qt5base-lts/tests/auto/gui/math3d/qquaternion/tst_qquaternion.cpp

1247 lines
36 KiB
C++
Raw Normal View History

/****************************************************************************
**
** Copyright (C) 2015 The Qt Company Ltd.
** Contact: http://www.qt.io/licensing/
**
** This file is part of the test suite of the Qt Toolkit.
**
** $QT_BEGIN_LICENSE:LGPL21$
** Commercial License Usage
** Licensees holding valid commercial Qt licenses may use this file in
** accordance with the commercial license agreement provided with the
** Software or, alternatively, in accordance with the terms contained in
** a written agreement between you and The Qt Company. For licensing terms
** and conditions see http://www.qt.io/terms-conditions. For further
** information use the contact form at http://www.qt.io/contact-us.
**
** GNU Lesser General Public License Usage
** Alternatively, this file may be used under the terms of the GNU Lesser
** General Public License version 2.1 or version 3 as published by the Free
** Software Foundation and appearing in the file LICENSE.LGPLv21 and
** LICENSE.LGPLv3 included in the packaging of this file. Please review the
** following information to ensure the GNU Lesser General Public License
** requirements will be met: https://www.gnu.org/licenses/lgpl.html and
** http://www.gnu.org/licenses/old-licenses/lgpl-2.1.html.
**
** As a special exception, The Qt Company gives you certain additional
** rights. These rights are described in The Qt Company LGPL Exception
** version 1.1, included in the file LGPL_EXCEPTION.txt in this package.
**
** $QT_END_LICENSE$
**
****************************************************************************/
#include <QtTest/QtTest>
#include <QtCore/qmath.h>
#include <QtGui/qquaternion.h>
// This is a more tolerant version of qFuzzyCompare that also handles the case
// where one or more of the values being compare are close to zero
static inline bool myFuzzyCompare(float p1, float p2)
{
if (qFuzzyIsNull(p1) && qFuzzyIsNull(p2))
return true;
return qAbs(qAbs(p1) - qAbs(p2)) <= 0.00003f;
}
static inline bool myFuzzyCompare(const QVector3D &v1, const QVector3D &v2)
{
return myFuzzyCompare(v1.x(), v2.x())
&& myFuzzyCompare(v1.y(), v2.y())
&& myFuzzyCompare(v1.z(), v2.z());
}
static inline bool myFuzzyCompare(const QQuaternion &q1, const QQuaternion &q2)
{
const float d = QQuaternion::dotProduct(q1, q2);
return myFuzzyCompare(d * d, 1.0f);
}
static inline bool myFuzzyCompareRadians(float p1, float p2)
{
static const float fPI = float(M_PI);
if (p1 < -fPI)
p1 += 2.0f * fPI;
else if (p1 > fPI)
p1 -= 2.0f * fPI;
if (p2 < -fPI)
p2 += 2.0f * fPI;
else if (p2 > fPI)
p2 -= 2.0f * fPI;
return qAbs(qAbs(p1) - qAbs(p2)) <= qDegreesToRadians(0.05f);
}
static inline bool myFuzzyCompareDegrees(float p1, float p2)
{
p1 = qDegreesToRadians(p1);
p2 = qDegreesToRadians(p2);
return myFuzzyCompareRadians(p1, p2);
}
class tst_QQuaternion : public QObject
{
Q_OBJECT
public:
tst_QQuaternion() {}
~tst_QQuaternion() {}
private slots:
void create();
void dotProduct_data();
void dotProduct();
void length_data();
void length();
void normalized_data();
void normalized();
void normalize_data();
void normalize();
void inverted_data();
void inverted();
void compare();
void add_data();
void add();
void subtract_data();
void subtract();
void multiply_data();
void multiply();
void multiplyFactor_data();
void multiplyFactor();
void divide_data();
void divide();
void negate_data();
void negate();
void conjugate_data();
void conjugate();
void fromAxisAndAngle_data();
void fromAxisAndAngle();
void fromRotationMatrix_data();
void fromRotationMatrix();
void fromAxes_data();
void fromAxes();
void rotationTo_data();
void rotationTo();
void fromEulerAngles_data();
void fromEulerAngles();
void slerp_data();
void slerp();
void nlerp_data();
void nlerp();
void properties();
void metaTypes();
};
// Test the creation of QQuaternion objects in various ways:
// construct, copy, and modify.
