Screen Print Separations in 2026: Manual, Automatic, or Send It Out?

Separations are where a screen printing job is decided. Get them right and the press run is uneventful. Get them wrong and you find out at the third colour, with the screens already burned.

There are three ways to get separations, and the useful question is not which is best but which suits the artwork in front of you.

The three routes

Automatic separation

Software splits the artwork into channels for you: spot colours, simulated process, index. Fast, cheap, and genuinely good on the artwork it suits.

Suits: flat vector artwork with defined colours. A three-colour logo separates perfectly in seconds, and paying a person to do it would be waste.

Falls apart on: gradients, photographic sources, and anything where the eye expects a smooth blend. Automatic separation of a gradient produces banding that looks like a mistake rather than a choice, because software cannot decide where one colour should stop.

Manual separation

A person decides the channels, the underbase, the trapping and the halftone angles.

Suits: photographic art, simulated process, tricky colours on dark garments, and any job where the press operator has an opinion about ink sequence.

Costs: time and skill, both scarce. This is why shops charge an art fee, and why the ones who do not charge it quietly absorb the cost instead.

Outsourcing

Suits: shops with a full press schedule and no dedicated art person, and jobs where the separation is worth more than an hour of the owner's evening.

Watch for: turnaround. A separation that arrives in three days is worthless on a two-day job, and that is the failure mode nobody advertises.

The test that tells you before you commit

Open the customer file and ask three questions in order.

Is it vector or raster? Vector with flat colours goes automatic. A raster of a logo goes to conversion first, because separating a bitmap of a logo separates its compression artefacts along with it.

How many real colours are there? Not how many the customer says. A JPG of a two-colour logo contains hundreds of colours, all the smeared intermediates between them. Screens do not print smears. Count the colours the design intends, not the colours the file contains.

Is there a gradient, and does it have to stay one? If yes, you are in simulated process or index, and that is a manual decision. If the gradient is decorative rather than essential, flattening it to two or three deliberate steps usually looks better on fabric than a halftone attempt at smoothness.

Why the file state matters more than the method

Every route above assumes the artwork can be separated. Much of what arrives cannot, and no separation method fixes it:

• A logo on a white rectangle separates the rectangle too. On a dark garment that prints as a white block around the design.
• JPG compression puts halos around every hard edge. Those halos become their own colour channel, and you print them.
• A 300 pixel source has no clean edges to separate. Enlarging it enlarges the mush.

The fix for all three is upstream: rebuild the artwork as vector with real flat colours before separating anything. Then automatic separation handles it, and you have saved the manual hour for the job that genuinely needs it.

Practical rule

Separate automatically when the artwork is vector and flat. Separate manually when the image is photographic or the garment is dark and the customer is fussy. Send it out when your press is full and the deadline is not tomorrow.

And before any of that, look at what the customer actually sent. You can run a file free on the screen print file converter to see whether it rebuilds into flat separable colours, or whether it is a job for a person. No account, and if the source is too far gone the preview shows you that too.