FAQ · Pantone · ΔE tolerance · ICC profiles · proofing rounds

Colour matching process — what determines the colour on a finished item

The colour on a finished promotional item is the result of a sequence of decisions that begins with the artwork and ends with the operator pressing the print, and every step in that sequence can move the colour away from what the buyer approved on the screen. This page walks through the colour matching process for custom merchandise in Armenia and the wider EAEU: how Pantone spot colours, CMYK process builds and ICC profiles actually behave on each decoration method, what ΔE tolerance means and what number is reasonable per substrate, what file format and what references to send the printer, and how a multi-round proofing workflow fits into a production schedule without becoming a project of its own.

Request a colour match How we structure programmes

Why colour matching is its own discipline in B2B merchandise

The colour on a finished item is the most visible thing on the item and the most commonly disputed thing at goods-in. A graphic designer working on a backlit display will see colour one way; a buyer looking at a printed swatch under office lighting will see it another; a screen printer holding a Pantone book on the press will see a third; a DTG operator looking at a rendered ICC profile will see a fourth; and a recipient looking at the finished product under daylight will see a fifth. The art of matching colour in this category is not a single decision but a long sequence of decisions made under different lighting conditions, against different reference systems, on different substrates, with each step introducing a small distortion that compounds across the line.

Two facts govern the conversation. First, no decoration method reproduces colour exactly: every method has a gamut, and every gamut has edges that the eye can see, which is why "Pantone match" is always a claim about being inside a tolerance band, not a claim of identity. Second, every substrate changes the colour: the cotton's optical whiteness, the polyester's sheen, the ceramic's glaze, the metal's anodised tint, and the leather's dye base all move the colour a measurable amount, which is why "the same Pantone" can render visibly differently on a white cotton tee, a heather cotton tee and a charcoal cotton tee. The matching process exists to make those distortions small enough not to matter for the brand.

The discipline that holds all of this together is named three different ways: colour management, colour fidelity and colour approval. Each name covers part of the work. The buyer-side question is which methods support which tolerances against which substrates, and which workflow actually lands the colour the brand expects on the first production run rather than on the second.

Reference system: Pantone Plus (coated/uncoated), Pantone TPX/TPG, Pantone C, custom library
Tolerance band: ΔE 2000 ≤ 2–3 on coated paper; ≤ 3–4 on textile; ≤ 5 on natural substrates
Standard proof format: AI/PDF/EPS with embedded fonts and Pantone swatches named
Recommended proofs: Digital mock-up + physical strike-off per SKU and per fabric colour
Approval cycle: 1–3 rounds; physical sample at round 1; production release at round 2 or 3
ICC profile family: ISO 12647-2 (offset); FOGRA51 / GRACoL2013; method-specific for DTG

Colour systems at a glance: PMS, CMYK, ICC and the fabric dip

Four colour systems are used in the conversation, and the right answer is usually which system fits the method and the substrate rather than which system is "best". Pantone (the Pantone Matching System, often shortened to PMS, currently issued as the Pantone Plus series with coated and uncoated books) is a spot colour system: each named colour is a specific ink mixed in the factory to a published formula, and the reference is the swatch book itself held against the substrate. CMYK is a process colour system: four standard inks (cyan, magenta, yellow and black) overprinted to build the visible gamut, with the visible range set by the press, the substrate and the profile. An ICC profile is a colour map that describes how a particular device (a monitor, a printer, a textile press, a heat press) renders colour against a reference, so that a digital file can be translated consistently across devices. A fabric dip is a piece of substrate that has been dyed to a Pantone reference or a custom colour, used as the substrate-side reference when the substrate's own colour response is significant.

