Why Is My Top Surface Rough and Puffy? Pillowing, Diagnosed in 3 Questions

The walls of your print are clean. The corners are sharp. Then the top face comes out like a quilted cushion — soft domes and sagging patches in a suspiciously neat grid, as if the surface had been stuffed. That defect has a name: pillowing. In a large analysis of community help posts it was the second most reported 3D printing problem of all — 1,917 of 20,000 posts, nearly tied with first-layer adhesion — yet, as the dataset's author notes, it gets a fraction of the coverage other problems get. This guide is the specific advice that is hard to find: three questions that route you to the cause, then the fix for that cause, with numbers for your filament.

A note on how this guide was written: this is a research-based troubleshooting guide compiled from published sources — Simplify3D's troubleshooting guide, Prusa's published filament parameters and stock profiles, a peer-reviewed infill study, and the FixMyPrint dataset — plus clearly labelled community advice. We did not personally test these steps on a printer. Where a number is community practice rather than a manufacturer or official figure, we say so.

Applies to: any FDM printer (bedslinger, CoreXY or delta) · PLA, PETG, ABS/ASA · PrusaSlicer, Bambu Studio, OrcaSlicer or Cura (setting names vary slightly). Evidence: manufacturer parameters, published troubleshooting guides, a peer-reviewed infill study and the FixMyPrint dataset — full sources at the end.

In this guide: Is it actually pillowing? · The 3-question diagnosis · Branch A: Support · Branch B: Cooling · Branch C: Heat and speed · Target numbers, by filament · FAQ

Key Takeaways

• The pattern is the diagnosis. Pillows follow the infill grid and appear only on top faces over a hollow interior. If your walls are rough too, it is probably not pillowing — see Step 0.
• The mechanism: air trapped in infill pockets warms and expands, pushing the still-soft top skin up into domes. A thin skin over wide, unsupported spans cannot resist.
• Three levers, in order: support (top layers and infill), cooling (a fan that is actually blowing), heat (nozzle temperature and top-layer speed).
• Working targets: 5–6 top layers at 0.2 mm (at least 1.0 mm of solid skin), 15–20% infill under large flat tops, and a pocket-breaking infill pattern — cubic or gyroid.
• Cooling is filament-specific. PLA wants the fan at 100% on top layers. ABS and ASA do not — on those, support is the lever, not the fan.
• Ironing does not fix pillowing. It re-heats the skin over the same trapped air. See What NOT to Do.

Step 0: Is It Actually Pillowing?

Four other faults get called "pillowing" and sent down this guide's fixes, where they do not belong. Match your surface first:

What you see	What it most likely is	Where to start
Domes, blisters or sagging patches that follow the infill pattern; only on top faces; walls clean	Pillowing — trapped hot air lifting a thin, under-supported top skin	This guide, three questions below
Thin, gappy, weak lines everywhere — walls and perimeters as well as the top; parts feel fragile	Under-extrusion — not enough plastic anywhere, not a top-surface problem	Extrusion checks (flow, partial clog), not top layers
Small random raised dots, often near the seam or retraction points, on any surface including walls	Blobs and zits — a retraction/seam issue; the dots ignore the infill pattern	Retraction and seam settings
Long streaks or drag lines in one direction across the final face; surface flat but scored — and ironing was switched on	Ironing marks — a finishing-pass problem, not a support problem	Ironing flow/speed, or switch ironing off
Uniform ripples and crowded, overlapping lines across the whole top, including solid areas and over the walls	Over-extrusion on the top — too much plastic, everywhere it lands	Flow rate, not top layers

One quick confirmation for row 1: true pillowing lives over the hollow interior — the pillows sit in the bays between infill lines. If the roughness also covers areas that are solid all the way down, you are in one of the other rows.

The 3-Question Diagnosis

Answer in order. Each question either routes you to a branch or clears the way to the next one.

Question 1: Is it only on large, flat top surfaces? If yes — small, narrow or curved tops come out fine — span and support are the prime suspects: go to Branch A first. If it happens on small tops too, check the fan question (Question 3) early: small tops print fast, layer on layer, with almost no time to cool.

Question 2: Are you running 4 or fewer top layers, or 15% or less infill? Check the actual numbers in your slicer, not the profile's name. If yes to either, pillowing is the expected outcome on a large flat top — Branch A is your fix, and the cheapest one in this guide. If both numbers already meet the table's targets, move on.