void tst_QQuaternion::create()
{
QQuaternion identity;
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(identity.x(), 0.0f);
QCOMPARE(identity.y(), 0.0f);
QCOMPARE(identity.z(), 0.0f);
QCOMPARE(identity.scalar(), 1.0f);
QVERIFY(identity.isIdentity());
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QQuaternion negativeZeroIdentity(1.0f, -0.0f, -0.0f, -0.0f);
QCOMPARE(negativeZeroIdentity.x(), -0.0f);
QCOMPARE(negativeZeroIdentity.y(), -0.0f);
QCOMPARE(negativeZeroIdentity.z(), -0.0f);
QCOMPARE(negativeZeroIdentity.scalar(), 1.0f);
QVERIFY(negativeZeroIdentity.isIdentity());
QQuaternion v1(34.0f, 1.0f, 2.5f, -89.25f);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), 1.0f);
QCOMPARE(v1.y(), 2.5f);
QCOMPARE(v1.z(), -89.25f);
QCOMPARE(v1.scalar(), 34.0f);
QVERIFY(!v1.isNull());
QQuaternion v1i(34, 1, 2, -89);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1i.x(), 1.0f);
QCOMPARE(v1i.y(), 2.0f);
QCOMPARE(v1i.z(), -89.0f);
QCOMPARE(v1i.scalar(), 34.0f);
QVERIFY(!v1i.isNull());
QQuaternion v2(v1);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v2.x(), 1.0f);
QCOMPARE(v2.y(), 2.5f);
QCOMPARE(v2.z(), -89.25f);
QCOMPARE(v2.scalar(), 34.0f);
QVERIFY(!v2.isNull());
QQuaternion v4;
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v4.x(), 0.0f);
QCOMPARE(v4.y(), 0.0f);
QCOMPARE(v4.z(), 0.0f);
QCOMPARE(v4.scalar(), 1.0f);
QVERIFY(v4.isIdentity());
v4 = v1;
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v4.x(), 1.0f);
QCOMPARE(v4.y(), 2.5f);
QCOMPARE(v4.z(), -89.25f);
QCOMPARE(v4.scalar(), 34.0f);
QVERIFY(!v4.isNull());
QQuaternion v9(34, QVector3D(1.0f, 2.5f, -89.25f));
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v9.x(), 1.0f);
QCOMPARE(v9.y(), 2.5f);
QCOMPARE(v9.z(), -89.25f);
QCOMPARE(v9.scalar(), 34.0f);
QVERIFY(!v9.isNull());
v1.setX(3.0f);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), 3.0f);
QCOMPARE(v1.y(), 2.5f);
QCOMPARE(v1.z(), -89.25f);
QCOMPARE(v1.scalar(), 34.0f);
QVERIFY(!v1.isNull());
v1.setY(10.5f);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), 3.0f);
QCOMPARE(v1.y(), 10.5f);
QCOMPARE(v1.z(), -89.25f);
QCOMPARE(v1.scalar(), 34.0f);
QVERIFY(!v1.isNull());
v1.setZ(15.5f);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), 3.0f);
QCOMPARE(v1.y(), 10.5f);
QCOMPARE(v1.z(), 15.5f);
QCOMPARE(v1.scalar(), 34.0f);
QVERIFY(!v1.isNull());
v1.setScalar(6.0f);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), 3.0f);
QCOMPARE(v1.y(), 10.5f);
QCOMPARE(v1.z(), 15.5f);
QCOMPARE(v1.scalar(), 6.0f);
QVERIFY(!v1.isNull());
v1.setVector(2.0f, 6.5f, -1.25f);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), 2.0f);
QCOMPARE(v1.y(), 6.5f);
QCOMPARE(v1.z(), -1.25f);
QCOMPARE(v1.scalar(), 6.0f);
QVERIFY(!v1.isNull());
QVERIFY(v1.vector() == QVector3D(2.0f, 6.5f, -1.25f));
v1.setVector(QVector3D(-2.0f, -6.5f, 1.25f));
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), -2.0f);
QCOMPARE(v1.y(), -6.5f);
QCOMPARE(v1.z(), 1.25f);
QCOMPARE(v1.scalar(), 6.0f);
QVERIFY(!v1.isNull());
QVERIFY(v1.vector() == QVector3D(-2.0f, -6.5f, 1.25f));
v1.setX(0.0f);
v1.setY(0.0f);
v1.setZ(0.0f);
v1.setScalar(0.0f);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v1.x(), 0.0f);
QCOMPARE(v1.y(), 0.0f);
QCOMPARE(v1.z(), 0.0f);
QCOMPARE(v1.scalar(), 0.0f);
QVERIFY(v1.isNull());
QVector4D v10 = v9.toVector4D();
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v10.x(), 1.0f);
QCOMPARE(v10.y(), 2.5f);
QCOMPARE(v10.z(), -89.25f);
QCOMPARE(v10.w(), 34.0f);
}
// Test the computation of dot product.
void tst_QQuaternion::dotProduct_data()
{
QTest::addColumn<float>("x1");
QTest::addColumn<float>("y1");
QTest::addColumn<float>("z1");
QTest::addColumn<float>("scalar1");
QTest::addColumn<float>("x2");
QTest::addColumn<float>("y2");
QTest::addColumn<float>("z2");
QTest::addColumn<float>("scalar2");
QTest::addColumn<float>("dot");
QTest::newRow("null")
<< 0.0f << 0.0f << 0.0f << 0.0f
<< 0.0f << 0.0f << 0.0f << 0.0f
<< 0.0f;
QTest::newRow("identity")
<< 0.0f << 0.0f << 0.0f << 1.0f
<< 0.0f << 0.0f << 0.0f << 1.0f
<< 1.0f;
QTest::newRow("unitvec")
<< 1.0f << 0.0f << 0.0f << 0.0f
<< 0.0f << 1.0f << 0.0f << 0.0f
<< 0.0f;
QTest::newRow("complex")
<< 1.0f << 2.0f << 3.0f << 4.0f
<< 4.0f << 5.0f << 6.0f << 7.0f
<< 60.0f;
}
void tst_QQuaternion::dotProduct()
{
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, scalar1);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, scalar2);
QFETCH(float, dot);
QQuaternion q1(scalar1, x1, y1, z1);
QQuaternion q2(scalar2, x2, y2, z2);
QCOMPARE(QQuaternion::dotProduct(q1, q2), dot);
QCOMPARE(QQuaternion::dotProduct(q2, q1), dot);
}
// Test length computation for quaternions.
void tst_QQuaternion::length_data()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QTest::addColumn<float>("x");
QTest::addColumn<float>("y");
QTest::addColumn<float>("z");
QTest::addColumn<float>("w");
QTest::addColumn<float>("len");
QTest::newRow("null") << 0.0f << 0.0f << 0.0f << 0.0f << 0.0f;
QTest::newRow("1x") << 1.0f << 0.0f << 0.0f << 0.0f << 1.0f;
QTest::newRow("1y") << 0.0f << 1.0f << 0.0f << 0.0f << 1.0f;
QTest::newRow("1z") << 0.0f << 0.0f << 1.0f << 0.0f << 1.0f;
QTest::newRow("1w") << 0.0f << 0.0f << 0.0f << 1.0f << 1.0f;
QTest::newRow("-1x") << -1.0f << 0.0f << 0.0f << 0.0f << 1.0f;
QTest::newRow("-1y") << 0.0f << -1.0f << 0.0f << 0.0f << 1.0f;
QTest::newRow("-1z") << 0.0f << 0.0f << -1.0f << 0.0f << 1.0f;
QTest::newRow("-1w") << 0.0f << 0.0f << 0.0f << -1.0f << 1.0f;
QTest::newRow("two") << 2.0f << -2.0f << 2.0f << 2.0f << std::sqrt(16.0f);
}
void tst_QQuaternion::length()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x);
QFETCH(float, y);
QFETCH(float, z);
QFETCH(float, w);
QFETCH(float, len);
QQuaternion v(w, x, y, z);
QCOMPARE(v.length(), len);
QCOMPARE(v.lengthSquared(), x * x + y * y + z * z + w * w);
}
// Test the unit vector conversion for quaternions.
