Prepare Color Separations for a Printed Prop Graphic

A prop only becomes real at the moment somebody has to name the inks. Until then the tag on the bottle is a composite — a picture your monitor assembled out of whatever numbers happened to be in the file. Afterward it is a stack of plates, one per ink, each making its own pass over the same sheet. That stack, not the picture, is what shows up in the test shot.

So the first question isn't "how do I separate this file." It's "what did the shop say it's printing with." Get that answer before you touch a swatch, because nearly every mistake in this job comes from preparing the wrong kind of output very carefully.

What follows is one invented job, worked through in order. Nothing here was printed, measured or run — the tag, the shop and the plates are fictional so that the checks have something concrete to attach to. Where a step rests on Adobe's published documentation rather than on invented detail, it's marked.

Ask what is being printed, and with which inks

Get five things in writing from whoever is producing the printed piece:

• the printing process — digital, offset, something else
• the substrate, by name and weight
• the finished dimensions
• the ink list: the actual inks, named
• the handoff they want — which file, in which format, with what settings

The ink list is the part people skip. It's also the part that decides everything downstream, because it tells you which colors will print as process builds and which will print as named spot inks. Those are different jobs. Four process inks reproduce a photograph well and make fine type miserable. A named spot ink prints a flat, exact color that process can only approximate, and it earns its keep when a design uses two or three colors and holds small type.

Two colors that look identical on screen are not necessarily the same output ink. A green set as a named spot prints from one plate at a strength the shop mixed. A green built from process percentages prints as dots from four plates and shifts with everything else on the sheet. Your eye can't tell the two swatches apart in the file. The plates can, and so can the invoice.

The invented job: a neck tag, 45 by 90 mm, on uncoated cream cover stock, appearing in a test shot. The shop's email names two inks — a green, by its own ink number, and black. In the file, the green is a named spot swatch and the black is the file's process black, which is the same plate the shop's list means by Black. Two colors, two plates. The tag has flat color, some very small type and one deliberate overlap. It is exactly the kind of prop that goes wrong quietly.

Now make a copy of the artwork for preparation and leave the approved composite alone. Prepress work means deleting objects and reassigning swatches, and you want the version the designer signed off on to still exist when someone asks what the tag was supposed to look like.

One thing to resist: running the document through a generic conversion preset before you understand the job. Some output settings quietly convert spot inks into process builds, which would take this two-ink tag and hand the shop four process plates it never quoted. The preset looks helpful. It's answering a different question.

Inspect the graphic one plate at a time

Adobe's help page on printing color separations in Illustrator (helpx.adobe.com/illustrator/using/printing-color-separations.html, documentation checked 2026-09-18) treats the preview, the output controls and the document ink options as three separate things — and that separation is the whole point of the page. Find the preview in the copy you're running rather than following a screenshot from an older release.

The preview does two useful things: it can isolate a single ink plate so you can look at it alone, and it can show how objects set to overprint will actually behave. That second one gets forgotten, and it's the one that answers a question the composite can't.

Walk the tag one plate at a time.

The green plate. Isolate green and the tag reduces to a solid band across the lower 28 mm and a leaf glyph at the upper left, sitting in a 60% tint of the same ink. That's everything on it. Worth noticing: the tint reads as a paler color in the composite, but it costs nothing extra. It's a screen of the same plate, not a third ink.

The black plate. Isolate black and you should see the wordmark, a tagline, a small line of format numerals and a 0.5 pt keyline running inside the trim — and nothing else you put there. Read the whole list of small marks, not just the biggest one. Fine type and hairlines are the first things to go missing in a separation, and the first things to look wrong on press. Marks you can't account for carry forward; the pass that hunts stray inks will name them.

The overlap. The lower third of the wordmark crosses the top edge of the green band. Isolate the green plate and look at the band where the letters sit. If the band is continuous underneath the wordmark, the black is set to overprint and the two inks will stack. If you see letter-shaped holes punched in the band, the black is knocking out — the green is removed there and the letters land on bare paper. Both arrangements are legitimate. They are not the same result. With a knockout, a hair of misregistration leaves a pale outline around every letter. With an overprint, the same shift hides inside the dark stack. The wordmark crosses the band precisely so the design can absorb that shift, and the composite shows you the same black letters either way. Only the isolated plate tells you which one you have.

