PDF/X-1a (ISO 15930-1, CMYK Print-Ready)

PDF/X-1a:2001 (ISO 15930-1) is the strictest and the oldest of the print PDF profiles, and it is the one print houses name most often when they ask for a "printable PDF". Its rule is simple: nothing may be left in the file that production would have to guess. All colour is in CMYK inks, there is no transparency, fonts are inside the file, trim and bleed boundaries are set, and the print condition is described by an embedded profile.

The difference from the neighbouring PDF/X on this site is the colour. That file is PDF/X-3: it stays in RGB with an embedded sRGB profile, and the conversion to ink is done by the print house's prepress. Here the conversion is already done: the file carries DeviceCMYK only, and no RGB is left anywhere — not in fills, not in labels, not in raster inserts. The print house just prints it. The print condition is ECI "ISO Coated v2 300%": ISO 12647-2 offset on coated paper with a 300% total ink limit. The profile is free and embedded right into the file; it is also declared as the output intent, so the sheet carries its print condition with it and does not depend on someone else's press settings. Because of the embedded profile the file is about 1.3 MB heavier than a plain PDF.

A word about sheet size. The 3 mm bleed — the allowance for an imprecise trim — is added OUTSIDE the ordered format, it is not cut out of it. The sheet is rendered 6 mm wider and taller than ordered, the TrimBox (the trim line) matches the ordered size and the BleedBox is the whole sheet. After trimming at the print house you hold exactly what you ordered: A4 stays A4 instead of turning into 291×204 mm. The map margin is measured from the trim line as well, so the map itself, the legend and the source credit all lie inside the finished sheet.

A word about black. Black and grey are separated onto a SINGLE ink — the K plate, not a mix of four. This is the same separation CorelDRAW's native export performs, and it is done for the sake of register: black text and hairlines are laid in one pass, so a shift between the press units cannot put a colour fringe around labels and thin lines. The second half is overprint: small black text and hairlines are marked to print ON TOP of the inks below instead of knocking a hole in them, so the same shift opens no white paper underneath them. Large black areas and display type still knock out — the ink beneath would show through them and change the black. There is deliberately no trapping: the print house's RIP does that for its own press and paper, and bad trapping is worse than none. Nor is there a colour proof, which is a physical sheet off a press rather than a property of a file. Semi-transparent fills are flattened against the page background beforehand — the standard forbids transparency. The finished file is formally validated and is not handed over if it fails: a file that calls itself PDF/X-1a and is not one gets turned away after the press time has been paid for.

Spot inks. Up to two inks of your own may be added to the four: you give the ink a name — "Die-cut contour", "Varnish", a corporate colour — its CMYK equivalent, and the map layer that prints in it. The name goes onto the plate, and the print house mixes the paint by it; the equivalent is what a four-colour run lays instead. We ship no library of ready-made commercial colours — the inks are the ones you describe. In the file this is a /Separation over DeviceCMYK, a construct PDF/X-1a explicitly allows.

The limits of the spot inks are stated up front. An ink is assigned to a whole layer, not to individual objects. The layer's labels are drawn by its own labelling engine in its own colour — they print in composite CMYK and not in the ink; the layer's translucent objects go to CMYK as well, because PDF/X-1a has no transparency, and a spot flattened against the background stops being an ink. A layer whose colour comes from a data-defined expression or hides inside an SVG symbol or a raster cannot carry an ink: there is no way to tell exactly what to paint, so such an order is refused, naming the layer and the reason, rather than half-painted. And if the layer draws nothing inside the area you ordered, no plate for that ink is added to the file and a warning comes with it — an empty plate is still mounted on press and still billed.

Two more grounds for a refusal, from the same rule. If the layer's colour coincides with something else on the sheet — the page background, the mandatory source credit, the legend, or another layer's labels — the ink is not assigned. Marks on a plate are told apart by colour alone, and a coincidence would carry the wrong thing onto it, up to the whole sheet. For the same reason a gradient fill, and a layer with a paint effect on it — blur, shadow, colourise — cannot carry an ink: there the colour is either not one colour, or the mark becomes a picture. Both are named in the refusal rather than printed by halves.