Engraved QR codes: what actually scans on metal, wood, glass, and stone

A client hands an agency a trophy, a stainless steel building sign, or a run of wine bottles and asks for a QR code lasered straight into the surface. It looks like the easy version of the job: no ink to fade, no laminate to peel, nothing for weather to attack. Then the proof comes back from the engraver and half the test phones won't read it.

The advice that works for printed QR codes, dark ink on a light background, a laminate for UV protection, a standard error correction level, was written for a different physical process than engraving. A laser doesn't lay pigment on top of a surface. It removes, melts, or discolours the surface itself, and the code that results is read by shadow and reflection rather than by colour. That difference changes almost every spec an agency would normally reach for by habit.

Why an engraved code is a different scanning problem

A printed QR code gets its contrast from light absorption: dark ink absorbs light, the paper or vinyl around it reflects light, and a phone's camera sees a clean difference between the two. An engraved or etched code has no ink. The dark and light regions are the same material, so the contrast has to come from somewhere else entirely, usually a change in surface relief (a recessed or raised cell scatters light differently to the flat field around it) or a change in reflectivity from heat discolouration. Barcode-verification vendors who build hardware for exactly this problem describe direct part marks as inherently lower-contrast than printed labels, because the mark and the background are the same substrate and the difference is created only by microrelief or a shift in how the surface reflects light.

That single fact explains most of what goes wrong with a first attempt. A design that would scan instantly in ink can be genuinely unreadable once the same pattern exists only as a change in surface texture, especially under the kind of flat, even lighting most people photograph things in.

Glare, not darkness, is usually the enemy

Ink-on-paper problems are almost always about darkness: not enough contrast, a logo eating too much of the pattern, a scan taken in low light. Engraved codes fail for a less intuitive reason: reflection angle.

Flat, undisturbed metal, glass, or a polished stone surface reflects light in a mirror-like way (specular reflection): light comes in at one angle and leaves at the matching angle, with almost nothing scattered elsewhere. A rough or engraved surface scatters light in many directions instead (diffuse reflection). The problem is that an engraved cell can throw a bright specular reflection straight back at a camera lens depending on the angle it's held, which can look brighter than the untouched surface around it. That's the opposite of what a decoder expects: the mark, not the background, becomes the "light" pixel, and the code refuses to read even though it's perfectly legible to a human eye tilted a few degrees differently.

This is a well-documented problem in industrial machine vision, not a fringe concern: barcode readers built for direct part marking use angled, off-axis lighting specifically to control it, rather than assuming a phone flash or a shop light held straight-on will work. A matte or bead-blasted finish, where practical, keeps the field of scattered light more consistent and gives a phone camera a fighting chance regardless of the angle someone holds it at.

There's a second, smaller effect worth knowing about: laser marking is a thermal process, and every mark has a heat-affected zone, a thin ring of material around the cut that absorbed enough heat to change colour or texture without being fully removed. A wider heat-affected zone means softer, blurrier module edges, which works against a decoder the same way a badly inked, dot-gain-heavy print run does: the sharp square-to-square transitions the algorithm is looking for get smeared.

Matching the laser to the material

Engraving equipment isn't one machine with one setting; the wavelength of light a laser produces determines what it can actually mark, and using the wrong type doesn't just give a worse result, it can fail to mark the surface at all.

• Fiber lasers (near-infrared, around 1064 nanometres) are absorbed well by metals, ferrous and non-ferrous alike, which makes fiber the standard choice for QR codes on stainless steel, aluminium, brass, and titanium.
• CO2 lasers (far-infrared, around 10.6 micrometres, roughly ten times fiber's wavelength) are absorbed well by organic materials: wood, leather, paper, and many plastics. Metal, by contrast, mostly reflects that wavelength rather than absorbing it, so a CO2 laser is a poor tool for marking metal.
• Glass is the material the simple "metal versus organic" rule doesn't cover well. Fiber lasers largely can't mark plain, uncoated glass, because glass doesn't absorb 1064nm light effectively. CO2 can produce a basic surface frosting, fine for a logo but limited for the fine detail a QR code needs. Specialist engravers reach for UV lasers (around 355 nanometres) for high-quality glass work, including engraving just below the surface, because the shorter wavelength breaks chemical bonds directly rather than burning the surface with heat, which lowers the risk of the microcracking that ruins a glass engraving.

