 - alpha) * background[rgbi]);\n";
} }
else else
{ {
os os
<< " if (alpha > 0)\n" << " if (alpha > 0)\n"
<< " a /= alpha;\n" << " a /= alpha;\n"
<< " else\n" << " else\n"
<< " a = 0;\n"; << " a = 0;\n";
} }
os << os <<
"\n" "\n"
" correctedChannels[rgbi] = (" << dataType << ")clamp(a, (real_bucket_t)0.0, (real_bucket_t)255.0);\n" " correctedChannels[rgbi] = (" << dataType << ")clamp(a, (real_bucket_t)0.0, (real_bucket_t)255.0);\n"
" }\n" " }\n"
"\n"; "\n";
//The CPU code has 3 cases for assigning alpha: //The CPU code has 3 cases for assigning alpha:
//[3] = alpha.//Early clip. //[3] = alpha.//Early clip.
@ -393,13 +384,12 @@ string FinalAccumOpenCLKernelCreator::CreateGammaCorrectionFunctionString(bool g
if (alphaAccum) if (alphaAccum)
{ {
os os
<< " correctedChannels[3] = (" << dataType << ")(alphaBase + (alpha * alphaScale));\n"; << " correctedChannels[3] = (" << dataType << ")(alphaBase + (alpha * alphaScale));\n";
} }
os << os <<
"}\n" "}\n"
"\n"; "\n";
return os.str(); return os.str();
} }
@ -411,63 +401,61 @@ string FinalAccumOpenCLKernelCreator::CreateGammaCorrectionFunctionString(bool g
string FinalAccumOpenCLKernelCreator::CreateCalcNewRgbFunctionString(bool globalBucket) string FinalAccumOpenCLKernelCreator::CreateCalcNewRgbFunctionString(bool globalBucket)
{ {
ostringstream os; ostringstream os;
os << os <<
"static void CalcNewRgb(" << (globalBucket ? "__global " : "") << "real4reals_bucket* oldRgb, real_bucket_t ls, real_bucket_t highPow, real4reals_bucket* newRgb)\n" "static void CalcNewRgb(" << (globalBucket ? "__global " : "") << "real4reals_bucket* oldRgb, real_bucket_t ls, real_bucket_t highPow, real4reals_bucket* newRgb)\n"
"{\n" "{\n"
" int rgbi;\n" " int rgbi;\n"
" real_bucket_t newls, lsratio;\n" " real_bucket_t newls, lsratio;\n"
" real4reals_bucket newHsv;\n" " real4reals_bucket newHsv;\n"
" real_bucket_t maxa, maxc;\n" " real_bucket_t maxa, maxc;\n"
" real_bucket_t adjhlp;\n" " real_bucket_t adjhlp;\n"
"\n" "\n"
" if (ls == 0 || (oldRgb->m_Real4.x == 0 && oldRgb->m_Real4.y == 0 && oldRgb->m_Real4.z == 0))\n"//Can't do a vector compare to zero. " if (ls == 0 || (oldRgb->m_Real4.x == 0 && oldRgb->m_Real4.y == 0 && oldRgb->m_Real4.z == 0))\n"//Can't do a vector compare to zero.
" {\n" " {\n"
" newRgb->m_Real4 = 0;\n" " newRgb->m_Real4 = 0;\n"
" return;\n" " return;\n"
" }\n" " }\n"
"\n" "\n"
//Identify the most saturated channel. //Identify the most saturated channel.
" maxc = max(max(oldRgb->m_Reals[0], oldRgb->m_Reals[1]), oldRgb->m_Reals[2]);\n" " maxc = max(max(oldRgb->m_Reals[0], oldRgb->m_Reals[1]), oldRgb->m_Reals[2]);\n"
" maxa = ls * maxc;\n" " maxa = ls * maxc;\n"
"\n" "\n"
//If a channel is saturated and highlight power is non-negative //If a channel is saturated and highlight power is non-negative
//modify the color to prevent hue shift. //modify the color to prevent hue shift.
" if (maxa > 255 && highPow >= 0)\n" " if (maxa > 255 && highPow >= 0)\n"
" {\n" " {\n"
" newls = 255.0 / maxc;\n" " newls = 255.0 / maxc;\n"
" lsratio = pow(newls / ls, highPow);\n" " lsratio = pow(newls / ls, highPow);\n"
"\n" "\n"
//Calculate the max-value color (ranged 0 - 1). //Calculate the max-value color (ranged 0 - 1).
" for (rgbi = 0; rgbi < 3; rgbi++)\n" " for (rgbi = 0; rgbi < 3; rgbi++)\n"
" newRgb->m_Reals[rgbi] = newls * oldRgb->m_Reals[rgbi] / 255.0;\n" " newRgb->m_Reals[rgbi] = newls * oldRgb->m_Reals[rgbi] / 255.0;\n"
"\n" "\n"
//Reduce saturation by the lsratio. //Reduce saturation by the lsratio.
" RgbToHsv(&(newRgb->m_Real4), &(newHsv.m_Real4));\n" " RgbToHsv(&(newRgb->m_Real4), &(newHsv.m_Real4));\n"
" newHsv.m_Real4.y *= lsratio;\n" " newHsv.m_Real4.y *= lsratio;\n"
" HsvToRgb(&(newHsv.m_Real4), &(newRgb->m_Real4));\n" " HsvToRgb(&(newHsv.m_Real4), &(newRgb->m_Real4));\n"
"\n" "\n"
" for (rgbi = 0; rgbi < 3; rgbi++)\n"//Unrolling and vectorizing makes no difference. " for (rgbi = 0; rgbi < 3; rgbi++)\n"//Unrolling and vectorizing makes no difference.
