kwin/effects/blur/blurshader.cpp
Philipp Knechtges ef9aeafe0d kwin: Optimizing Blur Part II
This patch adds an optional texture cache to the blur effect such that damaged windows in
front of the blurred region dont trigger a repaint of the whole blurred region which pretty
often results in a avalanche repaint of nearly the whole screen.

REVIEW: 101977
2011-09-18 02:01:31 +02:00

485 lines
13 KiB
C++

/*
* Copyright © 2010 Fredrik Höglund <fredrik@kde.org>
*
* 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; see the file COPYING. if not, write to
* the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
* Boston, MA 02110-1301, USA.
*/
#include "blurshader.h"
#include <kwineffects.h>
#include <kwinglplatform.h>
#include <QByteArray>
#include <QMatrix4x4>
#include <QTextStream>
#include <QVector2D>
#include <KDebug>
#include <cmath>
using namespace KWin;
BlurShader::BlurShader()
: mRadius(0), mValid(false)
{
}
BlurShader::~BlurShader()
{
}
BlurShader *BlurShader::create()
{
if (GLSLBlurShader::supported())
return new GLSLBlurShader();
return new ARBBlurShader();
}
void BlurShader::setRadius(int radius)
{
const int r = qMax(radius, 2);
if (mRadius != r) {
mRadius = r;
reset();
init();
}
}
void BlurShader::setDirection(Qt::Orientation direction)
{
mDirection = direction;
}
float BlurShader::gaussian(float x, float sigma) const
{
return (1.0 / std::sqrt(2.0 * M_PI) * sigma)
* std::exp(-((x * x) / (2.0 * sigma * sigma)));
}
QVector<float> BlurShader::gaussianKernel() const
{
int size = qMin(mRadius | 1, maxKernelSize());
if (!(size & 0x1))
size -= 1;
QVector<float> kernel(size);
const int center = size / 2;
const qreal sigma = (size - 1) / 2.5;
// Generate the gaussian kernel
kernel[center] = gaussian(0, sigma) * .5;
for (int i = 1; i <= center; i++) {
const float val = gaussian(1.5 + (i - 1) * 2.0, sigma);
kernel[center + i] = val;
kernel[center - i] = val;
}
// Normalize the kernel
qreal total = 0;
for (int i = 0; i < size; i++)
total += kernel[i];
for (int i = 0; i < size; i++)
kernel[i] /= total;
return kernel;
}
// ----------------------------------------------------------------------------
GLSLBlurShader::GLSLBlurShader()
: BlurShader(), shader(NULL)
{
}
GLSLBlurShader::~GLSLBlurShader()
{
reset();
}
void GLSLBlurShader::reset()
{
delete shader;
shader = NULL;
setIsValid(false);
}
bool GLSLBlurShader::supported()
{
if (!GLPlatform::instance()->supports(GLSL))
return false;
if (!ShaderManager::instance()->isValid())
return false;
(void) glGetError(); // Clear the error state
#ifndef KWIN_HAVE_OPENGLES
// These are the minimum values the implementation is required to support
int value = 0;
glGetIntegerv(GL_MAX_VARYING_FLOATS, &value);
if (value < 32)
return false;
glGetIntegerv(GL_MAX_FRAGMENT_UNIFORM_COMPONENTS, &value);
if (value < 64)
return false;
glGetIntegerv(GL_MAX_VERTEX_UNIFORM_COMPONENTS, &value);
if (value < 512)
return false;
#endif
if (glGetError() != GL_NO_ERROR)
return false;
return true;
}
void GLSLBlurShader::setPixelDistance(float val)
{
if (!isValid())
return;
QVector2D pixelSize(0.0, 0.0);
if (direction() == Qt::Horizontal)
pixelSize.setX(val);
else
pixelSize.setY(val);
shader->setUniform("pixelSize", pixelSize);
}
void GLSLBlurShader::setTextureMatrix(const QMatrix4x4 &matrix)
{
if (!isValid()) {
return;
}
shader->setUniform("u_textureMatrix", matrix);
}
void GLSLBlurShader::setModelViewProjectionMatrix(const QMatrix4x4 &matrix)
{
if (!isValid()) {
return;
}
shader->setUniform("u_modelViewProjectionMatrix", matrix);
}
void GLSLBlurShader::bind()
{
if (!isValid())
return;
ShaderManager::instance()->pushShader(shader);
}
void GLSLBlurShader::unbind()
{
ShaderManager::instance()->popShader();
}
int GLSLBlurShader::maxKernelSize() const
{
#ifdef KWIN_HAVE_OPENGLES
// GL_MAX_VARYING_FLOATS not available in GLES
// querying for GL_MAX_VARYING_VECTORS crashes on nouveau
// using the minimum value of 8
return 8 * 2;
#else
int value;
glGetIntegerv(GL_MAX_VARYING_FLOATS, &value);
// Maximum number of vec4 varyings * 2
// The code generator will pack two vec2's into each vec4.