The failure mode for choosing wrong is concrete, not theoretical. A budget diode laser, common in hobbyist machines and a different technology again from fiber or CO2, will pass straight through clear acrylic without marking it at all rather than producing a faint result. And on some plastics, CO2 has been reported to leave only a colourless, low-contrast mark where a near-infrared laser produces a genuine high-contrast one through a foaming or carbonisation reaction in the material. If a supplier quotes a job without asking what the substrate is, that's worth a follow-up question before the production run, not after.

Error correction: why the printed default falls short

QR codes carry spare data, calculated with Reed-Solomon error correction, so a partially damaged or obscured code can still decode correctly. There are four levels, and the percentage of the code's data that can be missing and still recover cleanly is fixed by the QR standard itself: L recovers around 7%, M around 15%, Q around 25%, and H around 30%.

Most QR generators default to M, which is a sensible middle ground for a clean printed sticker in a controlled environment. Denso Wave, the company that invented the QR code and co-develops the underlying standard, specifically recommends Q or H for codes that will live in dirty, damaged, or otherwise imperfect real-world conditions, and an engraved or etched mark is exactly that kind of environment even when nothing is technically wrong with it: surface texture, minor debris in the grooves, oxidation over time, and the glare problems above all read to a decoder as partial damage, whether or not a human would call the code "dirty."

The practical upshot: specify Q or H, not the platform default, for anything that's going to be lasered rather than printed. It's the single highest-leverage change available before a job goes to the engraver, and it costs nothing except slightly denser modules. If the design also needs a logo in the middle of the code, be aware that a logo and an engraving-grade error correction level are both spending from the same damage-tolerance budget, so stacking both without headroom is asking for trouble.

Sizing: what actually changes when there's no ink

The module-size and quiet-zone guidance written for print still applies in principle to an engraved code: the QR standard fixes a minimum quiet zone of four blank modules on every side of the code, and that number doesn't change because the marking method changed. What does change is how much margin an agency should build in above the bare minimum.

There's no ISO figure for a "minimum module size for engraving" the way there is for the quiet zone, and it's worth being direct about that: plenty of engraving-shop blogs quote a specific number, and they don't agree with each other, which is the tell that none of them are quoting a standard. The defensible version of the rule is a physics one: a module needs to be meaningfully larger than the laser's focused spot size, because a mark smaller than the beam's spot can't render as a crisp square, it renders as a blurred dot. Combined with the heat-affected-zone softening described earlier, most engravers working in fine fiber marking treat something in the region of a third to half a millimetre as a comfortable practical floor for a module on metal, not because a standard says so, but because that's roughly where the edges stop blurring into each other on the equipment they actually run. Treat any more precise number you're quoted as shop convention, and ask to see a test piece before committing to a print run's worth of plaques.

How the industry actually verifies these marks

Printed barcode and QR verification is governed by ISO/IEC 15415, which grades a symbol on a 4.0-to-0.0 scale (roughly A to F) based on contrast, modulation, and pattern integrity under a standard lighting setup. Direct part marks, whether dot-peened, etched, or laser-engraved, don't behave the same way under that standard lighting, for exactly the reflectivity reasons covered above. ISO/IEC TR 29158 exists to adapt that grading methodology for marks made directly into a surface rather than printed onto one: it modifies the same core parameters (cell contrast, cell modulation, fixed-pattern damage) and adds alternative illumination conditions suited to relief-based marks, rather than inventing a new scale from scratch.

Worth being precise about scope here: the standard's roots and most of its worked examples are in Data Matrix direct part marking for automotive and aerospace traceability, not QR code specifically. Machine-vision software vendors have since extended the same grading framework explicitly to QR codes, so it's genuinely applicable, just not "the QR standard" the way ISO/IEC 18004 is.

The part that matters most for an agency briefing a supplier is the lighting angles the standard defines for verification: a 90-degree diffuse, overhead angle; a 45-degree angle, the same angle used for standard printed-code verification; and a 30-degree low-angle, raking light. The lower the angle, the more a light source grazes the surface rather than hitting it straight on, and the longer a shadow it throws across any recessed cell, which is precisely what makes a shallow engraving or a dot-peened mark visible at all.

1. 90-degree overhead, diffuse light catches contrast created by colour or coating changes and produces almost no shadow, which is the easiest angle for a mark that relies on discolouration rather than depth.
2. 45-degree standard light is the same angle used to verify ordinary printed codes, and works reasonably well for a shallow, lightly-relieved mark.
3. 30-degree low-angle, raking light throws the longest shadows across a recess and is the angle that actually reveals a deep engraving or a dot-peened mark that looks nearly invisible under the other two.

