2D codes and watermarks: what packs and lines must deliver

Quick answer: If you have to pick one thing to prepare for, make it the GS1 Digital Link 2D code. It has a published date (end of 2027 for retail checkouts to be able to read it), and the work lands on your coding, printing and vision inspection steps. Digital watermarks are a different tool aimed at recycling sorters, and the evidence so far comes from trials, not routine production.

Both technologies put identity on the pack. They ask different things of the line, and they answer to different people: retailers and brand owners for the 2D code, recyclers and consortia for the watermark.

What a GS1 Digital Link code actually is

GS1 Digital Link writes identifiers such as the GTIN, batch and serial number as a web address, so one QR code or NFC tag can serve checkout and online information. A retail scanner reads the identifiers out of the address. A phone follows the address to a web page.

The commercial hook is “Sunrise 2027”. GS1 US describes it as the goal that US retail point-of-sale systems can scan 2D barcodes, such as QR codes with GS1 Digital Link and GS1 DataMatrix, by the end of 2027. Commentators are consistent on what it is not. A Forbes Business Council piece from September 2026 says it is neither a law nor a worldwide switch-off date, and a separate guide notes that the date for retiring 1D barcodes, which GS1 calls “sunset”, has not been fixed globally.

So for the transition, GS1 US says items will require dual marking with U.P.C. and 2D barcodes, and U.P.C. barcodes will continue to be accepted and scanned after the 2027 sunrise date. Dual marking is where packaging engineers feel it first: two symbols, one panel.

On the regulatory side, the EU Digital Product Passport (under ESPR) and PPWR may both touch how packs carry data. For the detail on PPWR, see ppwreu.org, and check the current text before you act on any date.

What the pack and the line have to deliver

A 2D code is only useful if it reads. That sounds obvious until you look at where it can go wrong: low contrast, a distorted symbol on a flexible film, a code printed too small for the imager. The guide site gs1digital.link lists open hurdles that include the fact that GS1 needs to extend the General Specifications to cover placement of a final 2D symbol, and of a 2D symbol alongside UPC/EAN. In other words, the placement rules for your pack format may still be moving, so confirm them with GS1 and your retail customers.

Checkout hardware matters too. A linear UPC can be read by a laser-based scanner, while 2D codes need an imager that captures the whole symbol. That is a retailer investment, not yours, but it explains why the timetable is staggered.

Verification is a standard, not a feeling

Print quality for 2D symbols on labels and packs is graded under ISO/IEC 15415. Sources differ on the count of graded parameters: one vendor guide says 10 optical parameters, while a verification whitepaper from Emptech says 8 different parameters. The count depends on how the parameters are grouped, so use the current standard text rather than either summary. What the sources agree on is the logic: the lowest individual grade becomes the overall grade.

A GS1 implementation guideline adds two points relevant to line engineers. It says that measurements should be made on the symbol in its final configuration, as it is intended to be scanned, because the production process strongly influences the values. And it says a minimum overall symbol grade for acceptability should be set before quality control inspection starts. For a Digital Link code on a printed pouch, that means grading the code on the actual film, not on the proof.

Omron’s technical material explains that an ISO 15426 compliant verifier grades both print quality and data format (GS1 data format being one example). The distinction between a reader and a verifier matters here. Adams Corp puts it simply: readers decode, while verifiers test the quality of the printed barcode and grade it against parameters that influence readability.

Where the camera sits

Cognex describes a verifier as typically placed where barcodes are printed or applied to products, with some plants using inline verifiers and others using offline stations for batch checks. EyeC, which sells print inspection systems, describes inline inspection that reads both barcodes and variable codes and, for variable codes, identifies duplicates. Duplicate detection becomes more relevant as Digital Link codes carry batch and serial data instead of a fixed GTIN.

If you are building the data string, the GTIN inside it still needs a valid check digit. Our GTIN validator covers that part. It does not validate the full Digital Link syntax; gs1digital.link offers a free validator for that.

Digital watermarks: a different job entirely

A digital watermark is an imperceptible pattern spread over the pack surface. The HolyGrail 2.0 initiative describes the codes as the size of a postage stamp, covering the surface of consumer goods packaging, readable in a sorting plant by a standard high resolution camera on the sorting line. The sorter then routes the pack by attributes such as food versus non-food.

For the printer, the supplier’s pitch is low friction. Digimarc says its watermarks are applied into the artwork of product labels without special inks or printing processes, with enhancement tools provided to supply chain partners. That is a vendor claim, and a packaging team will still want to see how the watermark behaves across its own substrates, inks and press conditions. In the Belgian trial, converter Constantia Flexibles integrates the watermark directly into the packaging design during printing.

Notice what is absent from the line. There is no coder to add, no extra camera at the filler, and no new data string on each pack in the SKU-level case. The work sits in artwork, prepress and a database linking each watermark to product attributes. Digimarc’s guide describes a parent-child link between the product and its bill of materials so recycling data can be attributed at brand and SKU level.

Four identification options side by side

Option Visible to shoppers? Read by What it asks of pack and line Evidence status
UPC/EAN (1D) Yes Laser or imager at checkout Existing print and verification routines Established; stays accepted after 2027 per GS1 US
QR with GS1 Digital Link Yes Imager at checkout; phones Valid URL structure, verified print quality (ISO/IEC 15415), space for dual marking Retail readiness target for end-2027; pilots in progress
GS1 DataMatrix Yes Imager at checkout; line scanners Same grading approach; placement per GS1 General Specifications Named by GS1 US as a 2D option for POS
Digital watermark No (imperceptible) High-resolution cameras at sorting plants Embedded in artwork or substrate; attribute database linked to SKU Industrial trials; Belgian flexibles trial started September 2026

The first three are for the checkout and the supply chain. The last is for the waste stream, and the two families can sit on the same pack without conflict, since one is a visible symbol and the other a pattern in the artwork.

