What glm::normalize does?

Question

I trying to write an MD5 loader in Java from C++ source but I cannot find out what is this line doing:

animatedJoint.m_Orient = glm::normalize(animatedJoint.m_Orient);

where the animatedJoint.m_Orient is vec4. What does it do?

Answer 1

It's normalizing the animatedJoint.m_Orient vector, by taking the normal of the vector and copying it back to the vector itself. The glm::normalize() method does not modify the object you pass to it.

Answer 2

What glm::normalize does?

Short answer: It normalizes a vector i.e. sets length to 1.

Details

A normalized vector is one which is often used just to denote pure directions without bothering about the magnitude (set to 1; hence their other, more common name unit vector) i.e. how far the vector pushes doesn't matter but in what direction does it point/push matters. This also simplifies computations -- both on paper and the machine (e.g. dot products become purely cosine's result, omission of division by length, etc.)

If v = <v.x, v.y, v.z> some non-unit vector i.e. a vector of length/magnitude not equal to 1, then to get normalized(v), we've to divide each of its component by its length.

vec3 normalize(const vec3 &v)
{
   float length_of_v = sqrt((v.x * v.x) + (v.y * v.y) + (v.z * v.z));
   return vec3(v.x / length_of_v, v.y / length_of_v, v.z / length_of_v);
}

An older term for a unit vector is direction cosines. Say vector v makes an angle α with X-axis, β with Y-axis and γ with Z-axis then its direction cosines or the unit vector along v is given by <cos α, cos β, cos γ>. This is helpful when we don't know the components of v but its angles with the cardinal axes.

The reason cosine function and unit vectors are related will be clear with a simple example in 2D which can be extended to higher dimensions. Say for a vector

v = <3, 4> = 3i + 4j (3 units along X-axis and 4 units along Y-axis)

we're to find the unit vector u along v.

length of v = √(3² + 4²) = 5
u = <3/5, 4/5>

Now the X component (along basis i) 3/5 is nothing but the length along the X-axis (adjacent) divided by the length of the vector (hypotenuse), since cos α = adj/hyp = 3/5, we would've arrived at the same result if we'd known α. The same holds for Y component (along basis j) too, which is nothing but cos β, where β is with respect to the Y-axis, or if you want to measure it with respect to the X-axis, then it'll be 90-β which is nothing but α, that's the reason we've v = <cos α, sin α>, the abscissa and ordinate of a point on the unit circle, the vector from the origin to a point on the circle with length (radius) 1.

Answer 3

Normalizes a vector, ie scales its elements so that returned vectors length is 1. Many graphic related functions require passed vectors to be normalized.

Answer 4

You can read more (and find answer) about this library here:

• home page: http://glm.g-truc.net/
• manual: https://glm.g-truc.net/0.9.3/glm-0.9.3.pdf
• method API description: https://glm.g-truc.net/0.9.2/api/index.html

It will help you to understand what this library is and how does it work.