How many top layers should I use?

Think in millimetres, not in layers. You want roughly 0.8mm of solid material on the top surface, so the layer count depends on your layer height: four layers at 0.2mm, six or seven at 0.12mm, three at 0.28mm. Most slicers expose this directly as a minimum shell thickness, and setting it there means the count corrects itself whenever you change layer height. Four at 0.2mm is the number that solves most pillowing on its own. Going far beyond about a millimetre of solid buys nothing — the first two layers do the bridging work, and everything above them is just smoothing.

What is pillowing on a 3D print?

Pillowing is the rippled, lumpy, sometimes perforated top surface you get when the first solid layer sags between the infill lines that are supposed to carry it. That layer is not printing on anything — it is bridging across the open cells of your sparse infill — so it is briefly molten plastic hanging in air. If it does not freeze before it droops, it bulges upward into small pillows as it cools and drags the layers above out of flat. It is the same physics as a sagging overhang, and it is caused by weak part cooling, too-sparse infill, or too few solid layers, in that order of likelihood.

Does higher infill fix holes in the top surface?

Often, but it is rarely the cheapest fix. Infill is the scaffolding the first solid layer bridges across, so raising it from 10% to 20% roughly halves the unsupported span and the layer has far less distance to sag over. The catch is that infill costs time and material through the entire part to fix a defect in the last millimetre of it. Adding one or two top layers, or turning the part cooling fan up, usually solves the same problem for a fraction of the print time. Raise infill when you have already done both and the surface is still dimpled.

Why is my top layer rough and ridged instead of holey?

That is over-extrusion, which is the opposite problem and needs the opposite fix. When each line lays down slightly more plastic than the space allows, the excess has nowhere to go but up, so adjacent lines pile against each other and the nozzle drags through the ridges it just made. Check three things in order: the flow or extrusion multiplier, the filament diameter in the slicer against what your calipers actually measure, and the nozzle temperature, since a too-hot melt spreads wider than its commanded width. Print a single-wall calibration cube and measure the wall — that number tells you the multiplier directly.

Can ironing fix a bad top surface?

Ironing improves a good top surface and hides nothing on a bad one. It works by running the nozzle back across the finished top with almost no extrusion, remelting the surface so the lines flow into each other, and it produces a genuinely glassy finish on flat upward faces. What it cannot do is fill holes: there is no material being added, so a gap stays a gap and the hot nozzle simply drags over it. Fix the extrusion, the cooling and the layer count first, then switch ironing on as a finish. Expect it to add real time to large flat tops, and expect a small blob where each pass starts.

When is a top surface defect worth outsourcing?

When the surface is the product. Cosmetic prototypes, display models, client-facing mock-ups and anything that gets photographed are all jobs where FDM's layered top surface is a ceiling rather than a setting — you can tune it until it is flat and it will still read as printed under raking light. Resin and powder-bed processes do not have a sparse-infill-to-solid transition at all, so the defect on this page cannot occur on them. A shop that runs both will quote the process that fits the finish you need; the directory at /providers lists providers by location and process.

Why 3D Print Top Layers Have Holes and Pillowing

Nine hours in, the walls are clean, the dimensions are right, and then the last millimetre ruins it. The 3D print top layers came out rippled and lumpy, or there are small holes scattered across the flat face, or you can see the infill pattern telegraphing through a surface that is supposed to be solid. It is the most disproportionate defect in FDM printing: two or three per cent of the part destroys the look of all of it.

The good news is that the top surface is a narrow, well-understood transition — the moment the printer stops laying sparse infill and starts trying to close it over — and each way it fails looks different. Read yours below, then work down the fixes. Five of the six cost nothing but a slicer setting.

This post contains affiliate links. If you purchase through these links, 3D Prototyping Hub may earn a small commission at no extra cost to you.

This post also contains Amazon affiliate links. As an Amazon Associate we earn from qualifying purchases.

Read the Top Surface First

Put the part under a bright light at a low angle and look along the surface rather than down at it.

