Commercial (Plastic) 3D Printing Options

When you think about 3D printing, you probably think of the Fused Deposition Modeling technique used by RepRap and similar desktop machines. In reality, there are quite a few different options from job shops like Shapeways, Xometry, and Fictiv. They offer a surprisingly wide range of properties and DFM concerns, and sometimes I have trouble keeping them all straight. Also, the DFM particulars are usually pretty intuitive if you just know how the process works, which can be tricky to remember for similarly named name-brand options like PolyJet Fusion and Multi Jet Fusion.

Here are some brief notes to help you avoid messing up your next rush order.


image of FDM printed part

FDM is also sometimes called Fused Filament Fabrication since Fused Deposition Modeling is still a Stratasys trademark even though the foundational patents have all expired. Either FDM or FFF, this is the standard extruded thermoplastic technique you’re probably used to thinking about. A hot extruder melts plastic forced in as a filament, and deposits it onto a part layer by layer. All kinds of machines exist, from $150 entry level desktop machines to $10k professional machines with interesting features like continuous reinforcement fiber inclusion (Markforged).


  • Mature tech
  • Tons of thermoplastic materials like PLA, PETG, ABS (cheap, relatively easy), polycarbonate, nylon (harder to print, but excellent strength and temperature tolerance) and even some exotic ultra high temperature materials like PEEK, along with niche materials like HIPS (dissolvable) and wax (melts out for investment casting)
  • Cheap machines, or reliable machines, or exotic machine options
  • Relatively safe and uncomplicated material handling: rolls of inert plastic and no additional consumables other than household cleaners. Some materials require good ventilation, but the only immediate health hazard is burns.


  • Thermoplastic materials ONLY. Notably different from injection molding, where thermoset plastics can be used, which don’t easily re-melt and are therefore much stronger in higher service temperatures. Any thermoplastic will necessarily have a temperature at which the part starts to soften and weaken, and worse, the printing temperature must be well in excess of this, making high temperature materials especially tough.
  • Surface finish. FDM parts have limited peak surface finish quality, and this quality is also dependent on orientation of the specific feature. One way to improve surface finish is a smaller layer height, which incurs a linear print time time penalty.


  • Print time: proportional primarily to volume of material printed.
  • Can trade detail for print time. For instance, better X-Y resolution for small features by using a smaller nozzle that takes longer to fill in each layer, or better Z resolution by using a smaller layer height (and more layers take longer).


image of SLA 3d printed part

“Stereolithography Apparatus” rather poorly describes both the process and the machine, and indeed, there are a range of processes even in the specific scope of SLA printing. All of them share in common building an object by selective UV curing of a liquid photopolymer resin.

Common advantages

  • Surface finish and fine detail are much much better than FDM, as all SLA processes have better resolution and smaller material error.
  • Cured resin mechanical properties tend to be less sensitive to temperature than thermoplastics.

Common disadvantages

  • Photopolymer resin is quite toxic. The fumes bother many people and skin contact can result variably in anywhere from no reaction, to mild allergic response, to severe chemical burn.
  • Post-processing of printed objects requires chemical solvents and cleansers, multiple steps, and additional equipment like ultrasonic cleaners and UV post-curing chambers. The process can be very messy without exceptional care.
  • UV cured resins tend not to be as strong as thermoplastics, though some resins can be as strong as some thermoplastics.
  • SLA processes generally require adding supports to the object rather than printing flat against the build plate as can be done with FDM for many models, for reasons of contact area and build reliability. This can be a pain in post-processing, and can sometimes negate some of the advantages in terms of surface finish quality.

Laser SLA

Until recently the most common kind of SLA printer, a laser SLA machine uses a small UV laser dot, scanned in X and Y across a clear window at the bottom of a tank of resin. This cures the layer up against the window, and a build plate to which the growing model is attached pulls the new layer up and away, letting more resin into the gap for the next layer. The big names here are the Formlabs Form 1/2/3 and the Peopoly Moai.


  • The galvanometers used to scan the laser dot have tremendous precision, so the feature resolution and dimensional accuracy in X and Y is very, very high. This can only be approximately matched by the other options below.
  • The build area has practical physical limits based on the angle of incidence of the laser dot, but relative to other SLA options, larger build areas can be achieved (150x150mm)


  • Because the laser dot has discrete size, there is a minimum feature size similar to FDM prints. It’s mostly not a practical concern, though.
  • Because a small dot has to be scanned around to cure the layer, and because curing requires a discrete dwell time of the dot on a particular volume, print time is proportional to the cured volume of the final model, similar to FDM, which is generally longer than most other SLA options.


