🎥 OpenGL Mathematics (glm) for `C`

The original glm library is for C++ only (templates, namespaces, classes...), this library targeted to C99 but currently you can use it for C89 safely by language extensions e.g __restrict

Documentation

Almost all functions (inline versions) and parameters are documented inside related headers.
Complete documentation: http://cglm.readthedocs.io

Note for previous versions:

_dup (duplicate) is changed to _copy. For instance glm_vec_dup -> glm_vec_copy
OpenGL related functions are dropped to make this lib platform/third-party independent
make sure you have latest version and feel free to report bugs, troubles
[bugfix] euler angles was implemented in reverse order (extrinsic) it was fixed, now they are intrinsic. Make sure that you have the latest version
[major change] by starting v0.4.0, quaternions are stored as [x, y, z, w], it was [w, x, y, z] in v0.3.5 and earlier versions
[api rename] by starting v0.4.5, glm_simd functions are renamed to glmm_
[new option] by starting v0.4.5, you can disable alignment requirement, check options in docs.

Note for C++ developers:

If you don't aware about original GLM library yet, you may also want to look at: https://github.com/g-truc/glm

Note for new comers (Important):

vec4 and mat4 variables must be aligned. (There will be unaligned versions later)
in and [in, out] parameters must be initialized (please). But [out] parameters not, initializing out param is also redundant
All functions are inline if you don't want to use pre-compiled versions with glmc_ prefix, you can ignore build process. Just incliude headers.
if your debugger takes you to cglm headers then make sure you are not trying to copy vec4 to vec3 or alig issues...
Welcome!

Note for experienced developers:

Since I'm testing this library in my projects, sometimes bugs occurs; finding that bug[s] and making improvements would be more easy with multiple developer/contributor and their projects or knowledge. Consider to make some tests if you suspect something is wrong and any feedbacks, contributions and bug reports are always welcome.

Allocations?

cglm doesn't alloc any memory on heap. So it doesn't provide any allocator. You should alloc memory for out parameters too if you pass pointer of memory location. Don't forget that vec4 (also quat/versor) and mat4 must be aligned (16-bytes), because cglm uses SIMD instructions to optimize most operations if available.

Returning vector or matrix... ?

Since almost all types are arrays and C doesn't allow returning arrays, so cglm doesn't support this feature. In the future cglm may use struct for some types for this purpose.

Other APIs like Vulkan, Metal, Dx?

Currently cglm uses default clip space configuration (-1, 1) for camera functions (perspective, extract corners...), in the future other clip space configurations will be supported

Like some other graphics libraries (especially OpenGL) this library use Column-Major layout to keep matrices in the memory.

In the future the library may support an option to use row-major layout, CURRENTLY if you need to row-major layout you will need to transpose it.

Features

general purpose matrix operations (mat4, mat3)
chain matrix multiplication (square only)
general purpose vector operations (cross, dot, rotate, proj, angle...)
affine transforms
matrix decomposition (extract rotation, scaling factor)
optimized affine transform matrices (mul, rigid-body inverse)
camera (lookat)
projections (ortho, perspective)
quaternions
euler angles / yaw-pitch-roll to matrix
extract euler angles
inline or pre-compiled function call
frustum (extract view frustum planes, corners...)
bounding box (AABB in Frustum (culling), crop, merge...)
project, unproject

You have two option to call a function/operation: inline or library call (link) Almost all functions are marked inline (alwaysinline) so compiler probably will inline. To call pre-compiled version, just use `glmc(c stands for 'call') instead ofglm_`.

  #include <cglm/cglm.h>   /* for inline */
  #include <cglm/call.h>   /* for library call (this also includes cglm.h) */

  mat4 rot, trans, rt;
  /* ... */
  glm_mul(trans, rot, rt);  /* inline */
  glmc_mul(trans, rot, rt); /* call from library */

Most of math functions are optimized manualy with SSE2 if available, if not? Dont worry there are non-sse versions of all operations

You can pass matrices and vectors as array to functions rather than get address.

  mat4 m = {
    1, 0, 0, 0,
    0, 1, 0, 0,
    0, 0, 1, 0,
    0, 0, 0, 1
  };

  glm_translate(m, (vec3){1.0f, 0.0f, 0.0f});

Library contains general purpose mat4 mul and inverse functions but also contains some special form (optimized) of these functions for affine transform matrices. If you want to multiply two affine transform matrices you can use glm_mul instead of glm_mat4_mul and glm_inv_tr (ROT + TR) instead glm_mat4_inv

/* multiplication */
mat4 modelMat;
glm_mul(T, R, modelMat);

/* othonormal rot + tr matrix inverse (rigid-body) */
glm_inv_tr(modelMat);

Contributors

This project exists thanks to all the people who contribute. [Contribute].