What you see	What it actually is	Section
Small holes in a regular pattern, infill visible through them	Too few solid layers over too sparse an infill	1, 2
Rippled, bulging, quilted — no holes, but nothing flat	Pillowing: the bridged layer sagged before it froze	3
Rough and ridged everywhere, lines piled against each other	Over-extrusion	4
Fine parallel gaps between lines that should be touching	Under-extrusion or a partial clog	5
Fuzzy, pitted, furry — and it hissed or popped while printing	Wet filament	6
Flat and sound, but dull and visibly striped	Normal FDM. This is what ironing is for	7

The distinction that matters most is holes versus ripples. Holes mean the solid layers ran out before they closed. Ripples mean the first one never sat flat in the first place, which is a cooling problem and not a layer-count one.

1. Count Solid Layers in Millimetres, Not in Layers

"Three top layers" is not a setting, it is a coincidence. What the part needs is about 0.8mm of solid material on top, and how many layers that takes depends entirely on your layer height — four at 0.2mm, six or seven at 0.12mm, three at 0.28mm. Someone drops their layer height to get a finer finish, the top-layer count stays at three, and the solid skin silently thins from 0.6mm to 0.36mm. That is when the holes appear.

Most slicers have a minimum shell or top-shell thickness field that enforces the millimetre figure and recalculates the count for you. Set it there and it stops being something you can forget.

Nozzle diameter moves the same number. A 0.6mm nozzle lays wider, thicker lines, so it closes a top surface in fewer layers but resolves a coarser one; a 0.4mm nozzle needs more layers to reach the same thickness. If you change nozzles regularly, how nozzle diameter changes a print covers the rest of that trade.

2. Infill Is the Scaffolding, and Sparse Infill Is a Long Bridge

The first solid layer is not printed on anything. It is bridged across the open cells of the infill below, and the width of those cells is set by your infill percentage. At 10% infill on a typical part, the nozzle is being asked to span several millimetres of nothing; at 25%, it spans a fraction of that. Same plastic, same fan, completely different odds.

So if extra top layers did not fix it, raise infill rather than stacking more solid on top of a hole. Two refinements worth knowing:

Pattern matters as much as density. Grid, lines and triangles present the top layer with a lattice of straight edges to land on. Gyroid at the same percentage gives it curved, wandering contact — beautiful inside the part, less helpful directly under a flat top.
Most slicers can densify only where it matters. Look for a solid-infill threshold, a top-infill support setting, or an internal-dense-layers option, which adds a denser band immediately under the top skin and leaves the rest of the part at 10%. It fixes the surface without paying for infill through the whole print — and print time is worth protecting, as why 3D prints take so long goes into.

3. Pillowing Is a Cooling Failure

If the top is rippled rather than holed, stop changing layer counts. That layer bridged across the infill as molten plastic, and molten plastic hanging in air does exactly what you would expect: it droops, and as it cools it pulls itself upward into little pillows between the supports. This is the same physics as sagging overhangs, on a smaller and flatter scale.

The fix is air, and in this order:

Confirm the part cooling fan is actually running at full speed over the top layers. Not the hotend fan — the one aimed at the print. Many profiles ramp the fan down for speed or for large layers, and the top skin is exactly where you do not want that.
Look at the duct. A printed duct that has warped, cracked, or sagged toward the heater block is a fan that is blowing hot air sideways. This is a common, invisible failure on machines that print PLA in a warm room all summer.
Replace the blower if it stalls under load. A blower that spins freely by hand and stalls against back-pressure moves almost no air. A 5015 blower and a fresh duct is one of the cheapest genuine upgrades on a budget machine, and top-surface quality is where you notice it first.
Slow the top solid layers down. Give the fan more time over the same distance. Dropping top-surface speed to around half your normal solid-infill speed is usually enough.