void tst_QQuaternion::normalized_data()
{
// Use the same test data as the length test.
length_data();
}
void tst_QQuaternion::normalized()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x);
QFETCH(float, y);
QFETCH(float, z);
QFETCH(float, w);
QFETCH(float, len);
QQuaternion v(w, x, y, z);
QQuaternion u = v.normalized();
if (v.isNull())
QVERIFY(u.isNull());
else
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(u.length(), 1.0f);
QCOMPARE(u.x() * len, v.x());
QCOMPARE(u.y() * len, v.y());
QCOMPARE(u.z() * len, v.z());
QCOMPARE(u.scalar() * len, v.scalar());
}
// Test the unit vector conversion for quaternions.
void tst_QQuaternion::normalize_data()
{
// Use the same test data as the length test.
length_data();
}
void tst_QQuaternion::normalize()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x);
QFETCH(float, y);
QFETCH(float, z);
QFETCH(float, w);
QQuaternion v(w, x, y, z);
bool isNull = v.isNull();
v.normalize();
if (isNull)
QVERIFY(v.isNull());
else
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(v.length(), 1.0f);
}
void tst_QQuaternion::inverted_data()
{
// Use the same test data as the length test.
length_data();
}
void tst_QQuaternion::inverted()
{
QFETCH(float, x);
QFETCH(float, y);
QFETCH(float, z);
QFETCH(float, w);
QFETCH(float, len);
QQuaternion v(w, x, y, z);
QQuaternion u = v.inverted();
if (v.isNull()) {
QVERIFY(u.isNull());
} else {
len *= len;
QCOMPARE(-u.x() * len, v.x());
QCOMPARE(-u.y() * len, v.y());
QCOMPARE(-u.z() * len, v.z());
QCOMPARE(u.scalar() * len, v.scalar());
}
}
// Test the comparison operators for quaternions.
void tst_QQuaternion::compare()
{
QQuaternion v1(8, 1, 2, 4);
QQuaternion v2(8, 1, 2, 4);
QQuaternion v3(8, 3, 2, 4);
QQuaternion v4(8, 1, 3, 4);
QQuaternion v5(8, 1, 2, 3);
QQuaternion v6(3, 1, 2, 4);
QVERIFY(v1 == v2);
QVERIFY(v1 != v3);
QVERIFY(v1 != v4);
QVERIFY(v1 != v5);
QVERIFY(v1 != v6);
}
// Test addition for quaternions.
void tst_QQuaternion::add_data()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QTest::addColumn<float>("x1");
QTest::addColumn<float>("y1");
QTest::addColumn<float>("z1");
QTest::addColumn<float>("w1");
QTest::addColumn<float>("x2");
QTest::addColumn<float>("y2");
QTest::addColumn<float>("z2");
QTest::addColumn<float>("w2");
QTest::addColumn<float>("x3");
QTest::addColumn<float>("y3");
QTest::addColumn<float>("z3");
QTest::addColumn<float>("w3");
QTest::newRow("null")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 0.0f << 0.0f
<< 0.0f << 0.0f << 0.0f << 0.0f
<< 0.0f << 0.0f << 0.0f << 0.0f;
QTest::newRow("xonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 0.0f << 0.0f << 0.0f
<< 2.0f << 0.0f << 0.0f << 0.0f
<< 3.0f << 0.0f << 0.0f << 0.0f;
QTest::newRow("yonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 1.0f << 0.0f << 0.0f
<< 0.0f << 2.0f << 0.0f << 0.0f
<< 0.0f << 3.0f << 0.0f << 0.0f;
QTest::newRow("zonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 1.0f << 0.0f
<< 0.0f << 0.0f << 2.0f << 0.0f
<< 0.0f << 0.0f << 3.0f << 0.0f;
QTest::newRow("wonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 0.0f << 1.0f
<< 0.0f << 0.0f << 0.0f << 2.0f
<< 0.0f << 0.0f << 0.0f << 3.0f;
QTest::newRow("all")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << 3.0f << 8.0f
<< 4.0f << 5.0f << -6.0f << 9.0f
<< 5.0f << 7.0f << -3.0f << 17.0f;
}
void tst_QQuaternion::add()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, w1);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, w2);
QFETCH(float, x3);
QFETCH(float, y3);
QFETCH(float, z3);
QFETCH(float, w3);
QQuaternion v1(w1, x1, y1, z1);
QQuaternion v2(w2, x2, y2, z2);
QQuaternion v3(w3, x3, y3, z3);
QVERIFY((v1 + v2) == v3);
QQuaternion v4(v1);
v4 += v2;
QVERIFY(v4 == v3);
QCOMPARE(v4.x(), v1.x() + v2.x());
QCOMPARE(v4.y(), v1.y() + v2.y());
QCOMPARE(v4.z(), v1.z() + v2.z());
QCOMPARE(v4.scalar(), v1.scalar() + v2.scalar());
}
// Test subtraction for quaternions.
void tst_QQuaternion::subtract_data()
{
// Use the same test data as the add test.
