While CMYK can produce bright, crisp colors, as done with this print, it is more difficult to create unique or specific colors.

Artist or Engineer?

Joe Clarke has spent the past 47 years in the lab and in the engineering department, in pre-press and on-press, as an R&D / technical researcher and as a manager of screen print production. Clarke has held executive positions as President of M&R Printing Equipment and as Vice-President at Wilflex [Poly One]. He has been granted a growing number of print-related patents, including one for High-Shear printing with Smilin'Jack - he is a member of the ASDPT, is an Associate Editor for NBM and an SGIA Fellow.

Clarke has presented hundreds of technical papers, written a couple books and published over 600 technical / management articles for which he has been awarded five Swormstedts; the international standard for excellence in technical writing.

Currently Joe Clarke is the President of CPR, a Chicago-based corporation which manufactures Synergy Inks including NexGen; environmentally & financially responsible T-Shirt inks. For more information on CPR, visit

T-shirt printers are the only medium out of digital, direct-to-garment, flexo, gravure, sheet-fed, web offset, and screen printing that don’t use either CMYK or RGB to create full-range, vibrant color. They used to say that four-color process doesn’t pop, but direct-to-garment printing and a select few screen printers have shown one can get exceptionally bright color with CMYK. It’s also difficult to provide an accurate proof for simulated process printing, but can we make a case for each?

There are three questions I ask of myself as a press operator when I believe I’m fully prepared to run a process job on press. What time will I start, what time will I be ready for approval from the boss, and what time will I finish printing the order? If I circumvent the triad of questions even once, I’m not fully prepared. Predictability, consistency, and quality are all suspect. Plus, I issue a red flag every time I use conditional phrasing, such as “if,” “then,” “considering,” “depending,” “if it’s like the last time,” and “God willing.” We value press time at $400 per hour, so for the stockholders, stakeholders, and software selection, these are three key decision makers for simulated or CMYK.


In deference to graphics printing, T-shirts rarely have an explicit, measurable specification within narrow tolerances for color. T-shirt prints don’t usually have to look accurate. They simply need to look good, and art often supersedes reality. Unlike six-color sheet-fed or six-color digital presses, we screen printers have nearly unlimited slots, and the cost of the extra screens doesn’t slow us down.

When a press is available for experimentation or qualification and the run length is so long, the cost of developing the color on press can be amortized. When the garment color is low value or highly saturated and we are without a calibrated white underbase and a set of transparent colors to print on, the base simulated makes sense. Or if most of the colors are out of gamut for CMYK++; metallic, fluorescent, and inorganics; ochre and burnt sienna; and reflex blue, these can’t be run with four transparents.

If you’re into squeeze and mash to print your dots, you can’t deal with the precision required for CMYK. With CMYK, color is built by combinations of imposed and juxtaposed dots.


If one needs to simulate reality, memory, or natural colors precisely, there’s arguably no more efficacious nor accurate reproduction method than CMYK. When press time is at a premium and the system is color managed, CMYK is the most cost-effective method on press.

Nonetheless, T-shirts aren’t always white, and certain colors are out of gamut—they cannot be achieved with the color set. Because you probably have a PMS color naming book, open it to the front and you will see CMYK, in addition to pantone green, reflex blue, and a baker’s dozen or so of others. No matter how hard we try, nor dim the lights, we can’t match these unique colors with CMYK.

Core components

We’re almost always stuck with the shirt quality—or lack thereof—because it’s far more costly than all other components combined: the frame plus mesh, stencil, film, blade, ink, and labor. If we were to print at 100 per hour, the press time would cost more than the shirt. When we print at 800 per hour, it costs far less.

The point is: don’t be foolish. Get the finest core components you can find, and you’ll save money in the long and short run and increase your capacity, customer relations, and bank account.


If you want to do CMYK++ on dark shirts, find a low surface tension, high-shear white with lots of electrons on its gelled surface. If your supplier has to count them and get back to you, consider looking for an alternate supplier. High shear means the white runs at top speed on the press, which is the key to matte down and smoothness. Both are critical for CMYK++ on darks.

Cut a square for the underbase white and the visible whites will look fine as will most yellows and pastels. This geometry will ruin any and all high-saturation or low-value colors. Reflex blue, chocolate, Coke red, and Pepsi blue wreak havoc with natural image elements. The white has to be a dot pattern composited from the separation files so all colors have the right dot size at the deposit of gelled white. Further, if we match a proof, we need to consider the standard gain that’s built into most physical and digital proofing systems.


The world of printed color has evolved, catering to sheet-fed and web offset printing, which is where the biggest bucks have been. As we are sucked into digital, this offset convention still stands and proofs automatically to compensate for offset gain. Depending on lpi, stock, ink set, and methodology, a 1 percent tone gains to 2 percent, 10 percent to 18 percent, 25 percent to 38 percent, 50 percent to 68 percent, 75 percent to 84 percent, and 95 percent to 98 percent. Conversely, screen printing starts at 6 percent and may print a 10 percent as a 22 percent, a 50 percent at 52 percent, 75 percent at 82 percent, and a 95 percent as a solid. If we make a typical proof and want to match, we need to adjust or else we’ll have too little contrast, washed-out color, and too few subtleties.