The reason this matters beyond a verification lab: it's the same reason a client's phone under an overhead shop light might fail to scan a code that reads perfectly on a workbench near a window. If a supplier only ever tests under one fixed light, ask them to check the code held at a shallow angle to a lamp before signing off, because that's the condition most likely to expose a weak mark.

Laser power is a starting point, not a spec

It's tempting to ask a supplier for "the right wattage" the way you might ask for a Pantone reference. That number doesn't really exist. Laser manufacturers' own settings documentation, including official published guidance from equipment makers, consistently frames power and speed as a starting point that has to be tested and adjusted per machine and per material batch, not a fixed correct value. As a rough sense of scale rather than a spec: fiber lasers marking QR codes on stainless steel or aluminium commonly run somewhere in the 20 to 50 watt range depending on the alloy, coating, and whether the goal is a light surface anneal or a deeper cut, with higher power reserved for high-speed production lines rather than a one-off plaque. Treat any supplier who quotes one exact number with total confidence, without having run a test piece on your actual material, with a little scepticism.

Will it actually outlast a printed sticker?

Directionally, yes, and the reason is straightforward rather than mysterious: there's no ink layer to fade and no adhesive bond to fail. A vinyl QR sticker, even a UV-laminated one, is broadly understood to hold up for a few years outdoors before fading or lifting becomes a problem; an unlaminated paper code can visibly degrade within months of direct sun. An engraved or etched mark has nothing analogous to lose, so its functional lifespan tracks closer to the substrate's own lifespan than to a label's.

Be wary of a specific year count attached to this claim, though. The most concrete durability figure that circulates in this space, a 20-plus-year outdoor rating, belongs to photo-anodised aluminium asset tags (sold under names like Metalphoto), where the image is chemically sealed beneath the metal's own oxide layer. That's a genuinely tested, manufacturer-backed figure, but it describes a photochemical process, not laser engraving or etching, and the two shouldn't be conflated even though they end up looking similar. The honest claim for a laser-marked plaque or sign is qualitative: no fading, no peeling, resistant to the things that kill a sticker, without pretending to know a decade count nobody has actually tested for that specific process.

Where agencies are already doing this

This isn't a hypothetical niche. Industrial engravers already sell QR-marked trophies and awards that link to a video message or speech; metal signage suppliers sell weather-resistant QR plaques in stainless steel, brass, and aluminium explicitly on durability grounds; and laser-etched glass is a real, if pricier, alternative to a printed label on premium wine and spirits bottles. Warranty and asset-registration programmes are arguably the most serious use case: durable-goods manufacturers and industrial asset trackers are the exact audience ISO/IEC TR 29158 was built for, because a mark on a piece of factory equipment or a product housing genuinely does need to survive the life of the object.

There's also a real, established memorial products category, laser-etched plaques and medallions for headstones and memorial benches, sold by dedicated vendors as a permanent link to an obituary, photos, or a tribute page. It's a useful illustration of the whole argument in one product: nobody wants to re-engrave a headstone, but nobody wants the linked page to be frozen in 2026 either, which is exactly the tension the next section resolves. The same logic extends to any permanent outdoor signage an agency specifies in a durable material rather than a printed sign board.

The mark is permanent. The destination doesn't have to be

This is the point that makes engraving workable for a marketing use case rather than just an industrial one. A QR code doesn't encode a destination; it encodes a short string, usually a URL. If that URL points at a dynamic redirect rather than a hard-coded landing page, the pixel pattern lasered into the metal never has to change, even though where it sends people can be updated indefinitely from a dashboard.

That's the same mechanism behind why a printed QR code doesn't have to expire just because a campaign ends, and it's exactly how changing a QR code's destination without reprinting works for paper. Engraving simply raises the stakes: reprinting a sticker is an afternoon's job, but re-engraving a hundred trophies or a run of building signage is not, so the case for putting a short, dynamic redirect under a permanent mark is stronger here than almost anywhere else a QR code gets used. A memorial plaque can point at a simple tribute page today and a richer one built five years from now. A warranty tag can point at a registration form for one product generation and a support portal for the next. The plaque never needs a second trip to the engraver.

A pre-production checklist

Most of the failures above are cheap to catch before a full run and expensive to catch after one.

