3D Printing: FDM, SLA, SLS, MJF & Metal Technologies

3D printing, or additive manufacturing, builds a physical part layer by layer from a digital model, adding material only where it is needed rather than cutting it away from a solid block. For engineering and production work the relevant families are FDM (extruded filament), SLA (cured resin), SLS and MJF (fused nylon powder), and metal powder-bed fusion such as DMLS and SLM. Each family occupies a different position on the trade-off between accuracy, surface finish, strength, material range, and cost.

This guide explains how each technology works, what it does well, and, just as importantly, when it is the wrong choice. The goal is to help you pick a process from the part function, rather than force a part onto a machine that does not suit it. The detailed technology comparison lives in the FDM vs SLA vs SLS page, and cost drivers are covered separately in the 3D printing quote page.

How 3D printing works

Every 3D printing process follows the same logical flow, even though the physics differ. A solid CAD model is exported, most often as an STL mesh or a STEP file. Slicing software cuts that model into thin horizontal layers and generates the toolpath, speed, and material settings the machine will follow. The machine then builds the part one layer at a time, bonding each layer to the one below.

The layer height sets a hard limit on surface quality and Z resolution. A finer layer produces a smoother vertical surface and finer detail but takes longer to print. A coarser layer builds faster but leaves visible stair-stepping on slopes and curves. Most engineering processes run layer heights between 25 and 300 micrometers, depending on the technology.

Orientation drives everything

Before the part is printed, the build orientation is chosen, and that single decision affects accuracy, strength, surface finish, support volume, and cost. A face printed flat against the bed comes out smooth and accurate, while a face built up at a shallow angle shows stair-stepping. A tall, thin wall is faster to print lying down but may need supports; the same wall printed standing up avoids supports but is weaker across the layers.

Where supports are needed

Processes that deposit or cure material in open air, namely FDM and SLA, need support structures under overhangs steeper than about 45 degrees and under any isolated feature with nothing below it. Those supports are broken or dissolved away afterward, which adds labor and leaves marks. SLS and MJF avoid this entirely because the unfused powder bed surrounds and supports the part as it builds, which is why they excel at complex, interlocking geometry.

The digital-to-physical scale problem

The model is unitless until units are assigned. An STL file stores only numbers and does not know whether they mean millimeters or inches. If the units are wrong, the part comes out at the wrong scale, and because one inch equals 25.4 millimeters, a unit mismatch produces a part that is 25.4 times too large or too small. State the units in the file or the filename every time.

The main 3D printing technologies

The five families below cover the great majority of engineering 3D printing. Each is treated as a technology, not as a service claim, and each carries a clear use case and a clear failure mode.

FDM: fused filament fabrication

FDM, also called fused filament fabrication or FFF, pushes a thermoplastic filament through a heated nozzle and deposits it track by track, layer by layer. It is the cheapest and most widely available process, and it prints in tough, familiar materials such as PLA, PETG, ABS, and nylon.

Tolerances vary with the material. ABS typically holds about plus or minus 0.1 to 0.3mm, polycarbonate and similar engineering filaments about 0.2 to 0.5mm, nylon about 0.3 to 0.5mm, and flexible TPU about 0.5 to 1.0mm. As-built surface finish runs roughly Ra 4 to 12 micrometers, which shows visible layer lines.

The defining weakness of FDM is anisotropy. A printed part is typically 20 to 30 percent weaker across the layer direction than within the layer plane, because the bonds between layers are weaker than the material within a track. This means a part designed to carry load should be oriented so the stress runs along the layers, not across them.

Use FDM for low-cost, tough prototypes, jigs and fixtures, brackets, and housings where cosmetic finish is secondary. Do not use it when you need fine detail (SLA is better), when you need isotropic strength, or when the part must be smooth without post-processing.

SLA: cured resin printing

SLA, stereolithography, uses a light source, either a laser or a projected image, to cure a liquid photopolymer resin one layer at a time. It produces the finest detail and the smoothest surface of any common 3D printing process.

On a well-tuned resin printer, dimensional tolerance runs about plus or minus 0.05 to 0.15mm depending on part size, with small features held to about 0.02 to 0.06mm. Minimum walls reach about 0.2mm and minimum holes about 0.5mm. Surface finish is about Ra 0.5 to 2 micrometers, smooth enough to look injection-molded after a brief post-cure.