Question 3: Is your part-cooling fan actually blowing on the top layers? Not "is it enabled" — is air physically arriving? Ducts clog with filament wisps and fans weaken as they age. If the fan is off, low, or moving little air when the top layers print, go to Branch B. If support (Branch A) and cooling (Branch B) both check out and the pillows persist, the skin is staying soft too long — Branch C: heat and speed.

Why It Happens (30 Seconds of Physics)

Your part is a shell around a mostly hollow grid. As the top layers print, each bay of the grid becomes a sealed pocket of air, kept warm by the bed and the fresh plastic above it. Warm air expands; the top skin — a few tenths of a millimetre of just-printed plastic — is still soft, so the air pushes it up into a dome, and it cools in that shape. A peer-reviewed study of infill geometry describes the same mechanism — hot air rising to push the top layer up — and found cubic infill, which closes the interior into small cells instead of long connected channels, least prone to it: the lowest pillowing incidence of the patterns tested. Thin skins and wide spans make it worse: the skin must bridge from infill line to infill line, and the wider the gap and thinner the skin, the less it takes to lift it. Every fix below attacks one of three ingredients — the air's heat, the skin's softness, or the span.

Branch A: Support — the Skin Is Too Thin, or the Spans Too Wide

Signs: pillows map neatly onto the infill bays; worst in the middle of large flat areas, better near the walls where support is dense; you are at 3–4 top layers and/or 10–15% infill.

Fixes, in order:

1. Thicken the skin. Set 5–6 top layers at 0.2 mm layer height — a total top thickness of 1.0–1.2 mm. Published floors agree: All3DP's common-problems guide sets the minimum at 0.5 mm of solid top and around 6 layers, and Simplify3D names too few top layers as a primary cause. Think in millimetres: at 0.12 mm layers you need about 9 for the same 1.0 mm skin.
2. Feed the spans. If your infill is at 15% or below, raise it to 20% for the test print. Community experience on Prusa's forum is blunt — 10% is too low for large flat surfaces, 15–20% is the fix zone — and Simplify3D lists low infill among pillowing's main causes (typical range 10–25%). You are not filling the part; you are shortening the bridges the first top layer must cross.
3. Break up the pockets. Switch the infill pattern to cubic or gyroid. Long, connected patterns (lines, grid) leave long air channels that act as one big pocket. Cubic closes the interior into small cells — the pattern the MDPI study found least prone to pillowing — and gyroid does the same job.
4. If the skin separates from the walls (gaps at the perimeters rather than domes), raise the infill/perimeter overlap from the typical 15% to 20–30% — Simplify3D's adjustment and range.

Work them in this order and re-test with the same model — or faster, a small flat-topped test plate sliced at your real settings. Steps 1 and 2 settle most cases; change all four at once and you will not know which one did it.

Expected result: on the same test print, the domes flatten out; anything left is shallow and even rather than a quilted grid. If it still pillows at 5–6 top layers over 20% cubic or gyroid, the skin is staying soft too long — go to Branch B and prove the fan before changing anything else.

Branch B: Cooling — the Skin Never Gets to Set

Signs: Branch A numbers are already at target; pillows are worse on the side of the part furthest from the fan duct, or worse inside an enclosure (for PLA); the fan sounds weak or looks slow.

Fixes, in order:

1. Prove the fan, don't assume it. Command the part-cooling fan to 100% from the printer's menu and confirm air actually leaves the duct — a strip of tissue held under it should flutter hard. Clear any filament wisps and check the fan spins freely. A fan that turns but moves little air is a failed fan for this purpose.
2. Set the fan for the top layers, per filament. For PLA: 100% — Prusa's stock profile value, and All3DP's pillowing guidance is fans to top speed while top surfaces print. For PETG: up to 50% (stock profile: 30–50%). For ABS/ASA: leave it low, 0–30% — blasting those trades pillowing for warped, weakly bonded layers; their fix is Branch A plus an enclosure.
3. Check the slicer isn't holding the fan down. Profiles ramp the fan up over the first layers — correct at the bed, wrong if a "disable fan for first layers" setting or a low maximum-fan cap is still throttling it hundreds of layers later, when the top prints. By then it should be at your filament's full setting from the table below.

Expected result: with the fan proven and set for your filament, the next test top sets flatter — the side furthest from the duct improves most. If nothing changes with a verified fan at the right setting, heat is the remaining lever: Branch C.

