RGB and CMYK differ because a screen makes colour by adding red, green and blue light to black, while a press makes colour by laying cyan, magenta, yellow and black ink on white paper that subtracts light. The two systems cover different sets of colours: in one published comparison, sRGB reproduces 39.1% of a reference colour atlas while standard coated offset (FOGRA39) reproduces 26.9%, so bright screen blues, greens and oranges must be shifted to the nearest printable colour. That shift, made by an ICC profile conversion, is the colour change you see on paper.
- 39.1% vs 26.9%: share of the HLC colour atlas reproduced by sRGB and by coated FOGRA39 offset.
- 17.0%: uncoated paper's share once its own paper colour is counted.
- 350% total ink is the hard ceiling for coated sheet-fed work under ISO 12647-2 (300% uncoated).
- 100% K for text, a rich black such as 60/40/40/100 only for large solids.
- Design in RGB, convert at the end with the printer's own ICC profile.
The answer "design in CMYK for print" is correct and useless on its own, because it does not tell you which CMYK. A coated sheet-fed job, a newspaper and an uncoated letterhead each have their own gamut, ink limit and profile.
This post puts numbers on all three, using the press standard (ISO 12647-2) that Indian commercial printers quote when they say they print to "FOGRA" or "ISO coated".
What is the difference between RGB and CMYK?
RGB is additive: each pixel emits red, green and blue light, and full output of all three reads as white. CMYK is subtractive: each ink absorbs one band of light from white paper, and black ink adds depth and saves ink in shadows.
- Cyan absorbs red.
- Magenta absorbs green.
- Yellow absorbs blue.
The idea that three inks can make every colour was first sold as a business in the 1710s. In Amsterdam around 1710, the painter and engraver Jacob Christoph Le Blon made colour prints from three copper mezzotint plates, inked yellow, red and blue and printed one after another on the same sheet. In 1719 George I granted him patents for the process in London, where he set up The Picture Office to sell copies of famous paintings.
It failed as a business in 1725, the year he published Coloritto, arguing that three primaries could produce all visible colours. In practice his later prints often needed extra black and white plates. Three centuries on, the press still adds a K plate for the same reason.
Wikipedia: Jacob Christoph Le BlonWhy the K plate exists
The K exists for physical reasons, not tradition. Real cyan, magenta and yellow pigments are impure, so 100/100/100:
- prints a muddy brown-black;
- soaks the sheet with 300% ink;
- makes registration of fine text impossible.
A single black plate prints crisp text and lets the separation software swap equal parts of C, M and Y for K in neutral areas (grey component replacement), which cuts ink and drying problems.
Light, white point and viewing
A screen also has a white point and a brightness that paper does not. The original sRGB proposal (below) assumed a display of 80 cd/m² with a D65 white; a print is lit by whatever lamp is overhead and is judged under D50 in pressrooms.
That mismatch in viewing conditions explains part of the "it looked brighter on screen" complaint before gamut even comes into it.
The proposal that became IEC 61966-2-1 (1999). It fixed the primaries at the ITU-R BT.709 values (red x 0.640, y 0.330; green 0.300, 0.600; blue 0.150, 0.060), a D65 white point (x 0.3127, y 0.3290), a 80 cd/m² display, a 64 lux encoding environment and a transfer curve close to gamma 2.2, based on HP's measurement of an average CRT gamma of 2.2 (standard deviation about 0.2). The point was to give untagged web images a default meaning without embedding ICC profiles. The limit: it was built around 1990s CRTs, which is why its gamut is small next to modern phone panels and why "sRGB" files look duller on a P3 screen that ignores colour management.
Read the document (free)Why do bright colours look duller in print?
Because they sit outside the press's gamut, the 3D volume of colours its inks can make on that paper. Saturated sRGB blues, cyan-greens and vivid oranges have no matching ink mixture, so the conversion moves them to the nearest printable colour, which is always less saturated, darker or both.
The clearest public numbers I found come from the HLC Colour Atlas comparisons published by the German freie Farbe project. They count what share of the atlas's systematic colour samples each output can reproduce. Two columns matter: relative colorimetric (paper white is treated as white) and absolute colorimetric (the paper's own colour is kept, as a proof would show it).
See the data as a table
| Relative colorimetric (%) | Absolute colorimetric (%) | |
|---|---|---|
| sRGB screen | 39.1 | 39.1 |
| Inkjet proof (Epson SC-P7000) | 44.2 | 40.4 |
| PSO Coated v3 (FOGRA51) | 39.0 | 27.7 |
| Coated FOGRA39 | 34.4 | 26.9 |
| ISO Coated v2 | 37.3 | 27.2 |
| PSO Uncoated (FOGRA52) | 37.2 | 17.0 |
Source: freie Farbe e.V. (Holger Everding): Gamut comparisons with the HLC Colour Atlas · Percentages of HLC atlas coordinates. Absolute colorimetric keeps the paper colour, which is why uncoated paper drops most.