Masked SLA is a relatively recent entrant that takes advantage of cheap, ultra high resolution consumer LCD screens, mostly from cell phones. A popular 2k phablet panel of 120mm x 68mm size has 47µm pixels, which is actually quite a bit smaller than the smallest dot size of the leading laser SLA printer from Formlabs (70µm, if memory serves). A bright UV illuminator shines through this LCD onto the build plate, casting a shadow onto it. Where the LCD is “white,” resin cures into the layer.


  • Cost. All the components of a MSLA printer are very cheap due to high volume production in the market, so machines that leave out very advanced features can be cheaper than even the cheapest laser machines. Think $250-500 for MSLA, vs $1300 for the cheapest laser machine.
  • Build time. Because an entire layer is exposed at once, print time is proportional to model height only.


  • Build volume. The largest acceptably high-resolution panels available are cell phone sized. For comparison, the current MacBook Pro retina displays 115µm pixels (1/4 the resolution), and even the best competing laptop displays only hit about 90µm (4k at 15″). I wouldn’t expect to see comparable resolution, larger-size displays, as such small pixels are really only relevant in a device held inches from the face. That said, at least one or two MSLA manufacturers are working on printers with larger build areas and only slightly lower resolution, based on tablet display parts.
  • Build time. Because the build time is height-dependent, you might have a tiny model that must be printed vertically and therefore takes a long time, where a laser machine might actually take less.
  • Aliasing. Because MSLA printers have pixels, and because model features don’t necessarily align on pixel boundaries, lines where the resolution aliases into the model can appear on the surface finish. These are basically just like layer lines, except in the X and Y axes as well as the Z. Antialiasing CAN be employed to mitigate this effect, by effectively increasing X-Y resolution with partial curing. More info in this video from Autodesk.


  • Resolution. It’s pretty difficult to compare the resolution and achievable dimensional accuracy between laser and MSLA printers without diving down deep, deep rabbit holes, but especially with antialiasing, MSLA offers practical equivalency in most cases.


Image of a DLP micromirror array chip
DLP micromirror array chip

Digital Light Processing is a very specific technology that uses a micro-mirror array to create a projected image. A light source is focused on the mirror array, which reflects that source at each mirror site either towards a projection screen or off to the side instead. When people talk about DLP printers, they generally mean any printer using a projector, even if it’s not actually a DLP projector. An example DLP machine is the B9Creator.


  • Flexibility. Because you can move the projection source closer to or farther from the build plate (screen), you can trade print resolution for build area
  • Build time. Similar to MSLA – because a whole layer is exposed at once, some models will print much faster than laser.


  • Bulk. The projection machinery and optics tend not to be very compact, where a MSLA machine can be extremely small.
  • Cost. DLP and other projectors are generally pretty expensive, so printers built around them are more expensive too.
  • Calibration complexity. Compared to a MSLA printer where the pixel you light is the voxel you print, there’s a lot more room for calibration error on a projector-based system.

Honestly, I don’t see many DLP printers around, in either professional OR hobbyist settings. I’m not super sure how popular they ever were. Mostly it’s worth mentioning because sometimes people refer to MSLA as DLP, since they use basically the same printing process if not the same tech to get there.

Now it’s time to bring in some of the newer, high priced commercial contenders


Image of an SLS build in progress
SLS build in progress,

Selective Laser Sintering is one of the oldest forms of plastic additive manufacturing. In an SLS process, a thin layer of fine grained plastic powder is swept across the build area. A high energy laser scans around this layer, fusing the plastic grins where it hits. The build volume is lowered, and a new layer is swept across for the process to repeat.


  • Produces parts of similar or greater strength to FDM
  • Doesn’t require support structures, saving time and providing for a uniform surface finish.
  • Because no supports are required, the build volume can be packed both in X and Y, but also in Z with multiple potentially different parts.
  • Because the temperature of the build area needs to be precisely temperature controlled and uniform, warping is less of an issue than it can be for some materials on other thermal processes (for instance, FDM ABS).