The trade-offs are material and durability. Resins are brittle compared with engineering thermoplastics, and they age under UV and heat. A standard resin has a heat deflection temperature around 60 to 80 degrees Celsius, while a high-temperature resin can reach 200 to 280 degrees Celsius, though at higher cost and lower toughness. Parts also shrink slightly, about 0.1 to 0.3 percent, during post-curing.

Use SLA for appearance models, tight-tolerance features, patterns for casting, and small complex parts where surface quality matters. Do not use it for load-bearing or impact-loaded parts unless you have specified a tough or engineering resin, and do not assume a printed resin part is biocompatible or food-safe without material qualification.

SLS: selective laser sintering

SLS uses a laser to sinter, or lightly fuse, a bed of nylon powder, one thin layer at a time. Because the unfused powder surrounds the part as it builds, SLS needs no support structures, which makes it strong for complex geometry, living hinges, and thin walls.

Dimensional tolerance runs about plus or minus 0.3mm for features under 50mm and about 0.5mm for larger features. Minimum walls sit around 0.8mm for non-structural geometry and about 1.2mm for load-bearing walls. Overhangs are self-supporting above roughly 30 degrees, and bridges up to about 5mm can span without support.

The standard SLS material is nylon PA12, often in a glass-filled grade for added stiffness and heat resistance. Parts come out with a slightly grainy, matte surface that can be dyed or smoothed. SLS parts are notably more isotropic than FDM parts, which is why they are favored for functional prototypes and end-use brackets.

Use SLS for functional nylon parts, complex assemblies, snap-fits, and any geometry where support removal would be difficult. Do not use it for the finest cosmetic detail (SLA wins) or for transparent parts, and expect a higher unit cost than FDM for simple geometry.

MJF: multi jet fusion

MJF, Multi Jet Fusion, is the HP process that fuses nylon powder across the full build area using a fusing agent and infrared energy rather than a moving laser. It produces parts mechanically similar to SLS, in the same nylon PA12 family, often with a slightly faster build and a characteristic gray as-built color that is commonly dyed black.

Tolerance runs about plus or minus 0.2 to 0.3mm for features up to 100mm and about 0.3 percent for larger dimensions. Minimum walls are about 0.3mm in the XY plane and 0.5mm in Z, with minimum features near 0.5mm, minimum holes near 1.0mm, and minimum pins near 1.5mm. Layer thickness is about 80 micrometers, and as-built surface finish runs about Ra 4 to 9 micrometers.

For assembled or moving parts, MJF calls for clearances of about 0.3 to 0.6mm per wall so fused surfaces do not bind. Like SLS, MJF needs no supports, so it suits complex, interlocking, and nested geometry.

Use MJF when you want SLS-like nylon performance with consistent mechanical properties and a fast turnaround on nested builds. Do not expect a cosmetic surface without post-processing, and note that the gray base color limits some appearance uses unless dyed.

Metal additive: DMLS and SLM

Metal additive manufacturing, most often DMLS (direct metal laser sintering) or SLM (selective laser melting), uses a laser to melt a bed of metal powder, layer by layer, into a dense part. Common materials are aluminum AlSi10Mg, stainless 316L, and titanium Ti-6Al-4V.

Layer heights are fine, about 20 to 80 micrometers, and tolerance runs about 0.3 percent of the part dimension with a floor near 0.3mm. As-built surface finish is comparatively rough, around Ra 5 to 10 micrometers after shot peening, so critical faces almost always need machining. Parts also need stress-relief heat treatment and support removal.

Metal additive is the right answer for complex geometry that cannot be machined, such as internal cooling channels, lightweight lattice structures, and consolidated assemblies, and for low volumes where tooling does not pay off. It is the wrong answer for simple parts that machine easily, for high volumes where casting or machining is cheaper per part, or for any feature that needs a fine cosmetic or mating surface without post-machining. It is the most expensive of the five families, driven by powder cost, slow build rate, and extensive post-processing.

Choosing the right technology

The decision comes down to four questions: how accurate must it be, how strong, how many, and how cosmetic. A part that must hold a tight fit needs SLA or post-machined metal; a part that must take load needs SLS, MJF, or a properly oriented FDM print; a part needed in hundreds needs a process rethink toward molding; and a part that must look finished needs SLA or added post-processing.

Match the process to the function

For example, a drone airframe prototype that must be light, tough, and moderately accurate is a strong fit for SLS or MJF nylon, because it needs no supports and is reasonably isotropic. A jewelry master pattern that must capture fine detail and finish to a mirror surface is a fit for SLA, because no other common process matches its surface. A shop-floor bracket that only needs to hold a sensor and does not need to look good is a fit for FDM, because it is cheap and fast.