add_data();
}
void tst_QQuaternion::subtract()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, w1);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, w2);
QFETCH(float, x3);
QFETCH(float, y3);
QFETCH(float, z3);
QFETCH(float, w3);
QQuaternion v1(w1, x1, y1, z1);
QQuaternion v2(w2, x2, y2, z2);
QQuaternion v3(w3, x3, y3, z3);
QVERIFY((v3 - v1) == v2);
QVERIFY((v3 - v2) == v1);
QQuaternion v4(v3);
v4 -= v1;
QVERIFY(v4 == v2);
QCOMPARE(v4.x(), v3.x() - v1.x());
QCOMPARE(v4.y(), v3.y() - v1.y());
QCOMPARE(v4.z(), v3.z() - v1.z());
QCOMPARE(v4.scalar(), v3.scalar() - v1.scalar());
QQuaternion v5(v3);
v5 -= v2;
QVERIFY(v5 == v1);
QCOMPARE(v5.x(), v3.x() - v2.x());
QCOMPARE(v5.y(), v3.y() - v2.y());
QCOMPARE(v5.z(), v3.z() - v2.z());
QCOMPARE(v5.scalar(), v3.scalar() - v2.scalar());
}
// Test quaternion multiplication.
void tst_QQuaternion::multiply_data()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QTest::addColumn<float>("x1");
QTest::addColumn<float>("y1");
QTest::addColumn<float>("z1");
QTest::addColumn<float>("w1");
QTest::addColumn<float>("x2");
QTest::addColumn<float>("y2");
QTest::addColumn<float>("z2");
QTest::addColumn<float>("w2");
QTest::newRow("null")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 0.0f << 0.0f
<< 0.0f << 0.0f << 0.0f << 0.0f;
QTest::newRow("unitvec")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 0.0f << 0.0f << 1.0f
<< 0.0f << 1.0f << 0.0f << 1.0f;
QTest::newRow("complex")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << 3.0f << 7.0f
<< 4.0f << 5.0f << 6.0f << 8.0f;
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
for (float w = -1.0f; w <= 1.0f; w += 0.5f)
for (float x = -1.0f; x <= 1.0f; x += 0.5f)
for (float y = -1.0f; y <= 1.0f; y += 0.5f)
for (float z = -1.0f; z <= 1.0f; z += 0.5f) {
QTest::newRow("exhaustive")
<< x << y << z << w
<< z << w << y << x;
}
}
void tst_QQuaternion::multiply()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, w1);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, w2);
QQuaternion q1(w1, x1, y1, z1);
QQuaternion q2(w2, x2, y2, z2);
// Use the simple algorithm at:
// http://www.j3d.org/matrix_faq/matrfaq_latest.html#Q53
// to calculate the answer we expect to get.
QVector3D v1(x1, y1, z1);
QVector3D v2(x2, y2, z2);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
float scalar = w1 * w2 - QVector3D::dotProduct(v1, v2);
QVector3D vector = w1 * v2 + w2 * v1 + QVector3D::crossProduct(v1, v2);
QQuaternion result(scalar, vector);
QVERIFY((q1 * q2) == result);
}
// Test multiplication by a factor for quaternions.
void tst_QQuaternion::multiplyFactor_data()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QTest::addColumn<float>("x1");
QTest::addColumn<float>("y1");
QTest::addColumn<float>("z1");
QTest::addColumn<float>("w1");
QTest::addColumn<float>("factor");
QTest::addColumn<float>("x2");
QTest::addColumn<float>("y2");
QTest::addColumn<float>("z2");
QTest::addColumn<float>("w2");
QTest::newRow("null")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 0.0f << 0.0f
<< 100.0f
<< 0.0f << 0.0f << 0.0f << 0.0f;
QTest::newRow("xonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 0.0f << 0.0f << 0.0f
<< 2.0f
<< 2.0f << 0.0f << 0.0f << 0.0f;
QTest::newRow("yonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 1.0f << 0.0f << 0.0f
<< 2.0f
<< 0.0f << 2.0f << 0.0f << 0.0f;
QTest::newRow("zonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 1.0f << 0.0f
<< 2.0f
<< 0.0f << 0.0f << 2.0f << 0.0f;
QTest::newRow("wonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 0.0f << 1.0f
<< 2.0f
<< 0.0f << 0.0f << 0.0f << 2.0f;
QTest::newRow("all")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 4.0f
<< 2.0f
<< 2.0f << 4.0f << -6.0f << 8.0f;
QTest::newRow("allzero")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 4.0f
<< 0.0f
<< 0.0f << 0.0f << 0.0f << 0.0f;
}
void tst_QQuaternion::multiplyFactor()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, w1);
QFETCH(float, factor);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, w2);
QQuaternion v1(w1, x1, y1, z1);
QQuaternion v2(w2, x2, y2, z2);
QVERIFY((v1 * factor) == v2);
QVERIFY((factor * v1) == v2);
QQuaternion v3(v1);
v3 *= factor;
QVERIFY(v3 == v2);
QCOMPARE(v3.x(), v1.x() * factor);
QCOMPARE(v3.y(), v1.y() * factor);
QCOMPARE(v3.z(), v1.z() * factor);
QCOMPARE(v3.scalar(), v1.scalar() * factor);
}
// Test division by a factor for quaternions.
void tst_QQuaternion::divide_data()
{
// Use the same test data as the multiply test.
multiplyFactor_data();
}
void tst_QQuaternion::divide()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, w1);
QFETCH(float, factor);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, w2);
QQuaternion v1(w1, x1, y1, z1);
QQuaternion v2(w2, x2, y2, z2);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
if (factor == 0.0f)
return;
QVERIFY((v2 / factor) == v1);
QQuaternion v3(v2);
v3 /= factor;
QVERIFY(v3 == v1);
QCOMPARE(v3.x(), v2.x() / factor);
QCOMPARE(v3.y(), v2.y() / factor);
QCOMPARE(v3.z(), v2.z() / factor);
QCOMPARE(v3.scalar(), v2.scalar() / factor);
}
// Test negation for quaternions.
