Registration for Multi-Colour Work: Keeping Every Layer Aligned

A one-color job either lines up or it doesn't — there's nothing to misalign against. The moment a design needs a second color, registration becomes the thing that separates a clean multi-color print from one where every layer is individually fine but the whole design looks slightly off. It's also one of the more expensive mistakes to get wrong repeatedly, because a misregistered piece usually can't be fixed after the fact — it has to be scrapped and reprinted.

What registration actually means

Registration is keeping every color's screen positioned identically relative to the garment, print after print, so each layer lands exactly where it's supposed to relative to every other layer. A logo with a black outline and a red fill only looks clean if the red sits precisely inside the black outline on every single piece — not just on the first test pull, but on piece 80 of a 150-piece run, after the platen has been loaded and unloaded eighty times and the press has been running long enough for small mechanical play to creep in.

Why this is harder than it sounds

Every screen on a multi-color job needs to return to the exact same position relative to the garment on every single pull. Presses handle the bulk of this mechanically — registration pins, indexed platens, and locked screen clamps all exist specifically to make repositioning automatic rather than something a printer eyeballs each time. But mechanical systems still drift: a platen with worn-in play, a screen clamp that isn't fully tightened, or a garment loaded slightly crooked all introduce small errors that a rigid mechanical system doesn't fully prevent on its own. Registration marks — small crosshairs or targets printed just outside the visible design area on the first color, then checked against on every subsequent color — are the practical way a printer catches drift before it reaches a finished piece rather than after.

Where registration actually goes wrong

A few specific failure points come up more often than the rest. Garment loading is the most common: a shirt loaded even slightly crooked on the platen shifts every subsequent color relative to where the first color landed, and this is a per-piece risk that mechanical registration systems can't fully eliminate, because they register the screens to the platen, not the garment to the platen. Screen clamp looseness is another: a clamp that isn't fully tightened lets a screen creep slightly out of position over the course of a run, producing drift that gets worse the longer the run goes rather than showing up immediately. Stretch and shift in the garment itself under squeegee pressure is a third, subtler cause — heavier fabric or a looser knit can shift slightly under the pressure of a stroke, especially on a large print area, moving the garment relative to a screen that hasn't itself moved at all. And thermal drift on a long run is a fourth, easy to overlook: as a press and its screens warm up over hours of continuous use, very slight dimensional changes in the screen frame or mesh tension can accumulate into registration error late in a long run that wasn't present when the job started.

The real cost of registration drift

Take a 3-color job that needs 150 good finished shirts. If registration drift causes an 8% spoilage rate — a plausible figure for a shop still refining its process on a design with tight color-to-color tolerances — the shop actually needs to run about 164 blanks to end up with 150 sellable pieces, since spoiled pieces still consume a blank garment and press time even though they can't be sold. At a $3.50 blank cost, those roughly 14 wasted garments alone come to about $49, before counting the ink, labor, and screen time already spent on each one before it was scrapped. Once that job's full cost — three screens' setup cost, ink for three colors, labor, garments, and overhead — is spread only across the 150 pieces that actually sold, the true cost per good shirt comes out to roughly $6.21, meaningfully higher than the $5.68 raw cost-per-piece the same cost model would show if it were spread across all 164 printed pieces including the spoiled ones.

Tighten registration discipline enough to bring spoilage down to 2% instead of 8%, and the same job needs only about 154 blanks instead of 164, and the true cost per good shirt drops to roughly $5.85 — a real, if modest, difference on a single job, and a much larger one compounded across every multi-color job a shop runs over a year. Registration quality isn't just a visual-quality concern; it's a direct cost lever that most basic pricing models don't surface unless spoilage is deliberately factored in.

Building registration checks into the run, not just the setup

Checking registration once at the start of a run and trusting it for the rest of the job is a common shortcut, and it's the reason drift-related spoilage often clusters in the second half of a long run rather than spreading evenly throughout. A better habit is spot-checking registration marks periodically through a run — every 20 or 25 pieces on a longer job — rather than only at the start, catching drift early enough to correct a screen clamp or re-check platen alignment before it produces a dozen bad pieces in a row instead of one.

It's also worth inspecting registration marks under raking light (light coming across the surface at a low angle) rather than straight overhead lighting, since raking light makes small ridges and edges — including slight ink-layer misalignment — much easier to see than flat, direct light does. A misalignment that's genuinely hard to spot under normal shop lighting can be obvious once you change how the piece is lit to check it.

Pricing registration risk in rather than absorbing it silently

Given that even a well-run shop won't hit zero spoilage on multi-color work, it's worth building a realistic spoilage allowance into multi-color quotes rather than pricing as if every printed piece will be sellable. A shop that knows its own typical spoilage rate on multi-color jobs — tracked the same way ink use or reclaim cost should be tracked — can price a quote to actually cover the blanks, ink, and labor spent on the pieces that won't make it, rather than discovering after the job that the "profitable" quote quietly wasn't, once spoilage is accounted for. This is the same principle behind pricing reclaim and screen wear into a job's cost model rather than treating them as background overhead absorbed silently — registration spoilage belongs in the same category of real, trackable cost.

Manual presses carry more registration risk than automatics

Automatic and manual presses handle registration differently, and it's worth understanding why a manual multi-station press demands more deliberate registration discipline than an automatic one does. On an automatic press, the platens themselves index mechanically between print heads, so garment position relative to each screen is largely fixed by the machine once a piece is loaded correctly. On a manual press, the printer physically rotates the platen or the print head assembly between colors, and that manual rotation is one more point where small positioning error can creep in on every single piece, not just occasionally. This doesn't make manual presses unsuitable for multi-color work — plenty of quality multi-color printing happens on manual presses — but it does mean registration-mark checks and a settled, repeatable loading routine matter more on a manual setup than on an automatic one, where some of that risk is designed out mechanically.

Registration problems and squeegee problems get confused

A multi-color piece that just looks "off" isn't always a registration failure — sometimes the colors are perfectly aligned and the real issue is uneven deposit thickness between layers, which is a squeegee-technique problem, not a positioning one. The two are worth telling apart before reaching for a fix, since correcting the wrong one wastes time and can mask the actual cause. Raking light across the piece makes the distinction clear: a true registration miss shows as physical offset between layers, where a deposit-consistency problem shows as color layers that line up correctly but look uneven in coverage. The technique side of that distinction, including how to diagnose which lever — durometer, angle, or pressure — is actually responsible for an inconsistent deposit, is covered in squeegee durometer, angle, and pressure.

Improving registration is mostly about eliminating variables, not adding steps

The most effective registration improvements tend to be about removing sources of variation rather than adding new checks. A garment-loading jig or a consistent, marked loading position on the platen removes the "slightly crooked" variable at its source rather than catching it after the fact with a registration-mark check. Fully tightening screen clamps and periodically verifying they haven't loosened removes the slow-creep variable. And allowing a long run to reach thermal equilibrium before committing to a full print count — running a handful of warm-up pieces and checking registration again once the press has been running a while — catches thermal drift before it becomes a run-long problem instead of discovering it fifty pieces in.

Registration spoilage sits alongside reclaim cost and screen wear as one of the real costs a basic pricing model tends to miss until it's tracked deliberately — the full worked accounting of those is covered in costing a print run properly.