Merge branch '0.9.5' of https://github.com/g-truc/glm into 0.9.5

glm/ext.hpp

@@ -72,6 +72,7 @@
 #include "./gtc/matrix_transform.hpp"
 #include "./gtc/noise.hpp"
 #include "./gtc/quaternion.hpp"
+#include "./gtc/dual_quaternion.hpp"
 #include "./gtc/random.hpp"
 #include "./gtc/reciprocal.hpp"
 #include "./gtc/swizzle.hpp"

glm/gtx/dual_quaternion.hpp

@@ -0,0 +1,242 @@
+///////////////////////////////////////////////////////////////////////////////////
+/// OpenGL Mathematics (glm.g-truc.net)
+///
+/// Copyright (c) 2005 - 2013 G-Truc Creation (www.g-truc.net)
+/// Permission is hereby granted, free of charge, to any person obtaining a copy
+/// of this software and associated documentation files (the "Software"), to deal
+/// in the Software without restriction, including without limitation the rights
+/// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+/// copies of the Software, and to permit persons to whom the Software is
+/// furnished to do so, subject to the following conditions:
+/// 
+/// The above copyright notice and this permission notice shall be included in
+/// all copies or substantial portions of the Software.
+/// 
+/// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+/// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+/// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+/// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+/// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+/// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+/// THE SOFTWARE.
+///
+/// @ref gtx_dual_quaternion
+/// @file glm/gtx/dual_quaternion.hpp
+/// @date 2013-02-10 / 2013-02-20
+/// @author Maksim Vorobiev (msomeone@gmail.com)
+///
+/// @see core (dependence)
+/// @see gtc_half_float (dependence)
+/// @see gtc_constants (dependence)
+/// @see gtc_quaternion (dependence)
+///
+/// @defgroup gtc_dual_quaternion GLM_GTX_dual_quaternion
+/// @ingroup gtc
+/// 
+/// @brief Defines a templated dual-quaternion type and several dual-quaternion operations.
+/// 
+/// <glm/gtx/dual_quaternion.hpp> need to be included to use these functionalities.
+///////////////////////////////////////////////////////////////////////////////////
+
+#ifndef GLM_GTX_dual_quaternion
+#define GLM_GTX_dual_quaternion GLM_VERSION
+
+// Dependency:
+#include "../glm.hpp"
+#include "../gtc/half_float.hpp"
+#include "../gtc/constants.hpp"
+#include "../gtc/quaternion.hpp"
+
+#if(defined(GLM_MESSAGES) && !defined(glm_ext))
+#	pragma message("GLM: GLM_GTX_dual_quaternion extension included")
+#endif
+
+namespace glm{
+namespace detail
+{
+    template <typename T> 
+    struct tdualquat// : public genType<T, tquat>
+    {
+        enum ctor{null};
+
+        typedef T value_type;
+        typedef glm::detail::tquat<T> part_type;
+        typedef std::size_t size_type;
+
+    public:
+        glm::detail::tquat<T> real, dual;
+
+        GLM_FUNC_DECL size_type length() const;
+
+        // Constructors
+        tdualquat();
+        explicit tdualquat(tquat<T> const & real);
+        tdualquat(tquat<T> const & real,tquat<T> const & dual);
+        tdualquat(tquat<T> const & orientation,tvec3<T> const& translation);
+
+        //////////////////////////////////////////////////////////////
+        // tdualquat conversions
+        explicit tdualquat(tmat2x4<T> const & holder_mat);
+        explicit tdualquat(tmat3x4<T> const & aug_mat);
+
+        // Accesses
+        typename part_type & operator[](int i);
+        typename part_type const & operator[](int i) const;
+
+        // Operators
+        tdualquat<T> & operator*=(value_type const & s);
+        tdualquat<T> & operator/=(value_type const & s);
+    };
+
+    template <typename T> 
+    detail::tquat<T> operator- (
+        detail::tquat<T> const & q);
+    
+    template <typename T> 
+    detail::tdualquat<T> operator+ ( 
+        detail::tdualquat<T> const & q, 
+        detail::tdualquat<T> const & p); 