void tst_QQuaternion::negate_data()
{
// Use the same test data as the add test.
add_data();
}
void tst_QQuaternion::negate()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, w1);
QQuaternion v1(w1, x1, y1, z1);
QQuaternion v2(-w1, -x1, -y1, -z1);
QVERIFY(-v1 == v2);
}
// Test quaternion conjugate calculations.
void tst_QQuaternion::conjugate_data()
{
// Use the same test data as the add test.
add_data();
}
void tst_QQuaternion::conjugate()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, w1);
QQuaternion v1(w1, x1, y1, z1);
QQuaternion v2(w1, -x1, -y1, -z1);
QVERIFY(v1.conjugate() == v2);
}
// Test quaternion creation from an axis and an angle.
void tst_QQuaternion::fromAxisAndAngle_data()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QTest::addColumn<float>("x1");
QTest::addColumn<float>("y1");
QTest::addColumn<float>("z1");
QTest::addColumn<float>("angle");
QTest::newRow("null")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 0.0f << 0.0f;
QTest::newRow("xonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 0.0f << 0.0f << 90.0f;
QTest::newRow("yonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 1.0f << 0.0f << 180.0f;
QTest::newRow("zonly")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 0.0f << 0.0f << 1.0f << 270.0f;
QTest::newRow("complex")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 45.0f;
}
void tst_QQuaternion::fromAxisAndAngle()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, angle);
// Use a straight-forward implementation of the algorithm at:
// http://www.j3d.org/matrix_faq/matrfaq_latest.html#Q56
// to calculate the answer we expect to get.
QVector3D vector = QVector3D(x1, y1, z1).normalized();
const float a = (angle * M_PI / 180.0) / 2.0;
const float sin_a = std::sin(a);
const float cos_a = std::cos(a);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QQuaternion result(cos_a,
(vector.x() * sin_a),
(vector.y() * sin_a),
(vector.z() * sin_a));
result = result.normalized();
QQuaternion answer = QQuaternion::fromAxisAndAngle(QVector3D(x1, y1, z1), angle);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QVERIFY(qFuzzyCompare(answer.x(), result.x()));
QVERIFY(qFuzzyCompare(answer.y(), result.y()));
QVERIFY(qFuzzyCompare(answer.z(), result.z()));
QVERIFY(qFuzzyCompare(answer.scalar(), result.scalar()));
{
QVector3D answerAxis;
float answerAngle;
answer.getAxisAndAngle(&answerAxis, &answerAngle);
QVERIFY(qFuzzyCompare(answerAxis.x(), vector.x()));
QVERIFY(qFuzzyCompare(answerAxis.y(), vector.y()));
QVERIFY(qFuzzyCompare(answerAxis.z(), vector.z()));
QVERIFY(qFuzzyCompare(answerAngle, angle));
}
answer = QQuaternion::fromAxisAndAngle(x1, y1, z1, angle);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QVERIFY(qFuzzyCompare(answer.x(), result.x()));
QVERIFY(qFuzzyCompare(answer.y(), result.y()));
QVERIFY(qFuzzyCompare(answer.z(), result.z()));
QVERIFY(qFuzzyCompare(answer.scalar(), result.scalar()));
{
float answerAxisX, answerAxisY, answerAxisZ;
float answerAngle;
answer.getAxisAndAngle(&answerAxisX, &answerAxisY, &answerAxisZ, &answerAngle);
QVERIFY(qFuzzyCompare(answerAxisX, vector.x()));
QVERIFY(qFuzzyCompare(answerAxisY, vector.y()));
QVERIFY(qFuzzyCompare(answerAxisZ, vector.z()));
QVERIFY(qFuzzyCompare(answerAngle, angle));
}
}
// Test quaternion convertion to and from rotation matrix.
void tst_QQuaternion::fromRotationMatrix_data()
{
fromAxisAndAngle_data();
}
void tst_QQuaternion::fromRotationMatrix()
{
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, angle);
QQuaternion result = QQuaternion::fromAxisAndAngle(QVector3D(x1, y1, z1), angle);
QMatrix3x3 rot3x3 = result.toRotationMatrix();
QQuaternion answer = QQuaternion::fromRotationMatrix(rot3x3);
QVERIFY(qFuzzyCompare(answer, result) || qFuzzyCompare(-answer, result));
}
// Test quaternion convertion to and from orthonormal axes.
void tst_QQuaternion::fromAxes_data()
{
QTest::addColumn<float>("x1");
QTest::addColumn<float>("y1");
QTest::addColumn<float>("z1");
QTest::addColumn<float>("angle");
QTest::addColumn<QVector3D>("xAxis");
QTest::addColumn<QVector3D>("yAxis");
QTest::addColumn<QVector3D>("zAxis");
QTest::newRow("null")
<< 0.0f << 0.0f << 0.0f << 0.0f
<< QVector3D(1, 0, 0) << QVector3D(0, 1, 0) << QVector3D(0, 0, 1);
QTest::newRow("xonly")
<< 1.0f << 0.0f << 0.0f << 90.0f
<< QVector3D(1, 0, 0) << QVector3D(0, 0, 1) << QVector3D(0, -1, 0);
QTest::newRow("yonly")
<< 0.0f << 1.0f << 0.0f << 180.0f
<< QVector3D(-1, 0, 0) << QVector3D(0, 1, 0) << QVector3D(0, 0, -1);
QTest::newRow("zonly")
<< 0.0f << 0.0f << 1.0f << 270.0f
<< QVector3D(0, -1, 0) << QVector3D(1, 0, 0) << QVector3D(0, 0, 1);
QTest::newRow("complex")
<< 1.0f << 2.0f << -3.0f << 45.0f
<< QVector3D(0.728028, -0.525105, -0.440727) << QVector3D(0.608789, 0.790791, 0.0634566) << QVector3D(0.315202, -0.314508, 0.895395);
}
void tst_QQuaternion::fromAxes()
{
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, angle);
QFETCH(QVector3D, xAxis);
QFETCH(QVector3D, yAxis);
QFETCH(QVector3D, zAxis);
QQuaternion result = QQuaternion::fromAxisAndAngle(QVector3D(x1, y1, z1), angle);
QVector3D axes[3];
result.getAxes(&axes[0], &axes[1], &axes[2]);
QVERIFY(myFuzzyCompare(axes[0], xAxis));
QVERIFY(myFuzzyCompare(axes[1], yAxis));
QVERIFY(myFuzzyCompare(axes[2], zAxis));
QQuaternion answer = QQuaternion::fromAxes(axes[0], axes[1], axes[2]);
QVERIFY(qFuzzyCompare(answer, result) || qFuzzyCompare(-answer, result));
}
// Test shortest arc quaternion.
