I am looking for a way to apply non-affine/perspective 3D transformations to images within an HTML canvas. I have tried using matrix3d() on a div element, but since matrix3d() can only be used on DOM elements, I am struggling to find a solution to apply these transformations to objects inside the HTML canvas. To accomplish this, I need to develop a script that can replicate the functionality of the matrix3d() transform. In particular, I am seeking clarity on the significance of the 16 values passed as arguments in the matrix3d() function and the mathematical operations they undergo.