+
+    template <typename T>
+    detail::tdualquat<T> operator* (
+        detail::tdualquat<T> const & q, 
+        detail::tdualquat<T> const & p); 
+
+    template <typename T> 
+    detail::tvec3<T> operator* (
+        detail::tquat<T> const & q, 
+        detail::tvec3<T> const & v);
+
+    template <typename T> 
+    detail::tvec3<T> operator* (
+        detail::tvec3<T> const & v,
+        detail::tquat<T> const & q);
+
+    template <typename T> 
+    detail::tvec4<T> operator* (
+        detail::tquat<T> const & q, 
+        detail::tvec4<T> const & v);
+
+    template <typename T> 
+    detail::tvec4<T> operator* (
+        detail::tvec4<T> const & v,
+        detail::tquat<T> const & q);
+
+    template <typename T> 
+    detail::tdualquat<T> operator* (
+        detail::tdualquat<T> const & q, 
+        typename detail::tdualquat<T>::value_type const & s);
+
+    template <typename T> 
+    detail::tdualquat<T> operator* (
+        typename detail::tdualquat<T>::value_type const & s,
+        detail::tdualquat<T> const & q);
+
+    template <typename T> 
+    detail::tdualquat<T> operator/ (
+        detail::tdualquat<T> const & q, 
+        typename detail::tdualquat<T>::value_type const & s);
+} //namespace detail
+
+    /// @addtogroup gtc_dual_quaternion
+    /// @{
+
+    /// Returns the normalized quaternion. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T> 
+    detail::tdualquat<T> normalize(
+        detail::tdualquat<T> const & q);
+
+    /// Returns the linear interpolation of two dual quaternion. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T>
+    detail::tdualquat<T> lerp (
+        detail::tdualquat<T> const & x, 
+        detail::tdualquat<T> const & y,
+        typename detail::tdualquat<T>::value_type const & a); 
+
+    /// Returns the q inverse. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T> 
+    detail::tdualquat<T> inverse(
+        detail::tdualquat<T> const & q);
+
+    /// Extracts a rotation part from dual-quaternion to a 3 * 3 matrix. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T> 
+    detail::tmat3x3<T> mat3_cast(
+        detail::tdualquat<T> const & x);
+
+    /// Converts a quaternion to a 2 * 4 matrix. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T> 
+    detail::tmat2x4<T> mat2x4_cast(
+        detail::tdualquat<T> const & x);
+
+    /// Converts a quaternion to a 3 * 4 matrix. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T> 
+    detail::tmat3x4<T> mat3x4_cast(
+        detail::tdualquat<T> const & x);
+
+    /// Converts a 2 * 4 matrix (matrix which holds real and dual parts) to a quaternion. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T> 
+    detail::tdualquat<T> dualquat_cast(
+        detail::tmat2x4<T> const & x);
+
+    /// Converts a 3 * 4 matrix (augmented matrix rotation + translation) to a quaternion. 
+    /// 
+    /// @see gtc_dual_quaternion
+    template <typename T> 
+    detail::tdualquat<T> dualquat_cast(
+        detail::tmat3x4<T> const & x);
+
+    /// Dual-quaternion of floating-point numbers. 
+    /// 
+    /// @see gtc_dual_quaternion
+    typedef detail::tdualquat<float> dualquat;
+
+    /// Dual-quaternion of half-precision floating-point numbers.
+    /// 
+    /// @see gtc_dual_quaternion
+    typedef detail::tdualquat<detail::half>	hdualquat;
+
+    /// Dual-quaternion of single-precision floating-point numbers. 
+    /// 
+    /// @see gtc_dual_quaternion
+    typedef detail::tdualquat<float>	fdualquat;
+
+    /// Dual-quaternion of double-precision floating-point numbers. 
+    /// 
+    /// @see gtc_dual_quaternion
+    typedef detail::tdualquat<double>	ddualquat;
+
+    /// Dual-quaternion of low precision floating-point numbers.
+    /// 
+    /// @see gtc_dual_quaternion
+    typedef detail::tdualquat<lowp_float>		lowp_dualquat;
+
+    /// Dual-quaternion of medium precision floating-point numbers. 
