Fourteen years ago I wrote one of the most-read posts on this blog on DMLS surface finishing. It still gets traffic every week in 2026. The methods I covered then — abrasive blast, shot peen, electrochemical polishing, extrude hone, electroplating, optical polish, MMP, and CNC finishing — are all still in use. But the costs have shifted, two meaningful new processes have emerged, and the decision framework for which finish to use for which application has sharpened considerably.
Here is the 2026 update, with current pricing, new methods added, and a straightforward framework for choosing.
The starting point: as-built DMLS surface roughness
DMLS parts come off the build plate with a surface roughness of roughly 150–400 Ra-µinch (3.75–10 Ra-µm) depending on orientation, layer thickness, and alloy. Downskin surfaces (overhangs facing away from the build plate) are rougher than upskin surfaces. This hasn't fundamentally changed since 2010, though the newer 30 µm and 20 µm layer thickness builds give modestly better out-of-the-box surfaces than the 40 µm norms we were running back then.
Target finishes range from "raw" (~350 Ra-µinch / 8.75 Ra-µm) all the way to super mirror finish (1 Ra-µinch / 0.025 Ra-µm). Which one you need drives everything downstream.
Finishing method 1: Abrasive blast (grit & ceramic)
Still the baseline. Blast media is propelled against the part under pressure, producing a uniform matte or satin finish of roughly 150 Ra-µinch (3.75 Ra-µm). This remains the standard "included" finish most DMLS service bureaus provide at no additional charge.
2026 cost: typically bundled with the base part cost. Standalone pricing when needed runs $15–$40 per part depending on size.
When to use: general-purpose finishing. Default choice unless the application specifically demands otherwise.
Finishing method 2: Shot peen
Shot peening uses media impact to create a compressive residual stress layer in the part surface. It's a mechanical property treatment as much as a finishing operation — depending on parameters and alloy, it can increase fatigue life by 100–1000%. Finish improvement is a side benefit.
2026 cost: $25–$75 per part for a standard shot peen cycle, more for controlled-intensity Almen strip verified work.
When to use: fatigue-critical parts — turbine components, suspension components, orthopedic implants. Not primarily a cosmetic finish.
Finishing method 3: Electrochemical polishing
Electrochemistry removes material from the surface through a controlled anodic process (the opposite of electroplating). The result is a bright, deburred, passivated surface at relatively low cost for parts that aren't dimensionally tight.
2026 cost: $35–$150 per part depending on complexity and target finish.
When to use: stainless steel and cobalt chrome parts that need corrosion resistance and visual appeal but don't have tight tolerance requirements. Medical instruments, food-contact components.
Finishing method 4: Abrasive flow machining (extrude hone, AFM)
Abrasive media is extruded through the part — one-way or two-way — to polish internal passages, bends, cavities, and edges. The classic application is interior surfaces that nothing else can reach.
2026 cost: $150–$500 per part depending on size and complexity. Cost premium remains high relative to external finishes.
When to use: internal conformal cooling channels, flow passages in fuel injectors or hydraulic manifolds, heat exchanger internals. Any geometry where you cannot physically reach the critical surface by other means.
Finishing method 5: Electroplating
Electrical current deposits a metal layer onto the DMLS part — commonly nickel, copper, chrome, or gold — to provide corrosion resistance, wear resistance, lubricity, or cosmetic finish. Can also build up thickness on undersized features.
2026 cost: $40–$200+ per part depending on plating material and thickness.
When to use: adding properties the base alloy doesn't have. Jewelry (gold or rhodium plate over cobalt chrome), decorative hardware (chrome plating), electrical contacts (gold plating), and parts needing specific wear or corrosion properties.
Finishing method 6: Optical polish (hand finishing)
Hand polishing with progressively finer abrasives to achieve mirror finishes down to 1 Ra-µinch (0.025 Ra-µm). The best way to get a brilliant finish on complex geometry, but inconsistent part-to-part and material-intensive — 0.003"–0.010" of surface material is removed depending on geometry. If pursued, design must offset critical features with extra material to maintain tolerance.
2026 cost: $150–$800+ per part depending on complexity and desired final finish.
When to use: low-volume, cosmetic applications where appearance matters more than cost per part. Prototype jewelry, display components, high-end consumer products.
Finishing method 7: Micro Machining Process (MMP)
MMP is a mechanical-physical-chemical surface treatment in a batch tank, selectively applying different roughness profiles to targeted surfaces. The 2010 numbers were $500–$1,000 for sample testing, $3,000+ for batch runs — expensive to qualify, cost-competitive at volume.
2026 cost: sample testing now $750–$1,500. Production batch costs scale with part volume, roughly $8–$35 per part at 100+ unit volumes.
When to use: production runs of complex parts requiring selective tolerance-grade finishing, particularly aerospace components with internal passages and critical external surfaces simultaneously.
Finishing method 8: CNC finishing / machining
Precision contour milling to achieve tight tolerances on specific features. 3-axis, 4-axis, and 5-axis setups can hit tolerances as tight as 0.00004" (1 µm) on critical features. Requires pre-build planning — you must add machining stock to any surface you plan to post-machine.
2026 cost: highly variable. Hourly rates run $75–$180 per hour depending on machine class and location, with first-article setup adding $200–$600.
When to use: any feature requiring sub-thousandth tolerance. Mating surfaces, sealing faces, bearing bores. Nearly all production DMLS parts have at least one feature requiring this.
New in 2026, method 9: Hirtisation (electrochemical pulse-reverse)
Not available when I wrote the original post. Hirtisation (originated by Hirtenberger) is an electrochemical pulse-reverse process specifically engineered for AM parts, especially internal and lattice geometries that no other method can reach. It removes partially-melted powder, smooths internal channels, and produces finishes competitive with electrochemical polishing — but on surfaces that were previously impossible to access.
2026 cost: $100–$400 per part depending on size and complexity.
When to use: complex lattice structures, internal passages too narrow for AFM, medical implants where every internal surface matters. Particularly effective on Ti-6Al-4V, Inconel 718, and stainless steels.
New in 2026, method 10: Laser polishing
Also new since the original post. A controlled laser re-melts the surface microlayer to achieve finishes in the 10–30 Ra-µinch range without material removal. Still relatively specialized, but increasingly offered by production bureaus for tool steel and medical implant applications.
2026 cost: $60–$250 per part.
When to use: cases where you need an improved finish but can't afford dimensional change — the laser polish essentially re-flows the surface rather than removing material.
A simple decision framework for 2026
- Need a baseline usable surface? Abrasive blast. Standard, bundled, done.
- Need fatigue life? Shot peen.
- Need corrosion resistance or visual appeal, no tight tolerance? Electrochemical polish.
- Internal passages that nothing else can reach? Hirtisation first, AFM second.
- Mirror finish for display or prototype? Optical polish.
- Volume production with selective tolerance finishing? MMP.
- Any critical dimensional feature? CNC post-machine, with the stock planned into the build.
- Adding properties the base alloy lacks? Electroplating.
- Surface improvement without material loss? Laser polishing.
How this fits into the broader 2026 DMLS picture
Surface finishing is one of several areas where metal AM has quietly matured since the early 2010s. If you want the broader market view — what's changed in materials, build volumes, cost curves, quality systems, and the service bureau model — I wrote a companion post this week here.
The short version: finishing isn't the bottleneck it used to be. The bottleneck now is honest design for additive, realistic post-processing planning, and sourcing from a supplier who can execute the full stack without 20 handoffs. If you want help thinking through any of that for a current program, that is what I do.
Posts