return value / 2;
#endif
}
void GLSLBlurShader::init()
{
QVector<float> kernel = gaussianKernel();
const int size = kernel.size();
const int center = size / 2;
QByteArray vertexSource;
QByteArray fragmentSource;
// Vertex shader
// ===================================================================
QTextStream stream(&vertexSource);
stream << "uniform mat4 u_modelViewProjectionMatrix;\n";
stream << "uniform mat4 u_textureMatrix;\n";
stream << "uniform vec2 pixelSize;\n\n";
stream << "attribute vec4 vertex;\n";
stream << "attribute vec4 texCoord;\n\n";
stream << "varying vec4 samplePos[" << std::ceil(size / 2.0) << "];\n";
stream << "\n";
stream << "void main(void)\n";
stream << "{\n";
stream << " vec4 center = vec4(u_textureMatrix * texCoord).stst;\n";
stream << " vec4 ps = pixelSize.stst;\n\n";
for (int i = 0; i < size; i += 2) {
float offset1, offset2;
if (i < center) {
offset1 = -(1.5 + (center - i - 1) * 2.0);
offset2 = (i + 1) == center ? 0 : offset1 + 2;
} else if (i > center) {
offset1 = 1.5 + (i - center - 1) * 2.0;
offset2 = (i + 1) == size ? 0 : offset1 + 2;
} else {
offset1 = 0;
offset2 = 1.5;
}
stream << " samplePos[" << i / 2 << "] = center + ps * vec4("
<< offset1 << ", " << offset1 << ", " << offset2 << ", " << offset2 << ");\n";
}
stream << "\n";
stream << " gl_Position = u_modelViewProjectionMatrix * vertex;\n";
stream << "}\n";
stream.flush();
// Fragment shader
// ===================================================================
QTextStream stream2(&fragmentSource);
stream2 << "uniform sampler2D texUnit;\n";
stream2 << "varying vec4 samplePos[" << std::ceil(size / 2.0) << "];\n\n";
for (int i = 0; i <= center; i++)
stream2 << "const vec4 kernel" << i << " = vec4(" << kernel[i] << ");\n";
stream2 << "\n";
stream2 << "void main(void)\n";
stream2 << "{\n";
stream2 << " vec4 sum = texture2D(texUnit, samplePos[0].st) * kernel0;\n";
for (int i = 1; i < size; i++)
stream2 << " sum = sum + texture2D(texUnit, samplePos[" << i / 2 << ((i % 2) ? "].pq)" : "].st)")
<< " * kernel" << (i > center ? size - i - 1 : i) << ";\n";
stream2 << " gl_FragColor = sum;\n";
stream2 << "}\n";
stream2.flush();
shader = ShaderManager::instance()->loadShaderFromCode(vertexSource, fragmentSource);
if (shader->isValid()) {
QMatrix4x4 modelViewProjection;
modelViewProjection.ortho(0, displayWidth(), displayHeight(), 0, 0, 65535);
ShaderManager::instance()->pushShader(shader);
shader->setUniform("texUnit", 0);
shader->setUniform("u_textureMatrix", QMatrix4x4());
shader->setUniform("u_modelViewProjectionMatrix", modelViewProjection);
ShaderManager::instance()->popShader();
}
setIsValid(shader->isValid());
}
// ----------------------------------------------------------------------------
ARBBlurShader::ARBBlurShader()
: BlurShader(), program(0)
{
}
ARBBlurShader::~ARBBlurShader()
{
reset();
}
void ARBBlurShader::reset()
{
#ifndef KWIN_HAVE_OPENGLES
if (program) {
glDeleteProgramsARB(1, &program);
program = 0;
}
setIsValid(false);
#endif
}
bool ARBBlurShader::supported()
{
#ifdef KWIN_HAVE_OPENGLES
return false;
#else
if (!hasGLExtension("GL_ARB_fragment_program"))
return false;
(void) glGetError(); // Clear the error state
// These are the minimum values the implementation is required to support
int value = 0;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_PARAMETERS_ARB, &value);
if (value < 24)
return false;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_TEMPORARIES_ARB, &value);
if (value < 16)
return false;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_INSTRUCTIONS_ARB, &value);
if (value < 72)
return false;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_TEX_INSTRUCTIONS_ARB, &value);
if (value < 24)
return false;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB, &value);
if (value < 4)
