Colour Separation for Screen Printing, and What Software Cannot Decide

Screen printing lays one flat ink per screen. Every screen is a setup cost and a registration risk. So the whole craft of preparing artwork for screen is deciding how few inks can carry the design — and that is a judgement, which is why fully automatic separation still produces files that cannot be printed.

The three separation approaches

1. 1**Spot colour.** Each flat colour in the design gets its own screen and its own mixed ink. Cleanest result, most accurate colour, best for logos and flat graphics. Cost scales directly with colour count.
2. 2**Simulated process.** A handful of inks printed as halftones that blend optically to suggest many more colours. This is how a photographic image gets onto a garment with six or eight screens instead of thirty.
3. 3**Four-colour process.** Cyan, magenta, yellow and black, as in paper printing. Works on white garments, struggles on dark ones, and is less used on apparel than the other two.

What automatic separation does well

Modern separation tools are genuinely good at the mechanical parts: pulling a clean spot separation out of vector artwork, generating a competent underbase for dark garments, and producing halftone channels at a specified angle and frequency. For flat vector logos, automation is close to a solved problem.

Where it goes wrong

The failures are all cases where the software optimises for visual similarity to the original file, and the printer needs something else.

• **Too many channels.** Software will happily hand you a fourteen-colour separation of an image that a human would build in seven. Every extra screen is setup cost, another registration alignment, and another thing to go wrong on press.
• **Halftones too fine for the fabric.** A dot frequency that resolves beautifully on paper disappears into the knit of a fleece. Fabric texture sets the practical ceiling, not the file.
• **Underbase that is too aggressive.** An automatic white underbase pulled from the full image makes prints heavy, stiff and plasticky. Choosing where the underbase does and does not go is what makes a print feel soft.
• **Ignoring trapping.** Adjacent colours need slight overlap so a fractional misregistration does not show garment through the gap. Software often omits it.

The question nobody asks the software

Separation software is never told the quantity, and quantity changes the right answer completely.

At 20 pieces, a seven-colour separation is absurd — the setup dominates the job and DTF will produce a better result for less. At 1,000 pieces, those seven screens amortise to almost nothing per piece and screen printing gives a better hand feel and better durability than any digital method. The same file deserves different treatment at different volumes, which is a commercial decision rather than a graphical one.

What to check on a separation proof

1. 1**Colour count against quantity.** Does the number of screens make sense for the run size?
2. 2**Underbase coverage.** Is white going down only where it is needed, or across the whole image?
3. 3**Halftone frequency against fabric.** Coarser for fleece and textured knits, finer for smooth combed cotton.
4. 4**Trapping on adjacent colours.** Small deliberate overlaps where two inks meet.
5. 5**The darkest and lightest areas.** Detail dropping out in shadows and blowing out in highlights is the usual sign of a separation tuned for a screen rather than for ink.

The practical position

Let software do the mechanical work and have a human make the decisions about colour count, underbase and halftone. If you are ordering rather than printing, you do not need to do any of this yourself — but knowing what a good separation looks like is what lets you tell a proof that will print well from one that will not.