Print the same model twice on the same machine, with the same slicer profile and the same filament. Give one of them a surface texture before you slice, and leave the other one smooth. Put both under a desk lamp. The smooth one looks 3D printed. The textured one looks like an object. Nothing about the print quality changed. The layer height, the wall count, the temperatures were identical.
That gap has bothered me for years, and the explanation turned out to have almost nothing to do with 3D printing. It is about how vision works.
Your eyes are a regularity detector
Human vision is very good at one specific thing: finding a regular pattern against a plain background. This is not an accident, it is most of what the early visual system is for. Edges, repetition, and periodic structure get amplified. Flat, featureless regions get ignored.
A layer line is close to a worst case for this. It is a periodic signal, evenly spaced ridges at a fixed pitch, sitting on a background with no other structure at all. High regularity, zero competing detail. That is exactly the configuration your visual system is tuned to flag. The ridges are physically tiny, often a fraction of a millimetre, but they are the only organised thing on the surface, so they win all of your attention.
The analogy that made it click for me is audio. A single repeating tone is glaringly obvious in a silent room and almost inaudible in a busy cafe. The tone did not get quieter in the cafe. The background got louder, and the tone stopped being the only structured thing in the signal. Layer lines are the tone. A smooth print is the silent room.
Two ways to fix a salient signal
Once you frame it as a salient periodic signal on an empty background, there are exactly two levers.
You can attack the signal. That is sanding, and it is expensive in the literal sense. The work scales with surface area, it is manual, it only reaches faces you can physically get an abrasive onto, and it is destructive. Every pass removes real geometry, so on anything organic you are trading layer lines for lost detail. You are spending effort to subtract structure, and structure is the thing you paid to print.
Or you can attack the background. Fill the empty surface with competing, irregular detail so the periodic layer signal is no longer the only pattern present. In signal terms you are raising the noise floor until the periodic component stops being salient. Wood grain, stone, stipple, knurling, any relief with no fixed pitch of its own. The layer lines are still physically there. They just stop being the thing your eye locks onto, because now they are competing with a surface full of structure instead of sitting alone.
The second lever is strictly cheaper. Adding texture is a preprocessing step on the mesh, it applies to every face regardless of reachability, and it is non destructive to the model’s intent. You are adding noise, which is easy, instead of removing signal, which is hard.
The world is full of this trick
Look at objects that are made of one material and never read as cheap or unfinished.
A knurled tool handle is covered in a deliberate cross pattern. You never inspect an individual groove. A milled coin edge is a ring of fine ridges that reads as quality, not as a defect. Camouflage works by the same principle in reverse: it hides a clear shape by drowning it in high frequency clutter so your regularity detector cannot lock on. Cast metal, leather, stone, bark. All of them are busy everywhere, and none of them show a single flaw, because there is no empty background for a flaw to stand against.
A 3D print with a full surface texture joins that category. A smooth print does not, and no amount of profile tuning moves it there, because the problem was never the quality of the layers.
Where this lands for anyone still sanding
I am not anti sanding. For a flat display part with a modern smooth aesthetic, sanding and priming gives you something texture cannot, and vapour smoothing is great on the right polymer. Those are real, and they are the minority of what most people print.
For everything else, the reframe is worth internalising. Stop treating the printed look as a defect to remove after the fact and start treating it as an empty surface to fill before the fact. Practically that means baking a texture into the geometry before you slice, so the finish is part of the print instead of a chore that comes after it. I do this in the browser now with a tool that displaces the mesh directly, though Blender with a displacement modifier gets you to the same place if you already live there. The specific tool matters far less than the change in framing.
The layers were never the problem. The empty space around them was.