The systems are not interchangeable. A Pantone spot colour cannot be reproduced exactly inside a CMYK press because CMYK has a smaller gamut than the Pantone library; the printed result is the CMYK approximation that the profile names. An ICC profile cannot make a CMYK file match a Pantone spot ink because the underlying process builds from four inks and the spot ink is a fifteenth ink in a separate run. A fabric dip cannot replace a Pantone book because the fabric's colour response changes with the dye lot, the finishing and the optical brighteners in the substrate. The buyer who is told that a Pantone can be reproduced on DTG with an ICC profile has been told something that is technically plausible and practically wrong: the colour can be approximated within a tolerance band, but the Pantone reference is on paper and the print is on cotton, and the comparison the buyer makes at the goods-in point must be against the right reference for the right substrate.

The practical rule is that the colour system follows the method. Screen print and embroidery use Pantone spot inks and threads because the press loads discrete colours; DTG and UV print use CMYK process builds driven by an ICC profile because the printer builds the colour from the four inks it carries; sublimation uses a CMYK build with a heat-press profile because the colour is driven into the substrate by temperature and dwell time. A buyer who specifies a Pantone for an order that is being sublimated has named the wrong reference; the right reference is a CMYK build target or a previously approved physical print on the same substrate.

The fabric dip is the one element that survives across methods. A fabric dip is a small sample of the production blank (a cotton tee in the production colour, a polyester garment in the production colour) that has been dyed to a named Pantone reference or a custom colour, retained by the buyer and the factory, and used as the comparison reference at every proofing stage. The dip is the document that makes the conversation reproducible: it is the same dip on round one, round two and the production run, and the comparison the buyer makes is dip-to-print rather than screen-to-print or Pantone-book-to-print.

What each decoration method actually prints colour against

The four methods named below are not interchangeable and the colour workflow that fits each is different. Reading the table before the artwork is approved against the supplier saves a proofing round that was predictable from the beginning.

Screen printing — colour separations and PMS spot inks

Screen printing loads each colour as a separate ink pass through a separate mesh, which means the colour is decided at the artwork stage by the separation. A four-colour spot build (PMS 186 red, PMS 286 blue, PMS 116 yellow, black) uses four inks and four screens; a four-colour CMYK process build uses four inks and four screens at finer line counts; a half-tone build on dark garments adds an underbase white and a highlight white, which is technically a six-ink build. The colour the operator sees is the colour of the mixed ink on the substrate; the colour the buyer sees is the colour of the cured ink on the substrate after washing and after the substrate has settled into its finished state. Pantone spot colours are the right reference for spot builds; CMYK builds match to the printed proof at the press, not to a Pantone book.

DTG — ICC profile calibration and substrate-by-substrate rendering

DTG prints from the same CMYK ink set (plus white for dark garments) on every garment, and the colour the buyer sees depends on the ICC profile loaded for the fabric. A profile is built by printing a known target, measuring the printed result with a spectrophotometer, and computing the colour map that translates the digital file to the fabric's response. Cotton, poly-cotton and tri-blend fabrics all need different profiles because their optical whiteness and fibre response is different; a profile built on white cotton will print visibly cooler or warmer on a heather cotton, and a profile built on cotton will print incorrectly on a poly-cotton. The buyer should expect one ICC profile per fabric per supplier, kept on file, and a physical strike-off before the production run.

Sublimation — gamut, transfer paper and heat-press calibration

Sublimation is a CMYK process build printed first onto a transfer paper, then driven into the polyester or poly-blend substrate by heat and pressure in a calibrated heat press. The colour the buyer sees depends on three things: the printer's CMYK build, the substrate's polyester content (sublimation only bonds with polyester; cotton takes no dye and renders washed out), and the heat-press temperature/dwell/pressure profile. White polyester renders colour close to the source; light grey renders the colour underneath; dark fabric renders poorly and is not a viable substrate for sublimation. A Pantone reference on a transfer paper is the right starting point; the comparison reference for the finished product is a physical print on the same fabric.

