The model is not what the printer reads
A printer never sees your STL. It reads a list of coordinated movements with extrusion amounts attached. Slicing is the translation, and almost every quality decision happens during it.
Step one: orientation
Before anything is cut, the model is rotated. This one choice determines strength direction, where supports land, which surfaces come out smooth, and how long the print takes. We orient for load path first, then look at what that costs cosmetically.
Step two: layers
The model is intersected with a stack of horizontal planes. Each intersection is a set of closed loops. Thinner layers mean more of them: a 100 mm part at 0.2 mm is 500 layers, at 0.05 mm it is 2000, and roughly four times the print time.
Step three: perimeters and infill
Each loop is offset inward to make perimeter walls, usually two to four. What remains is filled with a repeating pattern at some density. Gyroid infill is our default for functional parts because it is isotropic and prints without sharp direction changes.
Step four: supports
Any region overhanging more than about 45 degrees needs something beneath it. Support placement is the most human part of slicing. Automatic generation is fast and puts supports on faces you care about. On detailed work we place them by hand.
Step five: the toolpath
Finally the loops become movements with speeds, accelerations, retraction and temperature commands. Seam placement, where each perimeter loop starts and ends, gets hidden on an internal corner if the geometry allows.
Why we do not just press the button
Every default in a slicer is a compromise tuned for an average part. Your part is not average. The difference between an acceptable print and a good one is almost always in these decisions rather than in the machine.