void tst_QQuaternion::rotationTo_data()
{
QTest::addColumn<QVector3D>("from");
QTest::addColumn<QVector3D>("to");
// same
QTest::newRow("+X -> +X") << QVector3D(10.0f, 0.0f, 0.0f) << QVector3D(10.0f, 0.0f, 0.0f);
QTest::newRow("-X -> -X") << QVector3D(-10.0f, 0.0f, 0.0f) << QVector3D(-10.0f, 0.0f, 0.0f);
QTest::newRow("+Y -> +Y") << QVector3D(0.0f, 10.0f, 0.0f) << QVector3D(0.0f, 10.0f, 0.0f);
QTest::newRow("-Y -> -Y") << QVector3D(0.0f, -10.0f, 0.0f) << QVector3D(0.0f, -10.0f, 0.0f);
QTest::newRow("+Z -> +Z") << QVector3D(0.0f, 0.0f, 10.0f) << QVector3D(0.0f, 0.0f, 10.0f);
QTest::newRow("-Z -> -Z") << QVector3D(0.0f, 0.0f, -10.0f) << QVector3D(0.0f, 0.0f, -10.0f);
QTest::newRow("+X+Y+Z -> +X+Y+Z") << QVector3D(10.0f, 10.0f, 10.0f) << QVector3D(10.0f, 10.0f, 10.0f);
QTest::newRow("-X-Y-Z -> -X-Y-Z") << QVector3D(-10.0f, -10.0f, -10.0f) << QVector3D(-10.0f, -10.0f, -10.0f);
// arbitrary
QTest::newRow("+Z -> +X") << QVector3D(0.0f, 0.0f, 10.0f) << QVector3D(10.0f, 0.0f, 0.0f);
QTest::newRow("+Z -> -X") << QVector3D(0.0f, 0.0f, 10.0f) << QVector3D(-10.0f, 0.0f, 0.0f);
QTest::newRow("+Z -> +Y") << QVector3D(0.0f, 0.0f, 10.0f) << QVector3D(0.0f, 10.0f, 0.0f);
QTest::newRow("+Z -> -Y") << QVector3D(0.0f, 0.0f, 10.0f) << QVector3D(0.0f, -10.0f, 0.0f);
QTest::newRow("-Z -> +X") << QVector3D(0.0f, 0.0f, -10.0f) << QVector3D(10.0f, 0.0f, 0.0f);
QTest::newRow("-Z -> -X") << QVector3D(0.0f, 0.0f, -10.0f) << QVector3D(-10.0f, 0.0f, 0.0f);
QTest::newRow("-Z -> +Y") << QVector3D(0.0f, 0.0f, -10.0f) << QVector3D(0.0f, 10.0f, 0.0f);
QTest::newRow("-Z -> -Y") << QVector3D(0.0f, 0.0f, -10.0f) << QVector3D(0.0f, -10.0f, 0.0f);
QTest::newRow("+X -> +Y") << QVector3D(10.0f, 0.0f, 0.0f) << QVector3D(0.0f, 10.0f, 0.0f);
QTest::newRow("+X -> -Y") << QVector3D(10.0f, 0.0f, 0.0f) << QVector3D(0.0f, -10.0f, 0.0f);
QTest::newRow("-X -> +Y") << QVector3D(-10.0f, 0.0f, 0.0f) << QVector3D(0.0f, 10.0f, 0.0f);
QTest::newRow("-X -> -Y") << QVector3D(-10.0f, 0.0f, 0.0f) << QVector3D(0.0f, -10.0f, 0.0f);
QTest::newRow("+X+Y+Z -> +X-Y-Z") << QVector3D(10.0f, 10.0f, 10.0f) << QVector3D(10.0f, -10.0f, -10.0f);
QTest::newRow("-X-Y+Z -> -X+Y-Z") << QVector3D(-10.0f, -10.0f, 10.0f) << QVector3D(-10.0f, 10.0f, -10.0f);
QTest::newRow("+X+Y+Z -> +Z") << QVector3D(10.0f, 10.0f, 10.0f) << QVector3D(0.0f, 0.0f, 10.0f);
// collinear
QTest::newRow("+X -> -X") << QVector3D(10.0f, 0.0f, 0.0f) << QVector3D(-10.0f, 0.0f, 0.0f);
QTest::newRow("+Y -> -Y") << QVector3D(0.0f, 10.0f, 0.0f) << QVector3D(0.0f, -10.0f, 0.0f);
QTest::newRow("+Z -> -Z") << QVector3D(0.0f, 0.0f, 10.0f) << QVector3D(0.0f, 0.0f, -10.0f);
QTest::newRow("+X+Y+Z -> -X-Y-Z") << QVector3D(10.0f, 10.0f, 10.0f) << QVector3D(-10.0f, -10.0f, -10.0f);
}
void tst_QQuaternion::rotationTo()
{
QFETCH(QVector3D, from);
QFETCH(QVector3D, to);
QQuaternion q1 = QQuaternion::rotationTo(from, to);
QVERIFY(myFuzzyCompare(q1, q1.normalized()));
QVector3D vec1(q1 * from);
vec1 *= (to.length() / from.length()); // discard rotated length
QVERIFY(myFuzzyCompare(vec1, to));
QQuaternion q2 = QQuaternion::rotationTo(to, from);
QVERIFY(myFuzzyCompare(q2, q2.normalized()));
QVector3D vec2(q2 * to);
vec2 *= (from.length() / to.length()); // discard rotated length
QVERIFY(myFuzzyCompare(vec2, from));
}
// Test quaternion creation from an axis and an angle.