This is where color management comes in. The CMS du jour is the refinement of a decade-old approach—adjust the file and film to anticipate the change in printed color based on one’s ability to produce a neutral gray tone scale. GRACol G7 is such a system wherein a fiducial image of a large color gamut is printed and measured. The measured data is used to allow the separations to help match a full tonal range grayscale.

A reasonable proof is a lot of help, but because of the shirt’s black or dark background, simultaneous contrast gives the image a much different appearance. It helps mitigate the contrast if you alter the background color of the proof to approximate the shirt color.

The proof should also consider and contain a color match for the background as well as the quality of white used as the underbase. For example, if the shirt is “black,” it’s likely to have a yellow cast. White ink either has a slight yellow, blue, or gray cast.


Everyone wants to know what densities to run. Here is the simplest answer. Pigments are not pure. Typical magenta, for example, contains tons of yellow and a dab of cyan. Cyan contains a bit of yellow and a little magenta. Yellow may have a spec of magenta or a spec of yellow, but it’s invariably the purest of the bunch.

So M=M+Y+c, C=C+y+m, Y=Y+[m‖c]. At a glance, we can see why yellow density is always lowest; cyan is slightly less than magenta; and magenta has the highest-printed, solid ink density. Specifics are based on pigment type and loading, sequence, and final deposit, but a generic set of optimal secondaries might be M~1.60, C~1.55, Y~1.30 and K~1.80.

To maximize and balance the color gamut with any set of inks, run the magenta at its maximum density with maximum saturation without entering masstone. Then, in sequence, overprint solid magenta and solid yellow so the secondary RED ~ 0 degrees in an HSB model. Overprint solid magenta and solid cyan so the secondary BLU ~ 270 degrees. The GRN will map around 155 degrees. This is errant, but we can’t take three impure primaries and make three accurate secondaries. If your customer is Pepsi or 7UP, you may want to vary the process to prioritize BLU or GRN.

We print white + CMY but we sell WRGBYG (WHT, RED=M+Y, GRN=C+Y, BLU=C+Y, YEL, NNG~.50C+.40M+37Y). Very few finished colors are pure primaries. If the secondaries and the near neutral grays are balanced throughout the tonal range, the colors within the gamut will be visually acceptable.

There’s always an issue of black text and transparent black halftone ink. With a color managed system, we can run the black at a sufficiently high density and put cyan only under the text to get a jet black without compromising the image, print dot and text in the same screen.

Sequence, trapping, and gloss

If the shirt is dark and the press color capacity is limited, it’s likely that we want to run black first, followed by white, and then use a first flash. If color capacity isn’t a problem, we print white, flash, and save black for later in the sequence. If we have two flashes available, we run in sequence: a one-hit white, yellow, cyan, magenta, flash, and black. The white must matte down and be smooth and electron rich. The colors must overprint to maintain our color-accurate secondaries and near neutral gray throughout the tonal range.

The gloss level of each color should diminish based on sequence, or the final image will look posterized when viewed at any angle but perpendicular. Each color should have a minimum surface tension, minimum fluid momentum, and maximum recovery rate.

Other considerations

The mesh count has little to do with the potential lpi. I recommend a thin thread with only sufficient count to be durable on press. For most readers, a 280/35 high modulus yarn PET works exceedingly well up to 85 lpi. If you need up to 130 lpi, use 380/30. On a 23" X 31" frame, the tension should be 25N/cm² at the four corners of the image and 32N/cm² in the center. The off contact should be between 0.125" and 0.175" for rougher or lower fabric mass garments.

The stencil should have an RzS1 garment flatness of less than 5.0m. RzS1 blade-side flatness is below the mesh also and -5.0m. The stencil should not be thicker than 5.0m. As you can see, it has to be thin and flat; therefore, the best stencil is a capillary film with a back coat of compatible emulsion once the film has dried.

Then, there are the enigmatic angles of the chaining direction of the dots, which if the gods of radial relationships don’t favor us that day, we have moiré. There are dozens of patterns of moiré, one of which is radial. If we maintain the dynamic shape of the mesh and dot and frequency, we will not have moiré.

The press must be calibrated so all four planes—carriage, blade edge, mesh, and platens—are all on the same plane. The flood bar should be polymeric, so it can be used to preload the mesh for the white. Otherwise, for the colors, it should have a gap above the mesh equal to the wet film thickness of less than 0.001" and run at maximum stroke speed.

The blades should be biaxial, run at 5 degrees with minimum pressure (~2 BAR or 30 PSI for our tension and off-contact gap), and they must run at maximum stroke speed. Biaxial blades have the unique ability to balance compressive force perpendicular to the mesh and shearing force parallel to the mesh. For the white, we need a 30 percent compressive and 70 percent shearing. The colors will print perfectly with 10 percent compression and 90 percent shearing force. The flood speed is excessive if the flooded layer contains craters, the print speed is excessive when the wet film thickness plateaus, and the edge acuity of the highlight dots diminish.

Before you opt for Photoshop or a cursory plug-in, decide if you’re a simulated artist or CMYK engineer. Pick a process of simulated or WCMYK++, color manage the system, develop a proof, and find or alter the inks to perform as indicated. Use a high-volume mesh with a spec stencil, calibrate the press, set the blades, and reap the benefits.