What the watermark trials have shown

The published HolyGrail 2.0 numbers come from several trials with different set-ups, so treat the chart below as a set of separate results, not a single measurement series. Each figure is a percentage reported by the initiative or its partners.

Reported HolyGrail 2.0 watermark sorting results (separate trials)
Reported HolyGrail 2.0 watermark sorting results (separate trials)

The 2022 semi-industrial test used approximately 125,000 packs from 260 SKUs at 3 m/s belt speed, and reported 99% detection, 95% ejection and 95% purity on average.

The larger test ran at Hündgen Entsorgung’s material recovery facility in Swisttal, Germany, which processes 100,000 tons of waste a year from three million households. Over 100 days it logged 5.66 million detections of 5,949 unique SKUs. Detection efficiency ranged from 87.9 to 93.8 percent for rigid packaging in single-pass sorting. The initiative expects two-pass sorting, common in plants, to improve that.

Flexibles are harder. For the baled, contaminated and aged streams assessed in February 2024, single-pass results were 89.9% detection efficiency, 75.1% sorting efficiency and 88.1% purity. The initiative itself notes that the conditions, with throughput up to 2,500 kg/h and repeated baling, were harsher than normal industrial settings.

The 2026 Belgian trial

The next step is the HolyGrail 2030 trial led by Fost Plus. Mondelez International, PepsiCo and Pladis are taking part, and the consortium describes it as the first national-scale European trial involving flexible post-consumer packaging collected through a household recycling system. Film sorted in Belgium goes to Hündgen’s German facility, where Pellenc ST cameras detect and read the watermarks.

The question being tested is whether watermarks can show a sorter that a flexible pack once held food, enough to support food-grade recycled content. No results from that trial were published in the sources reviewed here, so anything claimed about its outcome would be speculation.

Where AI and machine vision fit in

None of the 2D code or watermark sources I found describe AI doing the verification; grading under ISO/IEC 15415 is a rules-based measurement. AI shows up around it. Packaging Gateway, in a September 2026 feature, lists AI, robotics, machine vision, digital twins and connected production systems as tools for handling product variation, quality and unplanned downtime. That is the wider automation context in which a code verifier lives.

Investment appetite is there for the coding and printing step specifically. A market report citing PMMI beverage data says 68% of beverage companies expect moderate-to-significant increases in packaging machinery investment over the next two to three years, with coding, labeling and printing at 46% of planned purchase categories. Treat that as a signal from a secondary source, not a measured forecast for your plant.

For buying decisions, Packaging Gateway’s 28 October 2026 feature advises comparing, once production has stabilised, saleable output, downtime, reject rates, changeover times, resource consumption and operating costs against the agreed baseline. A verification camera that rejects more packs than the old process did is a reject-rate item, and it should show up in that comparison.

There is also a caution from the design side of AI. In a Packaging Digest interview published 27 October 2026, retired CPG packaging leader Jay Gouliard worries that leaning on AI for knowledge people should hold themselves might undermine capability-building in the person. For a team writing code specifications or vision inspection recipes, the point is practical: someone still has to understand why a symbol failed.

What is still unproven

Several gaps remain, and they differ by technology.

  • Placement rules for 2D codes. GS1 has said the General Specifications need extending for final and interim symbol placement (via gs1digital.link). Until then, label and pack layouts carry a risk of rework.
  • Retail readiness. Checkout imagers have to be rolled out and staff educated. The date is a goal, and individual retailers may set their own supplier requirements, as one compliance guide notes.
  • Watermarks on flexibles. Results to date are single-pass and below the rigid-pack figures for sorting efficiency. The Belgian trial is the first national-scale test, with outcomes pending.
  • Who pays and who benefits. HolyGrail 2.0’s stated aim includes proving the economic viability of the business case at large scale. That economic proof is not in the trial numbers above.
  • Scale of adoption. A trial with 79 or so participating organisations is large for the sector, but it is still a trial.

What this means for you

Packaging engineers and line teams. Ask which of your lines can print and verify a 2D symbol at line speed, on each film or board you run. Check where the verifier would sit (at the coder or at the end of the line) and whether it can grade to ISO/IEC 15415 with the data format check for GS1. Grade the code on the finished pack, since GS1’s guideline says the production process affects the result.

Brand and design teams. Plan the panel for dual marking now. Confirm placement with your key retail accounts before artwork is released, because the General Specifications were still being extended in the sources reviewed. A QR code that is not a GS1 Digital Link code will not do the same job at checkout, as the Forbes piece warns.

Procurement. When you request quotes for coders, vision systems or verifiers, ask suppliers for the standards each device grades to and for their test conditions. Ask your supplier for a quote and a trial plan; do not accept a rated speed on its own.

Sustainability and R&D. Digital watermarks are worth tracking if you run flexible food packaging and have recycled-content targets. Ask your converter how the watermark is integrated into the artwork, and wait for the Belgian trial outcome before building a business case. The lightweighting calculator and printable checklists can help with the parallel work on pack specs.

For any PPWR or EPR question this touches, check the current text at ppwreu.org with your regulatory colleagues.

Sources