Every pixel a computer draws is placed by math. The coordinate systems that say where things are, and the matrices that move them, are the first concepts to understand in graphics programming.
Pixel coordinates
The simplest coordinate system is the pixel grid of a screen or image. Positions are pairs of numbers, usually x across and y down, with the origin at the top-left corner in most screen-oriented systems. Different APIs and file formats can place the origin elsewhere, such as bottom-left, so reading the convention of the tool in hand is always the first step.
Normalized coordinates
Above the pixel grid, graphics systems often use normalized coordinates, where the visible area is expressed as a range such as zero to one regardless of the actual pixel size. Normalized device coordinates let a scene be described once and then mapped onto any resolution, which is why the same drawing code works on different screen sizes.
Transformations as matrices
Move, scale, and rotate are transformations, and each can be written as a matrix: a small grid of numbers that is multiplied with a point's coordinates to produce its new position. The advantage of the matrix form is composition. Several transformations can be multiplied together into a single matrix that does all of them at once, applied the same way to millions of vertices.
Why 4x4 matrices
In three-dimensional work, transformation matrices are usually 4 by 4. The fourth component, the homogeneous coordinate, is the trick that lets translation share the same matrix form as rotation and scale. Without it, moving an object would need special-case arithmetic; with it, every transformation is one kind of object that composes cleanly.
Model, view, projection
Transforms are conventionally organized in three layers. The model transform places an object within the world. The view transform positions the camera and expresses the world relative to it. The projection transform flattens the three-dimensional scene onto the two-dimensional screen. Keeping the layers separate makes cameras, objects, and lenses independently adjustable.
Right-handed versus left-handed
Three-dimensional APIs differ in their coordinate handedness, the convention that fixes which way axes point and which direction is a positive rotation. Some ecosystems are right-handed, others left-handed, and importing a model or a camera setting from one into the other can mirror the result. The practical skill is not memorizing every convention but checking which one each tool uses, and being consistent within a project.