+    /// 
+    /// @see gtc_dual_quaternion
+    typedef detail::tdualquat<mediump_float>	mediump_dualquat;
+
+    /// Dual-quaternion of high precision floating-point numbers. 
+    /// 
+    /// @see gtc_dual_quaternion
+    typedef detail::tdualquat<highp_float>		highp_dualquat;
+
+    /// @}
+} //namespace glm
+
+#include "dual_quaternion.inl"
+
+#endif//GLM_GTX_dual_quaternion

glm/gtx/dual_quaternion.inl

@@ -0,0 +1,426 @@
+///////////////////////////////////////////////////////////////////////////////////
+/// OpenGL Mathematics (glm.g-truc.net)
+///
+/// Copyright (c) 2005 - 2013 G-Truc Creation (www.g-truc.net)
+/// Permission is hereby granted, free of charge, to any person obtaining a copy
+/// of this software and associated documentation files (the "Software"), to deal
+/// in the Software without restriction, including without limitation the rights
+/// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+/// copies of the Software, and to permit persons to whom the Software is
+/// furnished to do so, subject to the following conditions:
+/// 
+/// The above copyright notice and this permission notice shall be included in
+/// all copies or substantial portions of the Software.
+/// 
+/// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+/// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+/// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+/// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+/// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+/// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+/// THE SOFTWARE.
+///
+/// @ref gtx_quaternion
+/// @file glm/gtx/quaternion.inl
+/// @date 2013-02-10 / 2013-02-13
+/// @author Maksim Vorobiev (msomeone@gmail.com)
+///////////////////////////////////////////////////////////////////////////////////
+
+#include <limits>
+
+namespace glm{
+namespace detail
+{
+    template <typename T>
+    GLM_FUNC_QUALIFIER GLM_CONSTEXPR typename tdualquat<T>::size_type tdualquat<T>::length() const
+    {
+        return 8;
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T>::tdualquat() : 
+          real(tquat<T>()),
+          dual(tquat<T>(tdualquat<T>::value_type(0),tdualquat<T>::value_type(0),tdualquat<T>::value_type(0),tdualquat<T>::value_type(0)))
+    {}
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T>::tdualquat
+    (
+        tquat<T> const & r
+    ) :
+        real(r),
+        dual(tquat<T>(tdualquat<T>::value_type(0),tdualquat<T>::value_type(0),tdualquat<T>::value_type(0),tdualquat<T>::value_type(0)))
+    {}
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T>::tdualquat
+    (
+        tquat<T> const & r,
+        tquat<T> const & d
+    ) :
+        real(r),
+        dual(d)
+    {}
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T>::tdualquat
+    (
+        tquat<T> const & q,
+        tvec3<T> const& p
+    ) :
+        real(q),
+        dual(-0.5f*( p.x*q.x + p.y*q.y + p.z*q.z),
+              0.5f*( p.x*q.w + p.y*q.z - p.z*q.y),
+              0.5f*(-p.x*q.z + p.y*q.w + p.z*q.x),
+              0.5f*( p.x*q.y - p.y*q.x + p.z*q.w))
+    {}
+
+    //////////////////////////////////////////////////////////////
+    // tdualquat conversions
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T>::tdualquat
+    (
+        tmat2x4<T> const & holder_mat
+    ) 
+    {
+        *this = dualquat_cast<>
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T>::tdualquat
+    (
+        tmat3x4<T> const & m
+    )
+    {