void tst_QQuaternion::fromEulerAngles_data()
{
QTest::addColumn<float>("pitch");
QTest::addColumn<float>("yaw");
QTest::addColumn<float>("roll");
QTest::newRow("null")
<< 0.0f << 0.0f << 0.0f;
QTest::newRow("xonly")
<< 90.0f << 0.0f << 0.0f;
QTest::newRow("yonly")
<< 0.0f << 180.0f << 0.0f;
QTest::newRow("zonly")
<< 0.0f << 0.0f << 270.0f;
QTest::newRow("x+z")
<< 30.0f << 0.0f << 45.0f;
QTest::newRow("x+y")
<< 30.0f << 90.0f << 0.0f;
QTest::newRow("y+z")
<< 0.0f << 45.0f << 30.0f;
QTest::newRow("complex")
<< 30.0f << 240.0f << -45.0f;
}
void tst_QQuaternion::fromEulerAngles()
{
QFETCH(float, pitch);
QFETCH(float, yaw);
QFETCH(float, roll);
// Use a straight-forward implementation of the algorithm at:
// http://www.j3d.org/matrix_faq/matrfaq_latest.html#Q60
// to calculate the answer we expect to get.
QQuaternion qx = QQuaternion::fromAxisAndAngle(QVector3D(1, 0, 0), pitch);
QQuaternion qy = QQuaternion::fromAxisAndAngle(QVector3D(0, 1, 0), yaw);
QQuaternion qz = QQuaternion::fromAxisAndAngle(QVector3D(0, 0, 1), roll);
QQuaternion result = qy * (qx * qz);
QQuaternion answer = QQuaternion::fromEulerAngles(QVector3D(pitch, yaw, roll));
QVERIFY(myFuzzyCompare(answer.x(), result.x()));
QVERIFY(myFuzzyCompare(answer.y(), result.y()));
QVERIFY(myFuzzyCompare(answer.z(), result.z()));
QVERIFY(myFuzzyCompare(answer.scalar(), result.scalar()));
{
QVector3D answerEulerAngles = answer.toEulerAngles();
QVERIFY(myFuzzyCompareDegrees(answerEulerAngles.x(), pitch));
QVERIFY(myFuzzyCompareDegrees(answerEulerAngles.y(), yaw));
QVERIFY(myFuzzyCompareDegrees(answerEulerAngles.z(), roll));
}
answer = QQuaternion::fromEulerAngles(pitch, yaw, roll);
QVERIFY(myFuzzyCompare(answer.x(), result.x()));
QVERIFY(myFuzzyCompare(answer.y(), result.y()));
QVERIFY(myFuzzyCompare(answer.z(), result.z()));
QVERIFY(myFuzzyCompare(answer.scalar(), result.scalar()));
{
float answerPitch, answerYaw, answerRoll;
answer.getEulerAngles(&answerPitch, &answerYaw, &answerRoll);
QVERIFY(myFuzzyCompareDegrees(answerPitch, pitch));
QVERIFY(myFuzzyCompareDegrees(answerYaw, yaw));
QVERIFY(myFuzzyCompareDegrees(answerRoll, roll));
}
}
// Test spherical interpolation of quaternions.
void tst_QQuaternion::slerp_data()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QTest::addColumn<float>("x1");
QTest::addColumn<float>("y1");
QTest::addColumn<float>("z1");
QTest::addColumn<float>("angle1");
QTest::addColumn<float>("x2");
QTest::addColumn<float>("y2");
QTest::addColumn<float>("z2");
QTest::addColumn<float>("angle2");
QTest::addColumn<float>("t");
QTest::addColumn<float>("x3");
QTest::addColumn<float>("y3");
QTest::addColumn<float>("z3");
QTest::addColumn<float>("angle3");
QTest::newRow("first")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 90.0f
<< 1.0f << 2.0f << -3.0f << 180.0f
<< 0.0f
<< 1.0f << 2.0f << -3.0f << 90.0f;
QTest::newRow("first2")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 90.0f
<< 1.0f << 2.0f << -3.0f << 180.0f
<< -0.5f
<< 1.0f << 2.0f << -3.0f << 90.0f;
QTest::newRow("second")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 90.0f
<< 1.0f << 2.0f << -3.0f << 180.0f
<< 1.0f
<< 1.0f << 2.0f << -3.0f << 180.0f;
QTest::newRow("second2")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 90.0f
<< 1.0f << 2.0f << -3.0f << 180.0f
<< 1.5f
<< 1.0f << 2.0f << -3.0f << 180.0f;
QTest::newRow("middle")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 90.0f
<< 1.0f << 2.0f << -3.0f << 180.0f
<< 0.5f
<< 1.0f << 2.0f << -3.0f << 135.0f;
QTest::newRow("wide angle")
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
<< 1.0f << 2.0f << -3.0f << 0.0f
<< 1.0f << 2.0f << -3.0f << 270.0f
<< 0.5f
<< 1.0f << 2.0f << -3.0f << -45.0f;
}
void tst_QQuaternion::slerp()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, angle1);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, angle2);
QFETCH(float, t);
QFETCH(float, x3);
QFETCH(float, y3);
QFETCH(float, z3);
QFETCH(float, angle3);
QQuaternion q1 = QQuaternion::fromAxisAndAngle(x1, y1, z1, angle1);
QQuaternion q2 = QQuaternion::fromAxisAndAngle(x2, y2, z2, angle2);
QQuaternion q3 = QQuaternion::fromAxisAndAngle(x3, y3, z3, angle3);
QQuaternion result = QQuaternion::slerp(q1, q2, t);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QVERIFY(qFuzzyCompare(result.x(), q3.x()));
QVERIFY(qFuzzyCompare(result.y(), q3.y()));
QVERIFY(qFuzzyCompare(result.z(), q3.z()));
QVERIFY(qFuzzyCompare(result.scalar(), q3.scalar()));
}
// Test normalized linear interpolation of quaternions.