Backers

Thank you to all our backers! 🙏 [Become a backer]

License

MIT. check the LICENSE file

Build

Unix (Autotools)

$ sh ./build-deps.sh # run only once (dependencies) [Optional].
$ # You can pass this step if you don't want to run `make check` for tests.
$ # cglm uses cmocka for tests and it may reqiure cmake for building it
$
$ sh autogen.sh
$ ./configure
$ make
$ make check # [Optional] (if you run `sh ./build-deps.sh`)
$ [sudo] make install

Windows (MSBuild)

Windows related build files, project files are located in win folder, make sure you are inside cglm/win folder. Code Analysis are enabled, it may take awhile to build

$ cd win
$ .\build.bat

if msbuild won't work (because of multi version VS) then try to build with devenv:

$ devenv cglm.sln /Build Release

Building Docs

First you need install Sphinx: http://www.sphinx-doc.org/en/master/usage/installation.html then:

$ cd docs
$ sphinx-build source build

it will compile docs into build folder, you can run index.html inside that function.

How to use

If you want to use inline versions of funcstions then; include main header

#include <cglm/cglm.h>

the header will include all headers. Then call func you want e.g. rotate vector by axis:

glm_vec_rotate(v1, glm_rad(45), (vec3){1.0f, 0.0f, 0.0f});

some functions are overloaded :) e.g you can normalize vector:

glm_vec_normalize(vec);

this will normalize vec and store normalized vector into vec but if you will store normalized vector into another vector do this:

glm_vec_normalize_to(vec, result);

like this function you may see _to postfix, this functions store results to another variables and save temp memory

to call pre-compiled versions include header with c postfix, c means call. Pre-compiled versions are just wrappers.

#include <cglm/call.h>

this header will include all headers with c postfix. You need to call functions with c posfix:

glmc_vec_normalize(vec);

Function usage and parameters are documented inside related headers. You may see same parameter passed twice in some examples like this:

glm_mat4_mul(m1, m2, m1);

/* or */
glm_mat4_mul(m1, m1, m1);

the first two parameter are [in] and the last one is [out] parameter. After multiplied m1 and m2 the result is stored in m1. This is why we send m1 twice. You may store result in different matrix, this just an example.

Example: Computing MVP matrix

Option 1

mat4 proj, view, model, mvp;

/* init proj, view and model ... */

glm_mat4_mul(proj, view, viewProj);
glm_mat4_mul(viewProj, model, mvp);

Option 2

mat4 proj, view, model, mvp;

/* init proj, view and model ... */

glm_mat4_mulN((mat4 *[]){&proj, &view, &model}, 3, mvp);

How to send matrix to OpenGL

mat4 is array of vec4 and vec4 is array of floats. glUniformMatrix4fv functions accecpts float* as value (last param), so you can cast mat4 to float* or you can pass first column of matrix as beginning of memory of matrix:

Option 1: Send first column

glUniformMatrix4fv(location, 1, GL_FALSE, matrix[0]);

/* array of matrices */
glUniformMatrix4fv(location, 1, GL_FALSE, matrix[0][0]);

Option 2: Cast matrix to pointer type (also valid for multiple dimensional arrays)

glUniformMatrix4fv(location, 1, GL_FALSE, (float *)matrix);

You can pass same way to another APIs e.g. Vulkan, DX...

Notes

This library uses float types only, does not support Integers, Double... yet
If headers are not working properly with your compiler, IDE please open an issue, because I'm using GCC and clang to test it maybe sometimes MSVC

TODO:

[ ] Unit tests (In Progress)
[ ] Unit tests for comparing cglm with glm results
[x] Add version info
[ ] Unaligned operations (e.g. glm_umat4_mul)
[x] Extra documentation
[ ] ARM Neon Arch (In Progress)

Introduction

🎥 OpenGL Mathematics (glm) for `C`

Documentation

Note for previous versions:

Note for C++ developers:

Note for new comers (Important):

Note for experienced developers:

Allocations?

Returning vector or matrix... ?

Other APIs like Vulkan, Metal, Dx?

Features

Contributors

Backers

Sponsors

License

Build

Unix (Autotools)

Windows (MSBuild)

Building Docs

How to use

Example: Computing MVP matrix

Option 1

Option 2

How to send matrix to OpenGL

Notes

results matching ""

No results matching ""

🎥 OpenGL Mathematics (glm) for C

Documentation

Note for previous versions:

Note for C++ developers:

Note for new comers (Important):

Note for experienced developers:

Allocations?

Returning vector or matrix... ?

Other APIs like Vulkan, Metal, Dx?

Features

Contributors

Backers

Sponsors

License

Build

Unix (Autotools)

Windows (MSBuild)

Building Docs

How to use

Example: Computing MVP matrix

Option 1

Option 2

How to send matrix to OpenGL

Notes

results matching ""

No results matching ""

🎥 OpenGL Mathematics (glm) for `C`