The plate nobody asked for. Now look for inks that shouldn't be there. In this invented file, a small roundel sits at the tag's top right — a solid warm-red circle, about 9 mm across, with "SMALL BATCH" set in black over it. It came out of a default swatch library a while back and nobody has thought about it since. Isolate magenta and there it is, alone on the plate. Yellow, the same. Cyan, a pale wash you'd never register in the composite. Black, the faint 10% ghost of the circle sitting behind the type — the unaccounted mark carried forward from the black plate.

That single object has turned a two-ink tag into a five-plate file: cyan, magenta, yellow, black, plus the green spot. Nobody chose that. The composite never showed it. The plate list did.

Keep output separate from preview

Here is the trap that catches people who have followed everything so far.

When you find the roundel, the preview offers a tempting fix: hide the magenta plate, hide the yellow, look at the composite again, and the tag is clean. Adobe's documentation states plainly why this doesn't work — hiding a separation in the preview does not disable its output. The visibility controls are a viewing tool. The plates still go.

Output is a separate set of decisions in a separate place, and that's where the file's ink list is actually settled: whether a plate prints at all, and whether a spot ink prints as itself or gets converted to a process build. Adobe's page lists the document ink options as their own area for exactly that reason.

So the remedy for an unrequested ink is one of two things: change the artwork, or get the shop's agreement to add the ink. Nothing you do to a preview changes the file.

In the invented job, the artwork changes. The designer deletes the roundel's red fill and the 10% black ghost sitting behind the type, keeps the words in black, and adds a 0.5 pt black ring so the badge still reads as a badge. The tag is back to two plates: the green spot and black. That correction is a design decision, which is why the designer makes it and not you.

Then come the output settings, and here you follow instructions rather than taste. If the shop asked for a print-ready PDF with the spot ink preserved, check that it was preserved by opening what you exported and confirming the ink list on the other side — don't assume the setting you clicked did what its label promised. If the shop asked for separations sent directly to a device, that's a different route, documented on the same page, and it isn't automatically the right handoff for every shop. If the shop supplied angles and trapping values, use theirs.

The last move is to read the delivered file the way you read the source: plate by plate, ink list against the agreed inventory. A preview eye icon is a habit. The file is the fact.

Reserve press-dependent questions for a proof

Plates tell you what is being printed. They cannot tell you what it will look like, and it's worth being precise with yourself about that gap, because it's the difference between asking the shop a useful question and telling the shop how to run its press.

What the inspection established: two plates exist, the green carries the band and the leaf tint, the black carries the type, the keyline and the badge, the overlap is an overprint, and the roundel's process inks are gone. That's a real, checkable result. It's also silent about everything that happens once ink meets paper.

What plate inspection can't tell you:

• whether that green, on uncoated cream stock, goes duller and darker than the swatch on your screen
• whether the 60% tint reads as a leaf or as a gray smudge
• whether the 0.5 pt keyline holds on this stock or drops out
• whether the overprinted wordmark looks black over the band or turns muddy where the inks stack
• whether two plates register closely enough on a tag this small to keep the overlap invisible

Those are the shop's questions and the answers live on paper. Ask for a proof — and ask which proof actually shows the thing you're worried about, since a digital proof and a press check don't answer the same question. Describe what you need to see rather than naming a method.

One habit worth keeping: view the proof the way the camera will. This tag is small in frame. On a bottle in a wide test shot it might occupy sixty pixels of height on a 1080-wide screen. Zooming to 400% will show you problems the audience can't see and hide the one they can. Hold the proof at shot scale and ask whether the tag reads as a two-color tag with a leaf and a badge. If the small numerals melt into the paper at that size, that's a real problem. If they're merely imperfect under a loupe, it isn't.

And don't hand a shop screen angles or trapping values it didn't ask for. Those depend on the press, the stock and how the shop runs its ink. You have the artwork and the plate inventory. The shop has the press. Ask the question your own evidence can't answer and let them answer it.

