How long will a 3D print take? The honest math
Every "how long does a 3D print take" answer online says it depends and stops there. Here's the actual model: print time is deposited plastic divided by your printer's real average flow rate, plus fixed overhead. Three numbers. This tool measures the first from your STL and stocks realistic values for the other two.
1. Deposited volume — measured from your mesh
The estimator reads every triangle in your STL and computes the exact enclosed volume (the signed-tetrahedron method — the same math slicers use for their "model volume" stat). But printers don't print solid parts: they print a shell of walls and skins around sparse infill. So deposited volume is:
deposited = shell (surface area × wall thickness + skins)
+ (solid volume − shell) × infill%
× support factor A typical decorative model at 2 walls / 15% infill deposits only 25–40% of its solid volume. That ratio — shown in the results panel — is why a "500 gram" looking model often prints with 150g of filament.
2. Real-world flow rate — not the spec sheet number
Hotend spec sheets advertise maximum volumetric flow — 32mm³/s for a Bambu X1C. No print averages that. Acceleration, direction changes, travels, small perimeters, and minimum layer times drag the whole-print average to 30–60% of the profile's cap. The estimator's rates are anchored to measured benchmark times: a stock Ender 3 prints a Benchy in ~114 minutes (≈1.4mm³/s effective), a Prusa MK4 in 32–40 minutes, a Bambu X1C standard profile in ~48, a Creality K1 in ~35.
The surprise in the data: on small parts, fast machines barely matter. A Benchy-sized model is cooling- and acceleration-bound, so everything modern converges to ~3.5mm³/s effective. The speed-class gap only opens up on larger prints, where a fast CoreXY sustains ~9mm³/s against a classic bedslinger's ~2.5. That's why this estimator blends its flow rate by deposited volume instead of using one fixed number per machine.
Material shifts these too: PETG profiles are flow-capped ~25% below PLA to control stringing, ABS ~15%, and TPU is capped brutally (~3.6mm³/s even on a Bambu) — near 3× PLA time. Finer layers cost time close to inverse-linearly: a 0.12mm Benchy takes ~1.6× the 0.2mm time, while 0.28mm draft saves ~30%.
3. Overhead — the forgotten minutes
Modern printers spend real time before the first line of plastic: bed heating, chamber soak, auto bed leveling, flow calibration. A Bambu machine's full calibration routine runs about 7 minutes; a classic bedslinger just heats and goes (~3 min). For a 20-minute part this overhead is a third of the job — which is why the estimator counts it once, separately, and why batch plates beat one-at-a-time reprints.
What this is for (and not for)
Use it for: quoting customers before you've sliced anything, sanity-checking a Thingiverse download before committing an evening, comparing two versions of a design, estimating filament needs for a batch, checking whether that 80% infill "solid" setting is really necessary (it isn't).
Don't use it for: billing by the minute, race-the-clock prints, or arguing with your slicer. Once you've sliced, the slicer wins — it simulated every move. The point of this tool is the 10 seconds before that, when you just need a real number fast. When you're ready to turn the estimate into a price, the Pricing Calculator takes grams and hours and returns what to charge.
Why STL files don't slice themselves into a time
An STL is just triangles — no units, no settings, no toolpaths. Two facts follow. First, the same STL prints in 40 minutes on one machine and 2 hours on another, so any estimator must ask what you're printing on. Second, an STL exported from inch-based CAD silently reads 25.4× too small; the tool detects suspiciously tiny models and offers the inch toggle instead of feeding you a 3-minute estimate for a 3-hour part.
If you want to go deeper on the mechanics — what actually consumes the hours, layer by layer — the companion guide "How long will it take?" — estimating 3D print time walks through it with worked examples.
Frequently asked questions
How accurate is this compared to my slicer?
Weight lands within about ±10% (the volume is computed exactly from your mesh — slicers themselves are only 3-5% off reality). Time is honest at ±30%, wider for TPU and heavily-supported prints. The estimator applies whole-print average deposition rates calibrated against measured benchmark times per printer class; a slicer simulates every toolpath move, so it will always be more precise — but for quoting a customer or checking whether a print finishes before bedtime, a 10-second estimate beats a 3-minute slice.
Is my STL file uploaded to a server?
No. The file is parsed entirely in your browser with JavaScript — it never leaves your device. That's also why it's fast: there's no upload wait, even for a 100MB mesh.
How does it estimate filament weight from an STL?
The mesh's exact solid volume is computed with the signed-tetrahedron method over every triangle. Deposited plastic is then modeled as shell (walls + top/bottom skins, from the surface area) plus sparse infill filling the interior. Multiply by the material's density — 1.24 g/cm³ for PLA, 1.27 for PETG, 1.05 for ABS — and you have grams. The same deposited volume divided by the cross-section of 1.75mm filament gives meters.
Why does the printer class matter so much?
Print time is deposition volume divided by real-world flow rate, and flow differs enormously between machines — but only on big parts. A stock Ender 3 Benchy takes ~114 minutes against ~35-48 on a fast CoreXY. On larger prints the gap widens further: a classic bedslinger sustains ~2-2.5mm³/s while an X1C-class machine approaches 9mm³/s. The estimator blends between small-part and large-part rates based on your model's actual deposited volume.
Does layer height change print time linearly?
Not quite. Halving layer height doubles the layer count, but outer walls usually print at a speed limited by quality, not flow — so the machine partly compensates on thicker layers by moving slower. The estimator applies a sub-linear correction calibrated against slicer outputs rather than assuming pure proportionality.
What about supports, brims, and rafts?
Pick a support level and the estimator adds a percentage to deposited volume — roughly 12% for light supports, 25% for moderate, 40% for support-heavy organic shapes. Brims and skirts are negligible (a gram or two) and are ignored. Rafts are not modeled; they add real material, so if you raft everything, mentally round up.
The size shown is way too small for my model — what happened?
Your STL was exported in inches. STL files carry no units, so a 2-inch part reads as 2mm unless told otherwise. Switch "STL units" to inches and the numbers correct themselves — the tool also warns you when a model is suspiciously tiny.
Can I estimate multi-part plates or batch prints?
Yes — set the quantity slider. Material scales linearly; the one-time heatup and calibration overhead is only counted once, which mirrors how batch plates behave on a real printer.