Embroidery — thread match and Pantone-to-thread conversion

Embroidery does not print colour as such; it lays thread, and the colour is the colour of the thread. There is no Pantone ink in embroidery because the design is stitched, not printed; the Pantone reference is converted to a thread reference using the thread manufacturer's colour library (Coats, Madeira, Sulky, Robison-Anton), and the closest match is named by thread code rather than by Pantone number. A five-colour logo on screen print is straightforward in CMYK process or five spot inks; the same logo in embroidery is five thread cones, five needles, a thread change every few metres, and potentially a digitising decision about how the colour overlap is rendered (a true mesh, a digitiser's trick with a darker or lighter thread to simulate the mesh, or a stitch order that changes the perceived mix).

What to send the printer — file format, fonts and embedded references

The artwork the buyer sends is the document the printer works from, and the format decisions in that file determine whether the colour can be matched at all. Three rules cover most of the cases. First, send a vector file wherever the artwork is a logo or typography: Adobe Illustrator (.ai) is the standard, but a PDF with the vector preserved or an EPS with the vector preserved is acceptable. Raster files (PNG, JPG) are acceptable only when the artwork is a photograph or a complex gradient that genuinely cannot be vectorised; a logo delivered as a raster is a logo that the printer will recreate or that the buyer will accept as best-effort.

Second, embed the fonts. A vector logo with text set in a typeface that is not converted to outlines or not supplied with the file will substitute at the print stage and will arrive with the wrong letterforms, which is the single most common avoidable error in artwork submission. The two acceptable solutions are to convert the text to outlines (a destructive operation that locks the type, so the buyer can no longer edit the text) or to supply the typeface file alongside the artwork file. Designers routinely convert to outlines for print because the conversion removes the editorial question from the printer's hands; buyers who want to retain editability should supply the font file and ask the printer to confirm receipt.

Third, name the colours in the file. A vector file that uses Pantone spot colours should have each Pantone reference named in the swatch panel and applied to the matching geometry; a vector file that uses CMYK process should have the CMYK build values visible and ideally the ICC profile embedded. Naming the colour is what lets the printer decide between loading a spot ink and building the colour in CMYK; an artwork with unnamed colour or with generic "red" controls is an artwork that the printer will interpret, and the interpretation will not match the brand without a physical proof.

The standard package sent to the printer combines: the source file in its editable format (AI preferred, or PDF with vectors intact); the artwork as a flattened PDF at the intended print size for review; the colour reference — a Pantone swatch naming each spot colour or a profile/build sheet for CMYK; the substrate specification (the blank good in its production colour); a vector or raster of any reference imagery with the colour as the buyer sees it on screen. The production run starts from the same file the proofing round started from, which is why naming the colours at the artwork stage is what links the proofing round to the production result.

Proofing rounds and the approval workflow

The proofing round exists because the colour on the finished item cannot be confirmed from the artwork file alone, and the number of rounds required to reach an acceptable match is the visible cost of the colour discipline a buyer has or has not put in place. A clean workflow has three rounds. Round one is a digital mock-up: a high-resolution render of the artwork on the substrate, generated by the factory's pre-press team using the ICC profile for the fabric and the Pantone references for spot colours. Round two is a physical strike-off: the actual production equipment on the actual production blank, with the actual ink or thread, cured and finished. Round three, where it is needed, is a corrected physical strike-off after revisions to the artwork, the colour or the fabric.

The first round is fast and cheap; the second round is the one that costs real money in materials and machine time; the third round is the one that indicates either an ambiguous brief or a material change between round two and the production run. A buyer who reaches round four on the same artwork has either named the wrong Pantone for the substrate, failed to specify a fabric colour change, or is approving round three against a different reference than was agreed at round one — every one of these is a fixable process error rather than a property of the decoration method.

The approval workflow that closes the round early is built around three checkpoints. The first checkpoint is a written acknowledgement that the digital mock-up has been reviewed against the named Pantone or profile reference, with the buyer stating approval or revisions; this is the document the printer uses to start round two. The second checkpoint is a physical sample at round two, with a sign-off that names the colour reference used at the comparison; without a named reference, the sign-off is anecdotal. The third checkpoint is a corrected physical sample at round three where applicable, with a sign-off that explicitly authorises the production run against that sample.

