When transforming a teapot (or some other object) through a series of translation/rotation/scaling operations, it would be less efficient to apply these transformations one by one to each vertex, i.e., K matrix-vector products for K transformation operations. Instead, the model matrix stores the product of all transformation matrices, so that performing the transformation requires a single matrix-vector product.
How do we compute the model matrix?
When transforming a teapot (or some other object) through a series of translation/rotation/scaling operations, it would be less efficient to apply these transformations one by one to each vertex, i.e., K matrix-vector products for K transformation operations. Instead, the model matrix stores the product of all transformation matrices, so that performing the transformation requires a single matrix-vector product.