A practical comparison framework

Read the families against five axes. On accuracy, SLA leads, then MJF, then SLS, then FDM, then metal additive before machining. On strength, metal leads, then SLS and MJF nylon, then FDM with a strong orientation penalty. On cost per part for simple geometry, FDM is lowest, then SLA for small parts, then SLS and MJF, then metal. On material range, FDM and metal offer the widest set, SLA is resin-bound, and SLS and MJF are largely nylon. On surface finish, SLA leads, then MJF and SLS with dyeing and smoothing, then FDM and metal additive with visible layer or bead texture.

When to step away from additive

Additive is not always the answer. If a part needs a sliding fit or a sealing surface, machine it, or print it and machine the critical face. If you need thousands of identical simple parts, injection molding or casting wins on unit cost. If the part is a flat plate or a bent sheet, laser cutting or sheet metal fabrication is faster and cheaper. Additive earns its place on complexity, on low volume, and on geometry that no subtractive process can reach.

3D printing materials

Material choice is tied to the process, because not every material prints on every machine. The two broad groups are polymers and metals.

Polymer materials

The common FDM filaments cover a useful spread of properties. PLA is easy to print and stiff but has a low heat deflection temperature around 55 degrees Celsius, so it is best for models, not hot environments. PETG sits in the middle with a heat deflection temperature near 70 degrees Celsius and better toughness than PLA. ABS reaches about 95 degrees Celsius and is tougher, but it needs a heated enclosure to print without warping. TPU is a flexible material sold across a Shore A 60 to 95 hardness range, good for gaskets and overmolds, but it needs dry storage because it absorbs moisture. Nylon PA12, the SLS and MJF workhorse, absorbs only about 1 percent moisture, far less than PA6 at about 9 percent, which keeps it dimensionally stable.

For higher loads, glass- or carbon-fiber-filled filaments add stiffness, but the abrasive filler wears standard nozzles and calls for a hardened steel nozzle. The very high-temperature engineering polymers, PEEK and PEI (Ultem), with heat deflection temperatures of 210 to 260 degrees Celsius, are printed only on specialized high-temp machines and are treated as an advanced topic.

Metal materials

The metal additive palette centers on three alloys. Aluminum AlSi10Mg is the general-purpose choice, light and weldable. Stainless 316L adds corrosion resistance for harsh environments. Titanium Ti-6Al-4V, with a tensile strength in the 895 to 1105 MPa range and a density about 60 percent of steel, is the aerospace and medical choice, though it machines slowly and conducts heat poorly, which makes machining of the printed blank demanding.

Accuracy and tolerances compared

No 3D printing process matches a machined tolerance out of the build chamber, and the differences between processes are large enough to drive the whole material and process decision. SLA holds the tightest tolerance of the polymer processes at about plus or minus 0.05 to 0.15mm, because the cure is precise and the resin shrinks predictably. MJF follows at about plus or minus 0.2 to 0.3mm up to 100mm, then SLS at about 0.3 to 0.5mm, then FDM, which ranges from 0.1mm on a well-tuned ABS print up to 1.0mm on flexible TPU.

Metal additive sits near 0.3 percent of the part dimension, which on a 100mm part is about 0.3mm, comparable to SLS but on a much harder material. For any mating or sealing surface, plan to machine the printed blank, because the as-built surface and tolerance will not meet a precision fit. The comparison page and the tolerance reference table carry the detailed numbers by material and feature size.

Post-processing and finishing

A printed part is rarely finished when it leaves the build chamber. Post-processing is where most of the time and cost sits, and planning for it up front is what separates a part that assembles cleanly from one that needs rework.

Polymer post-processing

FDM parts get their supports removed and their layer lines sanded or filled if a smooth surface is needed; vapor smoothing can gloss ABS, and a primer and paint finish is common on cosmetic housings. SLA parts are washed in solvent, have supports removed, and are post-cured under UV to reach full strength, with the shrinkage of 0.1 to 0.3 percent already accounted for in the design. SLS and MJF nylon parts get their trapped powder blasted out, then may be dyed, vapor-smoothed, or tumbled for a more uniform surface.

Metal post-processing

Metal additive parts almost always need stress-relief heat treatment to relieve residual stresses from the build, support and powder removal, and machining of any mating or sealing face. The as-built surface, around Ra 5 to 10 micrometers after shot peening, is too rough for a precision fit. Secondary operations such as bead blasting, anodizing (on aluminum), or passivation (on stainless) bring the finish up to cosmetic or functional standards.