+        *this = dualquat_cast(m);
+    }
+
+    //////////////////////////////////////////////////////////////
+    // tdualquat<T> accesses
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER typename tdualquat<T>::part_type & tdualquat<T>::operator [] (int i)
+    {
+        return (&real)[i];
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER typename tdualquat<T>::part_type const & tdualquat<T>::operator [] (int i) const
+    {
+        return (&real)[i];
+    }
+
+    //////////////////////////////////////////////////////////////
+    // tdualquat<valType> operators
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T> & tdualquat<T>::operator *=
+    (
+        value_type const & s
+    )
+    {
+        this->real *= s;
+        this->dual *= s;
+        return *this;
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER tdualquat<T> & tdualquat<T>::operator /=
+    (
+        value_type const & s
+    )
+    {
+        this->real /= s;
+        this->dual /= s;
+        return *this;
+    }
+
+    //////////////////////////////////////////////////////////////
+    // tquat<valType> external operators
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> operator-
+    (
+        detail::tdualquat<T> const & q
+    )
+    {
+        return detail::tdualquat<T>(-this->real,-this->dual);
+    }
+
+    template <typename T>
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> operator+
+    (
+        detail::tdualquat<T> const & q, 
+        detail::tdualquat<T> const & p
+    )
+    {
+        return detail::tdualquat<T>(q.real + p.real,q.dual + p.dual);
+    }
+
+    template <typename T>
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> operator*
+    (
+        detail::tdualquat<T> const & p, 
+        detail::tdualquat<T> const & o
+    )
+    {
+        return detail::tdualquat<T>(p.real * o.real,p.real * o.dual + p.dual * o.real);
+    }
+
+    // Transformation
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tvec3<T> operator* 
+    (
+        detail::tdualquat<T> const & q, 
+        detail::tvec3<T> const & v
+    )
+    {
+        const detail::tvec3<T> real_v3(q.real.x,q.real.y,q.real.z);
+        const detail::tvec3<T> dual_v3(q.dual.x,q.dual.y,q.dual.z);
+        return (cross(real_v3, cross(real_v3,v) + v * q.real.w + dual_v3) + dual_v3 * q.real.w - real_v3 * q.dual.w) * detail::tdualquat<T>::value_type(2) + v;    
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tvec3<T> operator* 
+    (
+        detail::tvec3<T> const & v,
+        detail::tdualquat<T> const & q
+    )
+    {
+        return inverse(q) * v;
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tvec4<T> operator* 
+    (
+        detail::tdualquat<T> const & q, 
+        detail::tvec4<T> const & v
+    )
+    {
+        return detail::tvec4<T>(q * detail::tvec3<T>(v), v.w);
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tvec4<T> operator* (
+        detail::tvec4<T> const & v,
+        detail::tdualquat<T> const & q)
+    {
+        return inverse(q) * v;
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> operator* 
+    (
+        detail::tdualquat<T> const & q, 
+        typename detail::tdualquat<T>::value_type const & s
+    )
+    {
+        return detail::tdualquat<T>(q.real * s, q.dual * s);
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> operator* 
+    (
+        typename detail::tdualquat<T>::value_type const & s,
+        detail::tdualquat<T> const & q 
+    )
+    {
+        return q * s;