return false;
if (glGetError() != GL_NO_ERROR)
return false;
return true;
#endif
}
void ARBBlurShader::setPixelDistance(float val)
{
#ifdef KWIN_HAVE_OPENGLES
Q_UNUSED(val)
#else
float firstStep = val * 1.5;
float nextStep = val * 2.0;
if (direction() == Qt::Horizontal) {
glProgramLocalParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, 0, firstStep, 0, 0, 0);
glProgramLocalParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, 1, nextStep, 0, 0, 0);
} else {
glProgramLocalParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, 0, 0, firstStep, 0, 0);
glProgramLocalParameter4fARB(GL_FRAGMENT_PROGRAM_ARB, 1, 0, nextStep, 0, 0);
}
#endif
}
void ARBBlurShader::bind()
{
#ifndef KWIN_HAVE_OPENGLES
if (!isValid())
return;
glEnable(GL_FRAGMENT_PROGRAM_ARB);
glBindProgramARB(GL_FRAGMENT_PROGRAM_ARB, program);
#endif
}
void ARBBlurShader::unbind()
{
#ifndef KWIN_HAVE_OPENGLES
int boundObject;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_PROGRAM_BINDING_ARB, &boundObject);
if (boundObject == program) {
glBindProgramARB(GL_FRAGMENT_PROGRAM_ARB, 0);
glDisable(GL_FRAGMENT_PROGRAM_ARB);
}
#endif
}
int ARBBlurShader::maxKernelSize() const
{
#ifdef KWIN_HAVE_OPENGLES
return 0;
#else
int value;
int result;
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_PARAMETERS_ARB, &value);
result = (value - 1) * 2; // We only need to store half the kernel, since it's symmetrical
glGetProgramivARB(GL_FRAGMENT_PROGRAM_ARB, GL_MAX_PROGRAM_INSTRUCTIONS_ARB, &value);
result = qMin(result, value / 3); // We need 3 instructions / sample
return result;
#endif
}
void ARBBlurShader::init()
{
#ifndef KWIN_HAVE_OPENGLES
QVector<float> kernel = gaussianKernel();
const int size = kernel.size();
const int center = size / 2;
QByteArray text;
QTextStream stream(&text);
stream << "!!ARBfp1.0\n";
// The kernel values are hardcoded into the program
for (int i = 0; i <= center; i++)
stream << "PARAM kernel" << i << " = " << kernel[center + i] << ";\n";
stream << "PARAM firstSample = program.local[0];\n"; // Distance from gl_TexCoord[0] to the next sample
stream << "PARAM nextSample = program.local[1];\n"; // Distance to the subsequent sample
// Temporary variables to hold coordinates and texture samples
for (int i = 0; i < size; i++)
stream << "TEMP temp" << i << ";\n";
// Compute the texture coordinates
stream << "ADD temp1, fragment.texcoord[0], firstSample;\n"; // temp1 = gl_TexCoord[0] + firstSample
stream << "SUB temp2, fragment.texcoord[0], firstSample;\n"; // temp2 = gl_TexCoord[0] - firstSample
for (int i = 1, j = 3; i < center; i++, j += 2) {
stream << "ADD temp" << j + 0 << ", temp" << j - 2 << ", nextSample;\n";
stream << "SUB temp" << j + 1 << ", temp" << j - 1 << ", nextSample;\n";
}
// Sample the texture coordinates
stream << "TEX temp0, fragment.texcoord[0], texture[0], 2D;\n";
for (int i = 1; i < size; i++)
stream << "TEX temp" << i << ", temp" << i << ", texture[0], 2D;\n";
// Multiply the samples with the kernel values and compute the sum
stream << "MUL temp0, temp0, kernel0;\n";
for (int i = 0, j = 1; i < center; i++) {
stream << "MAD temp0, temp" << j++ << ", kernel" << i + 1 << ", temp0;\n";
stream << "MAD temp0, temp" << j++ << ", kernel" << i + 1 << ", temp0;\n";
}
stream << "MOV result.color, temp0;\n"; // gl_FragColor = temp0
stream << "END\n";
stream.flush();
glGenProgramsARB(1, &program);
glBindProgramARB(GL_FRAGMENT_PROGRAM_ARB, program);
glProgramStringARB(GL_FRAGMENT_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB, text.length(), text.constData());
if (glGetError()) {
const char *error = (const char*)glGetString(GL_PROGRAM_ERROR_STRING_ARB);
kError() << "Failed to compile fragment program:" << error;
setIsValid(false);
} else
setIsValid(true);
glBindProgramARB(GL_FRAGMENT_PROGRAM_ARB, 0);
#endif
}