The signed sample retained against the production run is the document the goods-in inspection works from. The person inspecting the delivery should have the same physical sample, the same Pantone book or profile, and the same lighting (a D65 daylight booth is the standard for textile, a D50 booth for printed paper) that the buyer used at round two. A sign-off made under office lighting will not reproduce under D65 and vice versa, which is why the lighting at approval is the lighting at goods-in, or the comparison fails before it starts.

ΔE tolerance and reading a colour approval

ΔE (the E stands for Empfindung, the German word for perception) is a number that summarises the difference between two colours as the human eye would perceive it. The current standard, ΔE 2000 (CIEDE2000), corrects for the perceptual distortions in earlier formulas and is the metric most colour-managed workflows use. A ΔE of 1.0 is the threshold at which a trained eye under controlled lighting begins to see a difference between two colours side by side; a ΔE of 2.0 to 3.0 is the threshold at which an untrained eye under daylight begins to see a difference; a ΔE above 5.0 is a difference most viewers will call "different colour" rather than "slight variation".

Tolerance bands are stated per substrate because the substrate changes the colour. On coated paper, a ΔE of 2.0 to 3.0 against a Pantone reference is the working tolerance for a screen print or offset press; on uncoated paper the same Pantone is harder to hit and a ΔE of 3.0 to 4.0 is the working tolerance. On a bright white cotton, a ΔE of 3.0 to 4.0 against a Pantone book is reasonable; on a heather, a charcoal or a dyed substrate, the substrate's colour shows through the ink and a ΔE of 4.0 to 5.0 is the realistic tolerance. On natural leather, raw wood or unglazed ceramic, a ΔE above 5.0 is normal because the substrate's response is variable lot to lot, and the colour conversation moves from a Pantone book to a previously approved physical sample on a previously approved fabric lot.

A colour approval that names only the Pantone reference is a colour approval that does not name the tolerance; a colour approval that names the Pantone reference and the ΔE band, against a named substrate, is a colour approval that can be enforced at goods-in. The supplier should be able to state the tolerance band the production line is working against; the buyer should be able to state the tolerance band they expect; and the comparison reference at goods-in should be the same physical sample that was approved at the last proofing round, under the same lighting booth the buyer used at approval.

The acceptable answer to "what ΔE should we expect?" is a number between 2.0 and 4.0 for most textile and apparel orders, with the higher end of that range for dark or heather substrates and the lower end for white substrates. Numbers outside that range are usually visible to the recipient and should be the subject of a sign-off note rather than a silent acceptance. Numbers above that range on a claimed Pantone match usually indicate that the method or the substrate is not appropriate for that Pantone, and that is a conversation worth having before production rather than after delivery.

Special cases — fabric colour shifts, neon and metallic inks, post-print treatments

Five cases break the standard workflow and are worth flagging at the brief stage rather than discovering at goods-in. The first is fabric colour shifts between lots. Cotton and poly-cotton blanks are dyed in lots, and the dye lot varies in optical whiteness by a ΔE that the buyer can see between lots even when both are nominally "white". A buyer who orders across multiple production runs against nominally the same blank goods should request a current-lot swatch before the production release of each run, with the physical sample held as the comparison reference.

The second is colour changes on dark vs light fabric. A Pantone 186 red on white cotton is a screen print with a single colour pass on a white substrate; the same Pantone on a charcoal cotton requires an underbase white, the spot red, and a flash cure between layers, and the perceived red is duller because the substrate's grey absorbs some of the red's reflectance. The right comparison reference on a dark garment is a strike-off on the production fabric colour, not a Pantone book. A Pantone book on white paper under daylight is the worst reference for a comparison against a printed red on charcoal fabric.