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> operator/ 
+    (
+        detail::tdualquat<T> const & q, 
+        typename detail::tdualquat<T>::value_type const & s
+    )
+    {
+        return detail::tdualquat<T>(q.real / s, q.dual / s);
+    }
+
+    //////////////////////////////////////
+    // Boolean operators
+    template <typename T> 
+    GLM_FUNC_QUALIFIER bool operator==
+    (
+        detail::tdualquat<T> const & q1, 
+        detail::tdualquat<T> const & q2
+    )
+    {
+        return (q1.real == q2.real) && (q1.dual == q2.dual);
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER bool operator!=
+    (
+        detail::tdualquat<T> const & q1, 
+        detail::tdualquat<T> const & q2
+    )
+    {
+        return (q1.real != q2.dual) || (q1.real != q2.dual);
+    }
+}//namespace detail
+
+    ////////////////////////////////////////////////////////
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> normalize
+    (
+        detail::tdualquat<T> const & q
+    )
+    {
+        return q / length(q.real);
+    }
+
+    template <typename T>
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> lerp
+    (
+        detail::tdualquat<T> const & x, 
+        detail::tdualquat<T> const & y, 
+        typename detail::tdualquat<T>::value_type const & a
+    )
+    { // Dual Quaternion Linear blend aka DLB:
+        // Lerp is only defined in [0, 1]
+        assert(a >= T(0));
+        assert(a <= T(1));
+        const detail::tdualquat<T>::value_type k = dot(x.real,y.real) < detail::tdualquat<T>::value_type(0) ? -a : a;
+        const detail::tdualquat<T>::value_type one(1);
+        return detail::tdualquat<T>(x * (one - a) + y * k);
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> inverse
+    (
+        detail::tdualquat<T> const & q
+    )
+    {
+        const glm::detail::tquat<T> real = conjugate(q.real);
+        const glm::detail::tquat<T> dual = conjugate(q.dual);
+        return detail::tdualquat<T>(real, dual + (real * (-2.0f * dot(real,dual))));
+    }
+    
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tmat3x3<T> mat3_cast
+    (
+        detail::tdualquat<T> const & x
+    )
+    {
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tmat2x4<T> mat2x4_cast
+    (
+        detail::tdualquat<T> const & x
+    )
+    {
+        return detail::tmat2x4<T>( x[0].x, x[0].y, x[0].z, x[0].w, x[1].x, x[1].y, x[1].z, x[1].w );
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tmat3x4<T> mat3x4_cast
+    (
+        detail::tdualquat<T> const & x
+    )
+    {
+        detail::tquat<T> r = x.real / length2(x.real);
+        
+        const detail::tquat<T> rr(r.w * x.real.w, r.x * x.real.x, r.y * x.real.y, r.z * x.real.z);
+        r *= detail::tdualquat<T>::value_type(2);
+
+        const detail::tdualquat<T>::value_type xy = r.x*d.real.y;
+        const detail::tdualquat<T>::value_type xz = r.x*d.real.z;
+        const detail::tdualquat<T>::value_type yz = r.y*d.real.z;
+        const detail::tdualquat<T>::value_type wx = r.w*d.real.x;
+        const detail::tdualquat<T>::value_type wy = r.w*d.real.y;
+        const detail::tdualquat<T>::value_type wz = r.w*d.real.z;
+
+        const detail::tvec4<T> a(  
+                                    rr.w + rr.x - rr.y - rr.z, 
+                                    xy - wz,
+                                    xz + wy,
+                                  -(x.dual.w * r.x - x.dual.x * r.w + x.dual.y * r.z - x.dual.z * r.y) 
+        );
+