Plan finishing into the design

Because post-processing adds cost and lead time, design for it. Leave machining stock on critical faces, add clearance for support removal in deep pockets, and avoid features that trap powder or resin. A part that needs no secondary work is cheaper and faster than one that needs extensive finishing, so treat the as-printed condition as the starting point, not the end state.

Design for 3D printing

Good design for additive means working with the layer direction rather than against it. Walls should be a multiple of the nozzle or beam width so they print solid, and they should be oriented to carry load along the layers where possible. Holes print best vertical to the build, and small holes can be drilled to size after printing for a true fit.

Walls, overhangs, and bridges

Keep walls at or above the process minimum: about 0.8mm for SLS non-structural geometry, about 0.3mm in XY for MJF, about 0.2mm for SLA. Overhangs beyond about 45 degrees from vertical need supports in FDM and SLA, so orient the part or add a chamfer to avoid them. Bridges up to roughly 5mm can span in SLS without support; longer spans need a support strategy or a redesign.

Minimize supports and post-processing

Supports cost material, cost time, and leave marks, so the cheapest part is often the one with the fewest supports. Split a part along a flat plane, hollow thick sections, and add drainage holes for trapped powder or resin. For SLS and MJF, remember that trapped unfused powder must escape, or it stays sealed inside the part.

Watch the anisotropy

Because FDM is 20 to 30 percent weaker across the layers, orient load-bearing FDM parts so the load runs in-plane. If the load must run across the layers, either thicken the section, switch to SLS or MJF, or redesign the joint. Anisotropy is a design input, not a defect to ignore.

File format and units

STL is the universal 3D printing mesh format and is accepted by almost every slicer. It carries shape but no tolerance or units metadata, which is fine for printing but a problem if the file will also be machined. STEP is the preferred format when tolerance data matters or when the same file feeds both additive and subtractive work.

The single most important rule is to state the units. A file without explicit units is read against the supplier default, and a millimeter-versus-inch mistake produces a part at 25.4 times the intended scale. Put the units in the file properties and in the filename, and confirm them in the order, because this one mistake is the most common and the most expensive file-format error in custom manufacturing.

Applications: where each technology wins

The clearest way to choose a technology is to look at the job it does best. Each family has a characteristic set of applications where it outperforms the others.

FDM applications

FDM wins on low-cost, tough, non-cosmetic parts. Typical uses are early-stage prototypes, ergonomic mockups, jigs and fixtures for the shop floor, drill guides, and large housings where surface finish is secondary. It is the default first print for a new design because it is cheap and fast, and because the materials are familiar and easy to source.

SLA applications

SLA wins on detail, surface quality, and tolerance. Typical uses are appearance models for customer review, jewelry and dental patterns, fine-feature prototypes, master patterns for silicone molding and casting, and optical or transparent parts. Any application where the part must look finished, or where a feature is too fine for FDM, points to SLA.

SLS and MJF applications

SLS and MJF win on functional strength without supports. Typical uses are snap-fit housings, living hinges, complex assemblies with internal channels, brackets and mounts that carry load, and small-batch end-use parts. Because the powder bed supports the build, these processes also handle interlocking and nested geometry that FDM and SLA cannot print in one piece.

Metal additive applications

Metal additive wins on geometry that machining cannot reach. Typical uses are internal cooling channels in tooling, lightweight lattice structures for aerospace, consolidated assemblies that replace a weldment of many parts, and low-volume titanium or Inconel components for demanding environments. It is the choice when complexity or material performance matters more than unit cost, and when the alternative is a multi-part assembly.

When 3D printing is the right choice

3D printing earns its place when complexity, customization, or low volume makes tooling or machining uneconomic. A part with internal channels, a one-off bracket with a custom fit, or a short run of housings before tooling is committed are all strong additive candidates. The technology also shortens the loop between a design change and a physical part, which is why it dominates prototyping.

It loses to other processes at the extremes. For a tight sliding or sealing fit, CNC machining holds tolerances additive cannot. For hundreds or thousands of identical simple parts, injection molding drops the unit cost far below any print. For flat or bent sheet parts, laser cutting and sheet metal fabrication are faster and cheaper. And for the largest, simplest geometry, additive build volume and cost work against it. The right call is to match the process to the part function and volume, using the decision framework above, and to reach for additive where its strengths in complexity and low volume actually apply.