Three things stand out:
- Coated offset loses about a third of sRGB's colours on the absolute measure (26.9% against 39.1%).
- Uncoated paper is much worse once its own greyish, ink-absorbing surface is taken into account (17.0%).
- A good inkjet proofer actually beats sRGB, which is why a proof can show colours the final press cannot hit if the proof is not set to simulate the press profile.
Neither gamut contains the other, so you always need a defined conversion between them.This post, after the freie Farbe gamut comparisons
The gamut is not just smaller; it is a different shape. Screens reach further into saturated blue and green; CMYK reaches further into some cyans and deep yellows that sRGB cannot show. That is the answer to the Stack Exchange question "RGB and CMYK: why both?": neither contains the other, so you need a defined conversion between them.

How big is the screen gamut now?
Bigger than sRGB on most good phones, which widens the gap with print. On the CIE 1931 diagram, Rec. 709 (the same primaries as sRGB) covers 35.9% of the visible chromaticities, Adobe RGB 52.1%, DCI-P3 53.6% and Rec. 2020 75.8%, according to the figures Wikipedia attributes to a 2012 TV Technology article.
| Space | Where you meet it | Share of CIE 1931 diagram | White point |
|---|---|---|---|
| sRGB / Rec. 709 | Web default, most monitors, JPEGs | 35.9% | D65 |
| Adobe RGB (CSS a98-rgb) | Photo editing for print | 52.1% | D65 |
| DCI-P3 / Display P3 | Recent iPhones and many Android flagships, cinema | 53.6% (DCI-P3) | D65 for Display P3 |
| Rec. 2020 | HDR video mastering target | 75.8% | D65 |
| Coated CMYK (FOGRA39) | Sheet-fed offset on coated paper | Not a triangle; varies with lightness | Paper white, viewed at D50 |
CSS Color Module Level 4 now lets web designers ask for these spaces directly with color(display-p3 …), color(a98-rgb …) and color(rec2020 …), plus perceptual spaces like OKLCH. A brand colour picked in P3 on a phone can be well outside anything a press can print. If you work in OKLCH, the companion post on OKLCH and perceptual colour shows how to keep chroma inside a target gamut.
What do ISO 12647-2 and FOGRA mean when a printer quotes them?
ISO 12647-2 is the international standard for offset lithography process control; FOGRA is the German research institute whose characterisation data sets (FOGRA39, FOGRA51, FOGRA52 and others) describe what a press running to that standard should produce. When an Indian printer says "we print ISO coated", they mean their press is calibrated to one of these data sets, and you should convert your files with the matching ICC profile.
The 2013 edition of the standard sets limits that directly affect your artwork:
| Parameter (ISO 12647-2:2013) | Coated paper | Other (e.g. uncoated) paper |
|---|---|---|
| Total ink (tone value sum), sheet-fed | Should be under 330%, must not exceed 350% | Must not exceed 300% |
| Total ink, heat-set web | Must not exceed 300% | Must not exceed 270% |
| Screen frequency (periodic screens) | 120 to 200 lpi | 120 to 175 lpi |
| Screen angles, round dots | C, M and K 30° apart, yellow 15° from a neighbour, dominant colour at 45° | |
| Printing sequence | Black, cyan, magenta, yellow | |
The standard's own note gives the reason for the ink ceiling: above it, ink traps badly, transfers back and sets off onto the next sheet because it cannot dry. A design that converts to 380% in the shadows will either be clipped by the printer's profile or smudge on press.
Use this to check what you just read. Enter the CMYK values from the darkest area of your file (from the Info panel in Photoshop or the Separations Preview in InDesign) and compare the total with the ISO 12647-2:2013 limits.
Total ink coverage check (ISO 12647-2:2013)
The authors printed test charts on coated and uncoated stocks with three production digital presses (Canon imagePRESS C750, HP Indigo WS 6800, Xerox Versant 180), built ICC profiles and compared gamut volumes in CIELAB against the FOGRA reference sets. With ColorThink software, FOGRA39 (coated) measured 402,279 cubic CIELAB units against 181,382 for FOGRA29 (uncoated), so coated paper offers about 2.2 times the colour volume. The digital presses beat the FOGRA references in some regions and fell short in others, and the uncoated stocks gave the smallest gamuts on every machine. The limit: two gamut-volume programs gave noticeably different totals for the same profiles, so compare volumes only within one tool.