+        const detail::tvec4<T> b(
+                                    xy + wz, 
+                                    rr.w + rr.y - rr.x - rr.z, 
+                                    yz - wx,
+                                  -(x.dual.w * r.y - x.dual.x * r.z - x.dual.y * r.w + x.dual.z * r.x)
+        );
+
+        const detail::tvec4<T> c(
+                                    xz - wy, 
+                                    yz + wx, 
+                                    rr.w + rr.z - rr.x - rr.y,
+                                  -(x.dual.w * r.z + x.dual.x * r.y - x.dual.y * r.x - x.dual.z * r.w)
+        );
+
+        return detail::tmat3x4<T>(a,b,c);
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> dualquat_cast
+    (
+        detail::tmat2x4<T> const & x
+    )
+    {
+        return detail::tdualquat (
+            detail::tquat<T> ( x[0].w, x[0].x, x[0].y, x[0].z ),
+            detail::tquat<T> ( x[1].w, x[1].x, x[1].y, x[1].z )
+        );
+    }
+
+    template <typename T> 
+    GLM_FUNC_QUALIFIER detail::tdualquat<T> dualquat_cast
+    (
+        detail::tmat3x4<T> const & x
+    )
+    {
+        detail::tquat<T> real;
+
+        const detail::tdualquat<T>::value_type trace = x[0].x + x[1].y + x[2].z;
+        if(trace > detail::tdualquat<T>::value_type(0))
+        {
+            const detail::tdualquat<T>::value_type r = sqrt(detail::tdualquat<T>::value_type(1) + trace);
+            const detail::tdualquat<T>::value_type invr = detail::tdualquat<T>::value_type(0.5) / r;
+            real.w = detail::tdualquat<T>::value_type(0.5) * r;
+            real.x = (x[2].y - x[1].z) * invr;
+            real.y = (x[0].z - x[2].x) * invr;
+            real.z = (x[1].x - x[0].y) * invr;
+        }
+        else if(x[0].x > x[1].y && x[0].x > x[2].z)
+        {
+            const detail::tdualquat<T>::value_type r = sqrt(detail::tdualquat<T>::value_type(1) + x[0].x - x[1].y - x[2].z);
+            const detail::tdualquat<T>::value_type invr = detail::tdualquat<T>::value_type(0.5) / r;
+            real.x = detail::tdualquat<T>::value_type(0.5)*r;
+            real.y = (x[1].x + x[0].y) * invr;
+            real.z = (x[0].z + x[2].x) * invr;
+            real.w = (x[2].y - x[1].z) * invr;
+        }
+        else if(x[1].y > x[2].z)
+        {
+            const detail::tdualquat<T>::value_type r = sqrt(detail::tdualquat<T>::value_type(1) + x[1].y - x[0].x - x[2].z);
+            const detail::tdualquat<T>::value_type invr = detail::tdualquat<T>::value_type(0.5) / r;
+            x = (x[1].x + x[0].y) * invr;
+            y = detail::tdualquat<T>::value_type(0.5) * r;
+            z = (x[2].y + x[1].z) * invr;
+            w = (x[0].z - x[2].x) * invr;
+        }
+        else
+        {
+            const detail::tdualquat<T>::value_type r = sqrt(detail::tdualquat<T>::value_type(1) + x[2].z - x[0].x - x[1].y);
+            const detail::tdualquat<T>::value_type invr = detail::tdualquat<T>::value_type(0.5) / r;
+            x = (x[0].z + x[2].x) * invr;
+            y = (x[2].y + x[1].z) * invr;
+            z = detail::tdualquat<T>::value_type(0.5) * r;
+            w = (x[1].x - x[0].y) * invr;
+        }
+
+        const detail::tquat<T> dual;
+        dual.x =  0.5f*( x[0].w*real.w + x[1].w*real.z - x[2].w*real.y);
+        dual.y =  0.5f*(-x[0].w*real.z + x[1].w*real.w + x[2].w*real.x);
+        dual.z =  0.5f*( x[0].w*real.y - x[1].w*real.x + x[2].w*real.w);
+        dual.w = -0.5f*( x[0].w*real.x + x[1].w*real.y + x[2].w*real.z);