Branch C: Heat and Speed — the Skin Stays Soft Too Long

Signs: support and cooling are both verified, and it still pillows; your nozzle temperature sits at the hot end of your filament maker's range; small parts with very short layer times are rough on top as well.

Fixes, in order:

1. Drop the nozzle temperature 5°C and re-test. One step at a time, staying inside the maker's published range (see the table). Hotter plastic stays liquid longer and heats the pocket air more. The 5–10°C downward step is the standard tuning move in manufacturers' bridging guidance, and it applies here for the same reason: the top layer is a bridge that has to freeze.
2. Slow the top layers down for the test. Simplify3D lists high speed among pillowing's causes: a skin laid fast over a wide span gets dragged before it sets. Halving top-layer speed for one test print (community practice) separates "too fast" from "too hot".
3. Give small tops time. On small parts each top layer takes seconds, so every pass lands on a still-hot previous one. If your slicer has a minimum-layer-time setting, use it — each pass should land on a surface that has set.

Expected result: each 5°C step down visibly reduces the doming on the test print until the top sets flat. If you reach the bottom of the maker's temperature range and it still pillows, stop tuning settings — the cause is physical; see "When to Stop and Look at the Hardware" below.

The Target Numbers, by Filament

Everything above, in one table — the settings to test from when a large flat top pillows, not a claim that one magic profile fits every printer.

Setting	PLA	PETG	ABS / ASA
Top solid layers (at 0.2 mm layer height)	5–6	5–6	5–6
Total top thickness	1.0–1.2 mm	1.0–1.2 mm	1.0–1.2 mm
Infill minimum (large flat tops)	15–20%	15–20%	15–20%
Infill pattern	Cubic or gyroid	Cubic or gyroid	Cubic or gyroid
Nozzle temperature	215°C	230–250°C	250–265°C
Bed temperature	50–60°C	70–90°C	100–110°C
Part-cooling fan (top layers)	100%	Up to 50%	0–30% — enclosure, not fan

Where the numbers come from — read this before treating the table as law. Nozzle and bed temperatures are Prusa's published parameters for its own filaments (PLA: 215°C / 50–60°C; PETG: 230–250°C / 70–90°C; ASA/ABS: 250–265°C / 100–110°C). Your spool's label outranks this table — brands differ, and Branch C's rule is to move in 5°C steps inside your maker's range. Fan figures are PrusaSlicer's stock profile values. Top-layer, infill and pattern targets are guide consensus and community practice, anchored on the floors quoted in Branch A. Notice what does not change by filament: skin thickness and support — those are universal; temperature and cooling are the filament-specific dials.

What the Community Data Actually Confirms

The FixMyPrint dataset (20,000+ community help posts, nearly 60,000 fix suggestions, threads classified by problem and by whether the asker later confirmed a fix) puts rough top surface and pillowing at 1,917 posts, the #2 problem overall. It does publish a confirmed-fix rate for the category: 18.6% of rough-top threads end with the asker confirming a fix. For scale, the best-confirmed category in the dataset — first-layer adhesion — is 23.3%.

Two honest readings of that number. First, it is a floor, not a failure rate: the dataset's author is explicit that most people fix their print and never come back to say so, so 18.6% confirmed does not mean 81.4% unsolved — it means at least 18.6% provably solved, the rest unknown. Second, and this matters more: the dataset does not publish a fix-by-fix breakdown for pillowing. (For stringing, per-fix rates do exist — our stringing guide ranked its steps by them.) For pillowing, nobody can honestly tell you "adding top layers fixes X% of cases" — that number does not exist in the largest dataset we could find. So this guide orders its fixes by mechanism and official guidance, not an invented success rate. If another site quotes you a precise pillowing cure rate, ask them for their dataset.

What NOT to Do

• Do not iron a pillowed top. Ironing drags a hot nozzle slowly across the final face; over trapped air pockets it re-heats the skin and the air under it, and community reports describe the puffing getting worse. Ironing finishes a top that is already flat but lined (Step 0, row 4) — it is not a cure for domes.
• Do not "fix" it with flow rate. Uniform ripples are the over-extrusion look-alike; flow is its lever. Cutting flow for true pillowing starves an already thin skin and widens the gaps between lines.
• Do not raise the nozzle temperature to melt the pillows flat. More heat means a softer skin and hotter pocket air — both ingredients of the defect, increased.
• Do not blast ABS or ASA with the part fan. You will trade a rough top for warped corners and splitting layers; their cooling row in the table is low on purpose.
• Do not keep adding top layers while leaving 10% grid infill under a dinner-plate lid. Skin and support are halves of the same fix; past about 1.2 mm, more skin buys almost nothing if the spans stay wide.
• Do not change five settings at once. If the print improves, you will never know which change did it.