  • X-Y resolution is limited by laser spot size and powder granularity
  • Build time is dictated by fused volume of plastic, and because powder will be trapped in internal voids, either drain holes must be built into the part or it must be printed solid, incurring a time penalty.
  • Machines are expensive, starting at $10k, because the lasers and machinery to direct them are expensive.
  • Larger industrial systems require special air handling to control the atmosphere around the powder bed at elevated temperature for long periods, contributing to bulk, complexity, and cost, though smaller systems might not have this issue.

Stratasys PolyJet Fusion

This is one of Stratasys’ latest offerings. It’s quite interesting in that, chemically, it’s a UV cured resin process like SLA, but mechanically it’s quite different. The machine basically uses inkjet technology to precisely spray a layer of photopolymer onto the build plate as it scans over, and cures this layer with UV light as soon as it lands. The first very interesting thing this allows is vanishingly thin layer heights, as low as 16µm (SLA printers generally operate around 50µm – much thinner is impractical both for reasons of Z axis precision and overall print time). The second is full color – by using pigment prime colors of resin, just like an inkjet, these can be mixed live on the object being built to create full color in all 3 dimensions, as well as clear. Not only can this be done on the surface of the object, but at any point inside it, so clear prints can show colored internal features, or prints can be made like everlasting gobstoppers where cutting into them reveals colored insides. Here’s a video about how it works.


  • Print time. Because the scan head moves fast and deposits material in a line, part geometry doesn’t necessarily impact layer time too much. The part can be laid down with very fine resolution, but a very broad brush. Because layers are thin, curing time per layer is fast.
  • Full color prints. No other tech can achieve this.
  • Resolution. Z resolution is extremely high at down to 16µm layer height, and X-Y resolution is comparable to an Inkjet printer (good enough for pictures, probably good enough for your objects)


  • Cost. The machines are proprietary and expensive. You’re probably not buying one yourself, and you’re paying your job shop for the machine time.
  • Part strength. Because these are made of photopolymer resin, they’re going to be similarly brittle. Stratasys DOES actually offer a variety of material options. But even if you trade color for strength, you could still probably do better with FDM or HP’s competitor, MJF, below.


  • Surface finish. The top surface is glass-smooth and shiny with a little bit of texturing, where the sides in the Z axis are almost fuzzy matte. I think this is because the layer self-levels, but exactly how far the top layer goes towards or over the edge of the layer below is variable. The end result is not totally uniform, just like FDM, which could be a turn off.

HP Multi Jet Fusion

It’s easy to confuse the two names, and actually the processes are somewhat similar too, though the final products are not. Multi Jet Fusion mixes inkjet tech similar to PolyJet Fusion, and a powdered bed process like SLS. A thin layer of powder is spread over the build volume. Instead of a laser selectively fusing the layer where the part is to be constructed, an inkjet head sweeps across and sprays either a fusion activating agent or a detailing agent, where the part is to be built, or just beyond the edge. Heat is then applied with infrared radiation to fuse the material where the activating agent was deposited. The detailing agent serves to make sure edges are defined crisply. With a layer fused, the bed lowers and a new thin layer is spread across the build area. Again, YouTube:


  • Finished part strength is EXCELLENT. On par with or perhaps even better than FDM, and I believe with better thermal properties as well.
  • Speed is pretty good
  • Good production volume, since there’s no time penalty to pack a build volume full of objects.
  • Better DFM parameters w.r.t. part geometry and surface finish, since supports aren’t (generally? ever?) required, since the powder material takes care of that itself.
  • Can deposit full color additives during printing, for results similar to PolyJet Fusion.


  • The only material options, for now, are nylon (PA 11 and 12), and apparently one kind of TPU if you want flexible parts.
  • Similar to PolyJet Fusion, Multi Jet is a new process with big heavy expensive proprietary machines. You’ll pay big for the machine time.


FDM is cheap and accessible, and produces strong, kinda ugly parts in up to tens of hours. SLA/MSLA is cheap and accessible, and produces weak, pretty parts with high resolution, detail, and accuracy in hours. PolyJet Fusion is expensive, and produces very accurate, pretty, full-color, weak parts in 1-day turn from a service provider. Multi Jet Fusion is expensive and produces accurate, strong, full color parts with attractive surface finish, with a potential volume discount from a service provider. SLS produces strong, medium-detail parts at high cost and relatively slow, probably from a service provider but maybe from a small machine in your own shop.

Change Log:
8/26/19 – Added some notes on SLS and corrections on MJF color support, DLP details, formatting clarifications.

Be First to Comment

Leave a Reply

Your email address will not be published. Required fields are marked *