+        return detail::tdualquat<T>(real,dual);
+    }
+
+}//namespace glm

test/gtx/CMakeLists.txt

@@ -5,6 +5,7 @@ glmCreateTestGTC(gtx_matrix_interpolation)
 glmCreateTestGTC(gtx_matrix_query)
 glmCreateTestGTC(gtx_multiple)
 glmCreateTestGTC(gtx_quaternion)
+glmCreateTestGTC(gtx_dual_quaternion)
 glmCreateTestGTC(gtx_rotate_normalized_axis)
 glmCreateTestGTC(gtx_rotate_vector)
 glmCreateTestGTC(gtx_scalar_relational)

test/gtx/gtx_dual_quaternion.cpp

@@ -0,0 +1,173 @@
+///////////////////////////////////////////////////////////////////////////////////////////////////
+// OpenGL Mathematics Copyright (c) 2005 - 2013 G-Truc Creation (www.g-truc.net)
+///////////////////////////////////////////////////////////////////////////////////////////////////
+// Created : 2013-02-10
+// Updated : 2013-02-11
+// Licence : This source is under MIT licence
+// File    : test/gtc/gtc_dual_quaternion.cpp
+///////////////////////////////////////////////////////////////////////////////////////////////////
+
+#include <glm/glm.hpp>
+#include <glm/gtx/dual_quaternion.hpp>
+#include <glm/gtc/matrix_transform.hpp>
+#include <glm/gtc/epsilon.hpp>
+#include <glm/gtx/euler_angles.hpp>
+
+#include <iostream>
+
+int myrand()
+{
+    static int holdrand = 1;
+    return (((holdrand = holdrand * 214013L + 2531011L) >> 16) & 0x7fff);
+}
+
+float myfrand() // returns values from -1 to 1 inclusive
+{
+    return float(double(myrand()) / double( 0x7ffff )) * 2.0f - 1.0f;
+}
+
+int test_dquat_type()
+{
+    glm::dvec3 vA;
+    glm::dquat dqA,dqB;
+    glm::ddualquat C(dqA,dqB);
+    glm::ddualquat B(dqA);
+    glm::ddualquat D(dqA,vA);
+    return 0;
+}
+
+int test_scalars() {
+    float const Epsilon = 0.0001f;
+
+    int Error(0);
+
+    glm::quat src_q1 = glm::quat(1.0f,2.0f,3.0f,4.0f);
+    glm::quat src_q2 = glm::quat(5.0f,6.0f,7.0f,8.0f);
+    glm::dualquat src1(src_q1,src_q2);
+
+    {
+        glm::dualquat dst1 = src1 * 2.0f;
+        glm::dualquat dst2 = 2.0f * src1;
+        glm::dualquat dst3 = src1;
+        dst3 *= 2.0f;
+        glm::dualquat dstCmp(src_q1 * 2.0f,src_q2 * 2.0f);
+        Error += glm::all(glm::epsilonEqual(dst1.real,dstCmp.real, Epsilon)) && glm::all(glm::epsilonEqual(dst1.dual,dstCmp.dual, Epsilon)) ? 0 : 1;
+        Error += glm::all(glm::epsilonEqual(dst2.real,dstCmp.real, Epsilon)) && glm::all(glm::epsilonEqual(dst2.dual,dstCmp.dual, Epsilon)) ? 0 : 1;
+        Error += glm::all(glm::epsilonEqual(dst3.real,dstCmp.real, Epsilon)) && glm::all(glm::epsilonEqual(dst3.dual,dstCmp.dual, Epsilon)) ? 0 : 1;
+    }
+
+    {
+        glm::dualquat dst1 = src1 / 2.0f;
+        glm::dualquat dst2 = src1;
+        dst2 /= 2.0f;
+        glm::dualquat dstCmp(src_q1 / 2.0f,src_q2 / 2.0f);
+        Error += glm::all(glm::epsilonEqual(dst1.real,dstCmp.real, Epsilon)) && glm::all(glm::epsilonEqual(dst1.dual,dstCmp.dual, Epsilon)) ? 0 : 1;
+        Error += glm::all(glm::epsilonEqual(dst2.real,dstCmp.real, Epsilon)) && glm::all(glm::epsilonEqual(dst2.dual,dstCmp.dual, Epsilon)) ? 0 : 1;
+    }
+    return Error;
+}
+
+int test_inverse() 
+{
+    int Error(0);
+
+    float const Epsilon = 0.0001f;
+
+    glm::dualquat dqid;
+    glm::mat4x4 mid(1.0f);
+
+    for (int j = 0; j < 100; ++j) {
+        glm::mat4x4 rot = glm::yawPitchRoll(myfrand() * 360.0f, myfrand() * 360.0f, myfrand() * 360.0f);
+        glm::vec3 vt = glm::vec3(myfrand() * 10.0f, myfrand() * 10.0f, myfrand() * 10.0f);
+
+        glm::mat4x4 m = glm::translate(mid, vt) * rot;