When to Stop and Look at the Hardware

If the table's settings are in place, all three branches are worked, and the same model pillows in two different filaments, stop tuning settings. The likely causes are now physical: a part-cooling fan that spins but moves little air (Branch B's tissue test is the check), a damaged or worn nozzle laying down uneven lines, or a corrupted slicer profile. That last one has a free test: slice the same model with your printer maker's stock profile, untouched. If it prints a clean top, your profile was the fault — rebuild from stock, adding changes back one at a time. And if it pillows on one model only, suspect the model itself: some designs have a top shell that is only a token thickness in the file, and no slicer setting fully compensates for a top face the designer made too thin. One more routing note: if the "pillows" are on the bottom face of the part, it is not pillowing at all — that is a first-layer problem, and it belongs in our Ender 3 first-layer guide.

FAQ

Q: How many top layers do I need to stop pillowing?
A: Work in thickness, not count: aim for 1.0–1.2 mm of solid top — 5–6 layers at 0.2 mm layer height, about 9 at 0.12 mm. The published floor (All3DP) is 0.5 mm and roughly 6 layers; the target here sits above it because large flat tops are the hard case. If you reach 1.2 mm and it still pillows at 20% cubic infill with the fan verified, the cause is in Branch C or the hardware section, not the layer count.

Q: Can pillowing happen at 100% infill?
A: Classic pillowing needs the air pockets, and at 100% infill there are none — the interior is solid, so there is no trapped air to lift the skin and no span to bridge. A rough top at 100% infill is a different fault wearing the same name: start with the over-extrusion row of the Step 0 table (flow rate) and check nozzle temperature.

Q: Does ironing fix a rough, puffy top?
A: No — it treats a different defect. Ironing smooths fine lines on a top that is already flat and supported. Over a pillowed top it presses a hot nozzle onto a warm-air dome and can puff it further. Fix the pillowing first (branches A–C); iron afterwards if fine lines remain.

Q: Why does it only happen on big flat tops, when the same settings print small parts perfectly?
A: Pocket size and span. A small top is mostly held up by the walls around it and bridges only short gaps, so the skin survives. A big flat top seals a large volume of warm air under a thin skin and asks that skin to cross the widest spans on the part — exactly the conditions in the physics section. This is why "it worked on the test cube" proves so little: the failure scales with the flat area.

Summary

Diagnose before you tune: confirm the pillows follow the infill and only the infill (Step 0) → ask the three questions → Branch A: 5–6 top layers (1.0–1.2 mm), 15–20% infill, cubic or gyroid, overlap 20–30% if the edges gap → Branch B: prove the fan blows, then set it by filament (PLA 100%, PETG up to 50%, ABS/ASA low) → Branch C: nozzle down in 5°C steps inside the maker's range, top layers slowed. One branch at a time, the same test print. This is the second most reported printing problem there is, and the least covered — which is why a diagnosis beats a list.

Sources: FixMyPrint — 20,000-post dataset analysis (problem counts, confirmed-fix rates, "fraction of the coverage" observation); Simplify3D Print Quality Troubleshooting Guide — pillowing (republished by 3D Prima: causes; overlap 15% → 20–30%); All3DP — 16 Common 3D Printing Problems (pillowing: fans to top speed, ≥6 top layers, ≥0.5 mm solid; copy on Scribd); Infrastructures (MDPI), peer-reviewed — infill geometry and the pillowing mechanism; cubic infill's lowest pillowing incidence; Solid Print3D — Why are there holes in my 3D print? (pillowing causes and fix order); Prusament PLA technical parameters (215°C / 50–60°C); Prusa's published PETG (230–250°C / 70–90°C) and ASA/ABS (250–265°C / 100–110°C) parameters; 3dx.info — slicer bridging/cooling guide (Prusa stock profile fan values: PLA 100%, PETG 30–50%, ABS 20–30%); community advice, labelled as such in the text: Prusa forum — "Corner Bubbling and Rough" (10% infill too low for large flat tops; 15–20% fix zone).

Related guides: First Layer Won't Stick on Your Ender 3? 8 Things to Check Before Touching Settings · Stringing after changing PLA filament? Check these 5 things first · How to dry PLA filament: every manufacturer's numbers, compared