Skip to content

Latest commit

 

History

History
64 lines (38 loc) · 4.76 KB

glm.md

File metadata and controls

64 lines (38 loc) · 4.76 KB
Raw

GLM Reference Doc

GLM (OpenGL Mathematics) is a C++ library that replicates much of the mathematics functionality of GLSL. Nearly all types and functions in the GLM core are identical to their equivalents in GLSL in terms of naming conventions and behavior (although there are a few differences), making it particularly easy to use alongside GLSL.

GLM Headers

The header file glm/glm.hpp includes the entire GLM core. If you only need certain features, they can be found in the following headers.

  • glm/vec2.hpp, glm/vec3.hpp, glm/vec4.hpp: GLSL's vector types.

  • glm/mat2x2.hpp, glm/mat2x3.hpp, glm/mat2x4.hpp, glm/mat3x2.hpp, glm/mat3x3.hpp, glm/mat3x4.hpp, glm/mat4x2.hpp, glm/mat4x3.hpp, glm/mat4x4.hpp: GLSL's matrix types. Note that while the GLSL type for a square 2x2 matrix (for example) is mat2, the corresponding GLM header name is mat2x2.hpp. Once the type has been included, however, glm::mat2 can still be used as an alias.

  • glm/common.hpp: GLSL's common functions, including min(), max(), abs(), mod(), floor(), ceil(), clamp(), etc. These functions operate component-wise on vectors.

  • glm/exponential.hpp: GLSL's exponential functions, including exp(), log(), sqrt(), pow(), etc. These functions operate component-wise on vectors.

  • glm/integer.hpp: GLSL's integer functions.

  • glm/matrix.hpp: GLSL's matrix functions, including transpose(), inverse(), and determinant().

  • glm/packing.hpp: GLSL's packing functions, which can convert between vectors and numeric types.

  • glm/trigonometric.hpp: GLSL's trig functions, including sin(), asin(), sinh(), and the equivalents for cos and tan, as well as the conversion functions degrees() and radians(). These functions operate component-wise on vectors, and all except radians() expect their arguments to be in radians.

  • glm/vector_relational.hpp: GLSL's vector comparison functions. Most of these, such as lessThan() and greaterThanEqual(), take two vectors of the same length and return a boolean vector (bvec) of the same length containing the results of each comparison. not() inverts a bvec, and any() and all() are used to obtain bools from bvecs.

Full GLM core documentation can be found here: http://glm.g-truc.net/0.9.7/api/a00162.html

GTC and GTX

GTC and GTX contain stable (GTC) and experimental (GTX) extensions to GLM that add functionality beyond what exists in GLSL. There are a number of extensions, but some of the most useful for this course are:

  • glm/gtx/transform.hpp: Contains the functions translate(), rotate(), and scale(), for constructing standard 4x4 transformation matrices. translate() and scale() each accept a vec3 containing the amount of translation/scaling in all three dimensions, while rotate() accepts an angle in radians and a vec3 to use as the axis of rotation.

  • glm/gtc/constants.hpp: Contains a number of useful mathematical constants, such as pi() and root_two().

  • glm/gtc/type_ptr.hpp: Contains the value_ptr() function, which returns a pointer to the start of a vector or matrix’s data (see the “Passing GLM Types to GLSL” section below). Also contains a number of functions such as make_mat2x4() that construct the specified type from a value pointer.

Documentation for all extensions can be found here: http://glm.g-truc.net/0.9.7/api/a00160.html (GTC) and here: http://glm.g-truc.net/0.9.7/api/a00161.html (GTX)

Swizzle Functions

GLM supports vector swizzling, as in GLSL, but the behavior here is slightly different. In GLSL, both of the following lines accomplish the same thing (switching the x and y components), where v1 and v2 are vec2s:

v2 = v1.yx;
v2.yx = v1;

In GLM, swizzling is handled by functions, so you must make sure to include the parentheses. The functions return a new vector without modifying the old one, so they will not work on the left-hand side of assignments:

v2 = v1.yx(); //same result as in GLSL
v2.yx() = v1; //does nothing

To enable swizzle functions, you must #define GLM_SWIZZLE before any GLM includes.

Passing GLM Types to GLSL

GLM vectors and matrices can easily be passed to shaders via the corresponding glUniform function: glUniform4fv() for a vec4, glUniformMatrix2x3fv() for a mat2x3, etc. For the value pointer parameter, you need a pointer to the start of the vector or matrix data, which can be obtained by dereferencing the first element of the vector or matrix, like so:

glUniformMatrix3fv(location, 1, GL_FALSE, &matrix[0][0]);

Alternatively, you can use value_ptr(), found in glm/gtc/type_ptr.hpp, which accomplishes the same thing:

glUniformMatrix3fv(location, 1, GL_FALSE, glm::value_ptr(matrix));