+
+        glm::quat qr = glm::quat_cast(m);
+
+        glm::dualquat dq(qr);
+        
+        glm::dualquat invdq = glm::inverse(dq);
+
+        glm::dualquat r1 = invdq * dq;
+        glm::dualquat r2 = dq * invdq;
+
+        Error += glm::all(glm::epsilonEqual(r1.real, dqid.real, Epsilon)) && glm::all(glm::epsilonEqual(r1.dual, dqid.dual, Epsilon)) ? 0 : 1;
+        Error += glm::all(glm::epsilonEqual(r2.real, dqid.real, Epsilon)) && glm::all(glm::epsilonEqual(r2.dual, dqid.dual, Epsilon)) ? 0 : 1;
+
+        // testing commutative property
+        glm::dualquat r (   glm::quat( myfrand() * glm::pi<float>() * 2.0f, myfrand(), myfrand(), myfrand() ),
+                            glm::vec3(myfrand() * 10.0f, myfrand() * 10.0f, myfrand() * 10.0f) );
+        glm::dualquat riq = (r * invdq) * dq;
+        glm::dualquat rqi = (r * dq) * invdq;
+
+        Error += glm::all(glm::epsilonEqual(riq.real, rqi.real, Epsilon)) && glm::all(glm::epsilonEqual(riq.dual, rqi.dual, Epsilon)) ? 0 : 1;
+    }
+
+    return Error;
+}
+
+int test_mul() 
+{
+    int Error(0);
+
+    float const Epsilon = 0.0001f;
+
+    glm::mat4x4 mid(1.0f);
+
+    for (int j = 0; j < 100; ++j) {
+        // generate random rotations and translations and compare transformed by matrix and dualquats random points 
+        glm::vec3 vt1 = glm::vec3(myfrand() * 10.0f, myfrand() * 10.0f, myfrand() * 10.0f);
+        glm::vec3 vt2 = glm::vec3(myfrand() * 10.0f, myfrand() * 10.0f, myfrand() * 10.0f);
+
+        glm::mat4x4 rot1 = glm::yawPitchRoll(myfrand() * 360.0f, myfrand() * 360.0f, myfrand() * 360.0f);
+        glm::mat4x4 rot2 = glm::yawPitchRoll(myfrand() * 360.0f, myfrand() * 360.0f, myfrand() * 360.0f);
+        glm::mat4x4 m1 = glm::translate(mid, vt1) * rot1;
+        glm::mat4x4 m2 = glm::translate(mid, vt2) * rot2;
+        glm::mat4x4 m3 = m2 * m1;
+        glm::mat4x4 m4 = m1 * m2;
+
+        glm::quat qrot1 = glm::quat_cast(rot1);
+        glm::quat qrot2 = glm::quat_cast(rot2);
+
+        glm::dualquat dq1 = glm::dualquat(qrot1,vt1);
+        glm::dualquat dq2 = glm::dualquat(qrot2,vt2);
+        glm::dualquat dq3 = dq2 * dq1;
+        glm::dualquat dq4 = dq1 * dq2;
+
+        for (int i = 0; i < 100; ++i) {
+            glm::vec4 src_pt = glm::vec4(myfrand() * 4.0f, myfrand() * 5.0f, myfrand() * 3.0f,1.0f);
+            // test both multiplication orders        
+            glm::vec4 dst_pt_m3  = m3 * src_pt; 
+            glm::vec4 dst_pt_dq3 = dq3 * src_pt;
+         
+            glm::vec4 dst_pt_m3_i  = glm::inverse(m3) * src_pt;
+            glm::vec4 dst_pt_dq3_i = src_pt * dq3;
+
+            glm::vec4 dst_pt_m4  = m4 * src_pt;
+            glm::vec4 dst_pt_dq4 = dq4 * src_pt;
+
+            glm::vec4 dst_pt_m4_i  = glm::inverse(m4) * src_pt;
+            glm::vec4 dst_pt_dq4_i = src_pt * dq4;
+
+            Error += glm::all(glm::epsilonEqual(dst_pt_m3, dst_pt_dq3, Epsilon)) ? 0 : 1;
+            Error += glm::all(glm::epsilonEqual(dst_pt_m4, dst_pt_dq4, Epsilon)) ? 0 : 1;
+            Error += glm::all(glm::epsilonEqual(dst_pt_m3_i, dst_pt_dq3_i, Epsilon)) ? 0 : 1;
+            Error += glm::all(glm::epsilonEqual(dst_pt_m4_i, dst_pt_dq4_i, Epsilon)) ? 0 : 1;
+        }
+    } 
+
+    return Error;
+}
+
+int main()
+{
+    int Error(0);
+
+    Error += test_dquat_type();
+    Error += test_scalars();
+    Error += test_inverse();
+    Error += test_mul();
+
+    //std::cout << "Errors count: " << Error << std::endl;
+    return Error;
+}