Transforming models

In the Rendering models section of the Quickstart guide you have learnt how to draw models on screen, but we have not explained you have you can control its position, scale and orientation in space.

When a mesh is rendered, each of its vertices goes through a series of matrix multiplications before it finally gets on the screen. These matrices are knows as the world, view and projection matrices. While the view and the projection matrix control where is the camera, which direction is it looking and how is the 3D scene projected onto the screen (i.e. orthographic or perspective projection), the world matrix is the one that moves, rotates and scales the models.

The world matrix

GML comes with a few built-in functions using which you can control mentioned matrices and you can find them in the official manual. As you may have noticed, we are using them in the Rendering models section too:

matrix_set(matrix_world, matrix_build(x, y, z, 0, 0, direction, 1, 1, 1));
modSword.submit();

The code above creates a new matrix using matrix_build, which leaves the scale of the model at 1, rotates the model around the Z axis and moves it to position [x, y, z]. The created matrix is then set as the current world matrix using matrix_set(matrix_world, ...) and then the model is rendered using the method submit. This is the general idea behind transforming models.

At some point you should also "reset" the world matrix to an identity matrix. Simply put, an identity matrix is a matrix that does not change a matrix/vector that it is multiplied with - the result of the multiplication is the same matrix/vector. After you call the following line of code, the things drawn will have their original position, rotation and scale:

matrix_set(matrix_world, matrix_build_identity());

Matrix utilities

The example code above is pretty straight forward, but sometimes you may need to do more complicated transformations that require multiple matrix multiplications. In that case matrix_build and matrix_multiply can feel a little clunky. For that reason BBMOD comes with a struct BBMOD_Matrix, which simplifies chained transformations into method calls. Here is how would you rewrite the code above using BBMOD_Matrix:

new BBMOD_Matrix()
    .RotateZ(direction)
    .Translate(x, y, z)
    .ApplyWorld();

The code became much more self-explanatory, without having to learn in which specific order is translation, rotation and scale executed when you use matrix_build. Not to mention that you do not need to fill in all 9 arguments even in case you just wanted to move a model on one axis. For the full list of BBMOD_Matrix methods, please see its documentation.

Do you find this page helpful?