The exception is ABS, ASA and polycarbonate, where cranking the fan cracks the part in a different way. With those, take the cooling fix off the table and solve it with layer count and infill density instead: five or six top layers and 25% infill, printed slowly in a warm chamber.

4. Rough and Ridged Means Too Much Plastic

If the top has no holes at all but feels and looks like corduroy that has been trodden on, you are over-extruding. Each line is slightly wider than the space the slicer allotted it, the surplus is pushed up rather than out, and the nozzle then ploughs through the ridges it has just laid.

Calibrate it rather than nudging it. Print a single-wall calibration cube — one perimeter, no top, no infill — and measure the wall in several places with a set of digital calipers. Divide the extrusion width your slicer commanded by the number you measured and multiply your flow by that ratio. Ten minutes, and the result is a number rather than a feeling.

Two related settings deserve the same skepticism. Check the filament diameter in your slicer against what the calipers say the filament actually is, averaged over several points on the spool. And check nozzle temperature: a melt at the top of the material's range spreads wider than one in the middle, so a profile inherited from a hot-running filament over-extrudes on a cool-running one.

5. Fine Gaps Between Lines Means Too Little

The opposite symptom — thin, even, parallel slots between lines that should have fused — is under-extrusion arriving at the surface where you finally notice it. The lines are narrower than commanded, so they never touch, and on a top layer there is nothing above to hide it.

That has its own diagnostic chain, covered in why 3D prints under-extrude: extruder tension, a slipping or worn drive gear, retraction settings, and volumetric flow at high speed. Top surfaces tend to show it first because solid infill demands the most consistent flow of anything in the print.

The cause worth ruling out immediately is a partial clog. A nozzle that is 80% clear prints walls that look fine and solid layers that do not, because the walls are slow and the solid infill is fast. A cleaning kit and a cold pull rules it out in fifteen minutes, and why nozzles clog covers what put the debris there in the first place.

6. Fuzzy and Pitted Means Wet Filament

A surface that is pockmarked, hairy and slightly rough everywhere — rather than failing in a pattern that follows the infill — is usually moisture. Water absorbed into the spool flashes to steam inside the hotend, blowing tiny craters in the extruded line and making the flow inconsistent from one second to the next. If the printer hisses, ticks or pops while running, that is the diagnosis confirmed.

Nylon, PETG, TPU and PVA drink from the air fastest; PLA takes longer but gets there, and a spool that has sat out for a winter is not the same material it was when you bought it. The same moisture is what makes a spool snap in the feed path.

Drying it is the only fix that works: a heated filament dryer at the material's recommended temperature for several hours, and printing straight out of it for anything hygroscopic. Sealed storage afterwards preserves the result but cannot produce it — our filament dry box guide covers both halves.

7. Then, and Only Then, Turn On Ironing

Once the surface is flat and sound, ironing is what makes it glassy. The nozzle runs back over the finished top with the extruder nearly stopped, remelting the skin so adjacent lines flow together. On flat upward faces the difference is dramatic.

It is a finish, not a repair. There is no new material, so it cannot close a hole — it just drags a hot nozzle over one. Expect it to add noticeable time on large flat tops, expect a small blob where each ironing pass begins, and drop the flow to a couple of per cent if you see material building up. For curved and angled tops, where ironing cannot reach, the answer is sanding and filling instead: that is the territory of our finishing and painting supplies guide.

When the Top Surface Is the Product

There is a line past which none of this is a settings problem. A layered top surface tuned to its best is still a layered top surface, and under raking light in a photograph it reads as printed. For a client-facing model, a display piece or a cosmetic prototype, that can be the whole verdict on the part.

Resin and powder-bed processes never transition from sparse infill to solid skin, so the defect described on this page cannot happen on them — which is the real reason cosmetic work moves off FDM rather than any argument about detail. The trade-offs are laid out in SLA vs FDM printing, and the shops that run both are in the provider directory, where you can compare them by location and process before you spend another nine hours finding out.

Hero photograph by Inés Álvarez Fdez via Unsplash.