The corrected inventory, and what still needs a proof

The handoff note for the invented tag reads like this:

Nettle Row neck tag, 45 × 90 mm, uncoated cream cover. Two plates: Green (spot, your ink number) and Black. Green plate: solid band, lower 28 mm; leaf glyph in a 60% tint of the same ink. Black plate: wordmark, tagline, format numerals, 0.5 pt keyline, and the badge — "SMALL BATCH" set in black inside a 0.5 pt black ring. Wordmark overprints the green band — confirmed by isolating the green plate; the band runs continuous beneath the letters. Composite and editable master attached. The roundel's process-color fill and its 10% black ghost have been deleted from the artwork; the badge survives as black only, and nothing else remains on the black plate beyond the marks listed above. The file's ink list is back to the green spot and black, with no cyan, magenta or yellow content. Open for your proof: tint value of the leaf on this stock, appearance of the overprint where the wordmark crosses the band, and whether the 0.5 pt keyline holds.

That's the whole job. Two plates, inspected one at a time, with the correction made in the art instead of hidden in a preview, and the press-dependent questions handed to the person holding the press. Notice what the note doesn't contain: no screen angles, no trapping values, no claim that the tag is press-ready. It says what the file holds and what a proof still has to settle.

The tag is still a composite on your screen at the end, and it always will be. What changed is that you now know which inks are going onto the paper — and which questions you've sent to a proof instead of answering yourself.

Frequently asked questions

What should be requested from the shop before preparing separations?

Get five things in writing: the printing process, the substrate by name and weight, the finished dimensions, the actual ink list with inks named, and the handoff they want—which file, in which format, with what settings. The ink list is the part often skipped and it decides what prints as process builds and what prints as named spot inks. Four process inks reproduce a photograph well but make fine type miserable; a named spot ink prints a flat, exact color that process can only approximate, and it earns its keep when a design uses two or three colors and holds small type.

Why inspect the artwork one plate at a time?

The preview can isolate a single ink plate and show how overprint objects behave. In the invented tag, isolating green shows the solid band, the leaf glyph in a 60% tint, and whether the band runs continuous beneath the wordmark. If it does, black is overprinting; if there are letter-shaped holes, black is knocking out. Both are legitimate but not the same result: knockout misregistration can leave a pale outline, while overprint can hide the shift inside the dark stack. The composite shows the same black letters either way; only the isolated plate tells you which arrangement you have.

What does the unrequested roundel reveal?

It reveals how a single object can turn a two-ink tag into a five-plate file. The roundel's red fill sits on magenta and yellow, a pale cyan wash, and a faint 10% black ghost behind the type, plus the green spot. Nobody chose that, and the composite never showed it; the plate list did. Hiding separations in the preview does not disable their output—visibility controls are a viewing tool, and the plates still go. The remedy is to change the artwork or get the shop's agreement to add the ink; nothing done to a preview changes the file.

What can plate inspection not tell you?

Plates can establish what is being printed, but not what it will look like. They cannot tell whether the green on uncoated cream stock goes duller or darker than the screen swatch, whether the 60% tint reads as a leaf or a gray smudge, whether the 0.5 pt keyline holds or drops out, whether the overprinted wordmark looks black over the band or turns muddy where the inks stack, or whether two plates register closely enough on a small tag to keep the overlap invisible. Those are press-dependent questions that live on paper. Ask for a proof, and ask which proof actually shows the thing you are worried about, because a digital proof and a press check do not answer the same question.

How should the proof be viewed and the handoff described?

View the proof the way the camera will. The tag may occupy only about sixty pixels of height on a 1080-wide screen in a wide test shot; zooming to 400% shows problems the audience cannot see and can hide the one they can. Hold the proof at shot scale and ask whether the tag reads as a two-color tag with a leaf and a badge; small numerals melting at that size is a real problem, while loupe-level imperfection is not. The handoff note should list the two plates and their contents, confirm the overprint by isolating the green plate, and state the open proof questions. It should not include screen angles, trapping values, or a claim that the tag is press-ready, because those depend on the press, stock, and how the shop runs its ink.