void tst_QQuaternion::nlerp_data()
{
slerp_data();
}
void tst_QQuaternion::nlerp()
{
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QFETCH(float, x1);
QFETCH(float, y1);
QFETCH(float, z1);
QFETCH(float, angle1);
QFETCH(float, x2);
QFETCH(float, y2);
QFETCH(float, z2);
QFETCH(float, angle2);
QFETCH(float, t);
QQuaternion q1 = QQuaternion::fromAxisAndAngle(x1, y1, z1, angle1);
QQuaternion q2 = QQuaternion::fromAxisAndAngle(x2, y2, z2, angle2);
QQuaternion result = QQuaternion::nlerp(q1, q2, t);
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
float resultx, resulty, resultz, resultscalar;
if (t <= 0.0f) {
resultx = q1.x();
resulty = q1.y();
resultz = q1.z();
resultscalar = q1.scalar();
} else if (t >= 1.0f) {
resultx = q2.x();
resulty = q2.y();
resultz = q2.z();
resultscalar = q2.scalar();
} else if (qAbs(angle1 - angle2) <= 180.f) {
resultx = q1.x() * (1 - t) + q2.x() * t;
resulty = q1.y() * (1 - t) + q2.y() * t;
resultz = q1.z() * (1 - t) + q2.z() * t;
resultscalar = q1.scalar() * (1 - t) + q2.scalar() * t;
} else {
// Angle greater than 180 degrees: negate q2.
resultx = q1.x() * (1 - t) - q2.x() * t;
resulty = q1.y() * (1 - t) - q2.y() * t;
resultz = q1.z() * (1 - t) - q2.z() * t;
resultscalar = q1.scalar() * (1 - t) - q2.scalar() * t;
}
QQuaternion q3 = QQuaternion(resultscalar, resultx, resulty, resultz).normalized();
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QVERIFY(qFuzzyCompare(result.x(), q3.x()));
QVERIFY(qFuzzyCompare(result.y(), q3.y()));
QVERIFY(qFuzzyCompare(result.z(), q3.z()));
QVERIFY(qFuzzyCompare(result.scalar(), q3.scalar()));
}
class tst_QQuaternionProperties : public QObject
{
Q_OBJECT
Q_PROPERTY(QQuaternion quaternion READ quaternion WRITE setQuaternion)
public:
tst_QQuaternionProperties(QObject *parent = 0) : QObject(parent) {}
QQuaternion quaternion() const { return q; }
void setQuaternion(const QQuaternion& value) { q = value; }
private:
QQuaternion q;
};
// Test getting and setting quaternion properties via the metaobject system.
void tst_QQuaternion::properties()
{
tst_QQuaternionProperties obj;
obj.setQuaternion(QQuaternion(6.0f, 7.0f, 8.0f, 9.0f));
QQuaternion q = qvariant_cast<QQuaternion>(obj.property("quaternion"));
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(q.scalar(), 6.0f);
QCOMPARE(q.x(), 7.0f);
QCOMPARE(q.y(), 8.0f);
QCOMPARE(q.z(), 9.0f);
obj.setProperty("quaternion",
QVariant::fromValue(QQuaternion(-6.0f, -7.0f, -8.0f, -9.0f)));
q = qvariant_cast<QQuaternion>(obj.property("quaternion"));
Make gui/math3d classes use float rather than qreal This corrects the mismatch between using floats for internal storage and qreal in the API of QVector*D which leads to lots of implicit casts between double and float. This change also stops users from being surprised by the loss of precision when using these classes on desktop platforms and removes the need for the private constructors taking a dummy int as the final argument. The QMatrix4x4 and QQuaternion classes have been changed to use float for their internal storage since these are meant to be used in conjunction with the QVector*D classes. This is to prevent unexpected loss of precision and to improve performance. The on-disk format has also been changed from double to float thereby reducing the storage required when streaming vectors and matrices. This is potentially a large saving when working with complex 3D meshes etc. This also has a significant performance improvement when passing matrices to QOpenGLShaderProgram (and QGLShaderProgram) as we no longer have to iterate and convert the data to floats. This is an operation that could easily be needed many times per frame. This change also opens the door for further optimisations of these classes to be implemented by using SIMD intrinsics. This needs to be applied in conjunction with https://codereview.qt-project.org/#change,33548 Task-number: QTBUG-21035 Task-number: QTBUG-20661 Change-Id: I9321b06040ffb93ae1cbd72fd2013267ac901b2e Reviewed-by: Lars Knoll <lars.knoll@nokia.com>
2012-08-20 19:55:40 +00:00
QCOMPARE(q.scalar(), -6.0f);
QCOMPARE(q.x(), -7.0f);
QCOMPARE(q.y(), -8.0f);
QCOMPARE(q.z(), -9.0f);
}
void tst_QQuaternion::metaTypes()
{
QVERIFY(QMetaType::type("QQuaternion") == QMetaType::QQuaternion);
QCOMPARE(QByteArray(QMetaType::typeName(QMetaType::QQuaternion)),
QByteArray("QQuaternion"));
QVERIFY(QMetaType::isRegistered(QMetaType::QQuaternion));
QVERIFY(qMetaTypeId<QQuaternion>() == QMetaType::QQuaternion);
}
QTEST_APPLESS_MAIN(tst_QQuaternion)
#include "tst_qquaternion.moc"