• Confirm the laser type matches the material: fiber for metal, CO2 for wood, leather, and most plastics, UV for glass.
• Set error correction to Q or H, not the platform default, before generating the code.
• Ask the engraver for a matte or bead-blasted finish over the code area where the material allows it, rather than leaving it polished.
• Size the module generously above a print-minded minimum; ask for a physical test piece rather than trusting a rendering.
• Test the physical proof under more than one lighting angle, including a shallow, raking angle, not just an overhead light.
• Point the code at a dynamic redirect rather than a hard-coded URL, so the destination can change even though the plaque can't.
• Run the same multi-phone, multi-lighting test pass an agency would use before any print run, on the actual physical piece, before committing to volume.

Frequently asked questions

Can you laser engrave a QR code?

Yes. Fiber lasers do it routinely on metal, CO2 lasers on wood, leather, and many plastics, and UV lasers on glass, but the settings, error correction level, and finish all need to be chosen for the specific material rather than copied from printed-code advice.

What's the minimum size for an engraved QR code?

There's no ISO-specified minimum the way there is for the quiet zone. The working principle is that a module needs to be several times larger than the laser's focused spot size to render as a crisp square rather than a blurred dot; in practice, many shops running fine fiber marking treat somewhere around a third to half a millimetre as a comfortable floor on metal, though that's shop convention rather than a published standard.

Which error correction level should I use for an engraved QR code?

Q or H. The QR standard's four levels recover roughly 7% (L), 15% (M), 25% (Q), or 30% (H) of the code's data if part of it is damaged or unreadable. Denso Wave, the QR code's creator, specifically recommends Q or H for codes exposed to dirty or imperfect real-world conditions, and an engraved mark counts even when nothing looks visibly wrong with it.

Why won't my engraved QR code scan?

Almost always glare or low contrast, not physical damage. A polished or curved surface can throw a bright specular reflection straight back at the camera, which a decoder reads as noise rather than as the code's pattern. A matte finish, a slightly larger module, and a Q or H error correction level fix the great majority of first-attempt failures.

What's the difference between a fiber and a CO2 laser for QR codes?

Fiber lasers run at roughly 1064 nanometres, a wavelength metals absorb well, which makes them the standard choice for stainless steel, aluminium, brass, and titanium. CO2 lasers run at roughly 10.6 micrometres, absorbed well by organic materials like wood and leather, but largely reflected by metal, so a CO2 laser is a poor choice for a metal QR code.

Can an engraved QR code still be dynamic?

Yes, and this is the detail that makes engraving practical for a marketing brief. The engraved pattern is just a short redirect string; ScanKit updates where that redirect points, so the physical mark never has to be re-cut when a campaign, a landing page, or a product line changes.

What is ISO/IEC TR 29158 and does it apply to QR codes?

It's the technical report that adapts standard 2D-code verification (ISO/IEC 15415) for direct part marks, marks made into a surface rather than printed onto one. Its roots are in Data Matrix marking for automotive and aerospace traceability, but the same grading approach, including its distinctive 30, 45, and 90-degree verification lighting angles, has been extended to cover QR codes as well.

Do engraved QR codes need a quiet zone?

Yes, the same four-module blank border the QR standard requires around any printed code applies to an engraved one. It's worth being generous with it in practice, because surface texture and heat discolouration around the mark can visually eat into a border that looks adequate on a design file.

Is engraving or printing better for outdoor durability?

Engraving, directionally, because there's no ink or adhesive layer to fail. Treat any specific year-count claim with caution: the widely cited 20-year figures in this space belong to a different, chemical (photo-anodising) process, not to laser engraving or etching.

Can smartphone cameras read laser-etched codes?

Yes, provided the mark has enough contrast and isn't drowned out by glare. A matte finish and testing the proof at more than one lighting angle before a production run resolves the great majority of readability problems phones run into.

The short version

Engraving and etching read contrast from surface relief and reflection, not from ink, which means the printed-code playbook needs real adjustments rather than a straight copy-paste: match fiber, CO2, or UV lasers to the material, set error correction to Q or H instead of the default M, size the module generously above print-minded habits, ask for a matte finish where the material allows it, and test the physical proof at more than one lighting angle before committing to a full run. None of that is exotic, and it's cheaper to specify correctly up front than to re-engrave a batch of plaques after the fact. Then put a dynamic redirect under the mark, not a hard-coded URL, so a genuinely permanent piece of metal, glass, or stone can still point somewhere new whenever the campaign behind it does.