The third is neon and fluorescent inks, which sit outside the Pantone library. The Pantone C and Pantone Plus libraries describe coated and uncoated spot inks; fluorescent yellows, oranges, pinks and greens are sold by the ink manufacturers as their own colourways (MagnaPrint fluorescent, Rutland fluorescent, etc.) and require a custom strike-off rather than a Pantone reference. The colour can be specified by reference to the ink manufacturer's sample card plus a ΔE band against an approved physical sample.

The fourth is metallic inks (gold, silver, copper, bronze), which are real-metal inks printed as a separate colour pass and shift visibly with viewing angle. The buyer should approve a physical strike-off at the same time of day and on the same lighting the recipient will see the item under, because metallic inks change with light direction and what looks right under a striplight at noon may look dull under office lighting at four in the afternoon.

The fifth is post-print treatments that move the colour. A garment-washed tee is the canonical example: the wash softens the print slightly and shifts the colour by a measurable ΔE. A foil-stamped logo is moved by the temperature and pressure of the stamping machine. A UV-varnished print changes the gloss profile of the printed area. Each of these is a separate specification in its own right, and the buyer should approve the post-print strike-off against the post-print reference, not against the Pantone book that was used for the pre-treatment artwork.

Questions buyers ask about colour matching

Can a Pantone spot colour be reproduced exactly on DTG or sublimation?

No, not exactly. Pantone spot colours are formulated as specific inks mixed in the factory, and DTG and sublimation are CMYK process methods that build colour from four inks. The achievable result is an approximation inside a tolerance band, typically a ΔE 2000 of 3 to 4 on bright-white cotton for DTG and 4 to 5 on light-grey polyester for sublimation, with wider tolerances on darker substrates. The right reference for a DTG or sublimation order is a CMYK build target or a previously approved physical sample on the same substrate, not a Pantone book held against the print.

How many proofing rounds should a reasonable order require?

One to three rounds for a clean order. Round one is a digital mock-up; round two is a physical strike-off; round three is a corrected physical strike-off where revisions are needed. More than three rounds usually indicates a brief error (wrong Pantone for the substrate, fabric colour not specified, reference light changed between rounds) rather than a property of the decoration method. A buyer who can name the Pantone references, the substrate colour, the substrate lot and the lighting at approval will typically reach sign-off in two rounds.

What is the right file format to send for a screen print?

An Adobe Illustrator file (.ai) is the standard. A PDF with vectors preserved or an EPS with vectors preserved is acceptable. Raster files (PNG, JPG) are acceptable only when the artwork is a photograph or a complex gradient that genuinely cannot be vectorised. Each Pantone reference should be named in the swatch panel; fonts should be either converted to outlines or supplied alongside the file; the artwork should be at the intended print size. The production run starts from the same file the proofing round started from.

Is Corp-Merch affiliated with, endorsed by, or an official supplier to the brands, software vendors, certification bodies, or trade associations referenced on this page?

No. Corp-Merch is an independent B2B merchandise supplier. References to Pantone LLC, Adobe Inc., the International Organization for Standardization (ISO 12647-2 colour management), the Fédération Internationale des Arts Graphiques standards (FOGRA), the Coats thread library, Madeira, Sulky, Robison-Anton, ink manufacturers and any brand, software vendor, certification body or trade association are informational only and imply no partnership, sponsorship, endorsement or official-supplier relationship. Quantities, tolerances and lead times describe typical commercial practice for this category rather than a fixed quotation.

This page is written for buyers, brand managers and production teams arranging custom merchandise in Armenia, the wider EAEU and international markets. Corp-Merch is an independent B2B merchandise supplier. Pantone references, ICC profiles, ΔE tolerances, file formats and proofing rounds describe typical commercial practice in the decoration industry; they are not a fixed specification and each order is set up against a written brief and a signed proofing round. Brand names, method names and certification references are used factually to identify the technology and standards, not to suggest any commercial relationship with their owners.