" newRgb->m_Reals[rgbi] *= 255.0;\n" " newRgb->m_Reals[rgbi] *= 255.0;\n"
" }\n" " }\n"
" else\n" " else\n"
" {\n" " {\n"
" newls = 255.0 / maxc;\n" " newls = 255.0 / maxc;\n"
" adjhlp = -highPow;\n" " adjhlp = -highPow;\n"
"\n" "\n"
" if (adjhlp > 1)\n" " if (adjhlp > 1)\n"
" adjhlp = 1;\n" " adjhlp = 1;\n"
"\n" "\n"
" if (maxa <= 255)\n" " if (maxa <= 255)\n"
" adjhlp = 1;\n" " adjhlp = 1;\n"
"\n" "\n"
//Calculate the max-value color (ranged 0 - 1) interpolated with the old behavior. //Calculate the max-value color (ranged 0 - 1) interpolated with the old behavior.
" for (rgbi = 0; rgbi < 3; rgbi++)\n"//Unrolling, caching and vectorizing makes no difference. " for (rgbi = 0; rgbi < 3; rgbi++)\n"//Unrolling, caching and vectorizing makes no difference.
" newRgb->m_Reals[rgbi] = ((1.0 - adjhlp) * newls + adjhlp * ls) * oldRgb->m_Reals[rgbi];\n" " newRgb->m_Reals[rgbi] = ((1.0 - adjhlp) * newls + adjhlp * ls) * oldRgb->m_Reals[rgbi];\n"
" }\n" " }\n"
"}\n" "}\n"
"\n"; "\n";
return os.str(); return os.str();
} }
@ -480,34 +468,31 @@ string FinalAccumOpenCLKernelCreator::CreateGammaCorrectionKernelString(bool alp
{ {
ostringstream os; ostringstream os;
string dataType; string dataType;
os << os <<
ConstantDefinesString(m_DoublePrecision) << ConstantDefinesString(m_DoublePrecision) <<
UnionCLStructString << UnionCLStructString <<
RgbToHsvFunctionString << RgbToHsvFunctionString <<
HsvToRgbFunctionString << HsvToRgbFunctionString <<
CalcAlphaFunctionString << CalcAlphaFunctionString <<
CreateCalcNewRgbFunctionString(true) << CreateCalcNewRgbFunctionString(true) <<
SpatialFilterCLStructString << SpatialFilterCLStructString <<
CreateGammaCorrectionFunctionString(true, alphaCalc, true, false);//Will only be used with float in this case, early clip. Will always alpha accum. CreateGammaCorrectionFunctionString(true, alphaCalc, true, false);//Will only be used with float in this case, early clip. Will always alpha accum.
os << "__kernel void " << (alphaCalc ? m_GammaCorrectionWithAlphaCalcEntryPoint : m_GammaCorrectionWithoutAlphaCalcEntryPoint) << "(\n" <<
os << "__kernel void " << (alphaCalc ? m_GammaCorrectionWithAlphaCalcEntryPoint : m_GammaCorrectionWithoutAlphaCalcEntryPoint) << "(\n" << " __global real4reals_bucket* accumulator,\n"
" __global real4reals_bucket* accumulator,\n" " __constant SpatialFilterCL* spatialFilter\n"
" __constant SpatialFilterCL* spatialFilter\n" ")\n"
")\n" "{\n"
"{\n" " int testGutter = 0;\n"
" int testGutter = 0;\n" "\n"
"\n" " if (GLOBAL_ID_Y >= (spatialFilter->m_SuperRasH - testGutter) || GLOBAL_ID_X >= (spatialFilter->m_SuperRasW - testGutter))\n"
" if (GLOBAL_ID_Y >= (spatialFilter->m_SuperRasH - testGutter) || GLOBAL_ID_X >= (spatialFilter->m_SuperRasW - testGutter))\n" " return;\n"
" return;\n" "\n"
"\n" " uint superIndex = (GLOBAL_ID_Y * spatialFilter->m_SuperRasW) + GLOBAL_ID_X;\n"
" uint superIndex = (GLOBAL_ID_Y * spatialFilter->m_SuperRasW) + GLOBAL_ID_X;\n" " __global real4reals_bucket* bucket = accumulator + superIndex;\n"
" __global real4reals_bucket* bucket = accumulator + superIndex;\n" //Pass in an alphaBase and alphaScale of 0, 1 which means to just directly assign the computed alpha value.
//Pass in an alphaBase and alphaScale of 0, 1 which means to just directly assign the computed alpha value. " GammaCorrectionFloats(bucket, &(spatialFilter->m_Background[0]), spatialFilter->m_Gamma, spatialFilter->m_LinRange, spatialFilter->m_Vibrancy, spatialFilter->m_HighlightPower, 0.0, 1.0, &(bucket->m_Reals[0]));\n"
" GammaCorrectionFloats(bucket, &(spatialFilter->m_Background[0]), spatialFilter->m_Gamma, spatialFilter->m_LinRange, spatialFilter->m_Vibrancy, spatialFilter->m_HighlightPower, 0.0, 1.0, &(bucket->m_Reals[0]));\n" "}\n"
"}\n" ;
;
return os.str(); return os.str();
} }
} }