Read the paper (free)
What kind of black should I use for CMYK print?
Use 100% K alone for body text and thin lines, and a rich black (100% K plus some cyan, magenta and yellow) only for large solid areas. Plain K over a big area prints as a dark charcoal, because process black ink is thin and translucent; a rich black such as 60/40/40/100 or 40/30/30/100 looks deep and stays under the ink limit.
The highest-voted answer on the Graphic Design Stack Exchange thread about black gives recipes in that range and warns that 0/0/0/100 is a dark grey over large areas. The arithmetic matters: 60/60/40/100 totals 260%, comfortably under the 300% limit for uncoated sheet-fed work. A "registration black" of 100/100/100/100 totals 400%, breaks every press limit in the table above and will misregister on small type, which is why it should never appear in artwork.
How does an ICC profile convert RGB to CMYK?
Through a device-independent middle step. The source profile maps your RGB numbers to CIE colour (Lab or XYZ); the destination profile maps that colour to the CMYK amounts this press and paper need to reproduce it. The rendering intent decides what happens to colours the press cannot reach.
- Assign the source. Tag the file with what it really is (sRGB, Adobe RGB, Display P3). An untagged file is treated as sRGB on the web.
- Choose the destination. Ask the printer which profile: FOGRA39 or FOGRA51 for coated sheet-fed, FOGRA52 for uncoated, or their own house profile.
- Pick an intent. Relative colorimetric keeps in-gamut colours exact and clips the rest to the edge; perceptual compresses the whole gamut so gradients survive. Photos usually look better perceptual; logos and flat brand colours relative colorimetric.
- Soft-proof. Preview the destination profile on screen with "simulate paper colour" on. This is the closest a monitor gets to the absolute-colorimetric numbers in the chart above.
- Check the totals. Read the ink sum in the darkest areas and make sure it respects the printer's limit.
A brand colour that matters more than any photo is a different problem: if it falls outside CMYK, you may need a spot ink. I compare the cost and the alternatives in is a Pantone spot colour worth it, and the file-format side, vector against raster for printing logos, in vector vs raster logos for print.
Should I design in RGB or CMYK?
Design in RGB and convert to the printer's CMYK profile at the end, unless the job is print-only with heavy black type. Working in a wide RGB space keeps your options open for screen, social and print versions, and modern layout software converts at export with the right profile.
Work directly in CMYK when the piece is a print-only layout with fine black text, when the printer gives you a specific profile and ink limit, or when you need exact CMYK builds for brand colours that the brand book specifies. Whatever you choose, never let the printer convert an untagged RGB file with an unknown default; that is where the worst surprises come from.
What's new
- The CSS color() function for wide-gamut colour becomes Baseline "widely available" across major browsers web-features (W3C WebDX), Nov 2025
- Safari 18.4 adds Display P3 and alpha options to the HTML colour picker WebKit blog, May 2025
- CSS Color Module Level 4 published as a new Candidate Recommendation Draft W3C, Nov 2026
Questions people ask
Why does my printed photo look darker than on my phone?
Three reasons stack up. The phone screen emits light at high brightness while paper only reflects room light; the phone may show a wider P3 gamut than your file was converted from; and the press adds dot gain, where ink spreads into paper and darkens mid-tones. Lower screen brightness while editing and soft-proof with the printer's profile.
Can I convert a logo from RGB to CMYK without losing colour?
Only if the colour is inside the CMYK gamut. Many reds, oranges and dark greys are; saturated blues, greens and purples usually are not. For out-of-gamut brand colours, define a separate CMYK build that matches visually, or use a spot ink.
Is CMYK the same on every printer?
No. CMYK numbers are instructions, not colours. 40/0/100/0 prints differently on coated sheet-fed, uncoated paper and newsprint. That is why profiles such as FOGRA39 and FOGRA52 exist, and why you convert for one specific printing condition.
What is total ink coverage?
The sum of C, M, Y and K percentages at any point. ISO 12647-2:2013 caps it at 350% for coated sheet-fed work (with a target under 330%), 300% for heat-set web and uncoated sheet-fed, and 270% for uncoated heat-set web.
Sources
- ISO 12647-2:2013 Graphic technology: Process control for offset lithographic processes (official preview) (2013; source for ink limits, screen angles, screen frequency, print sequence)
- ISO: ISO 12647-2:2013 catalogue page
- freie Farbe e.V.: Gamut comparisons with the HLC Colour Atlas (source for the chart percentages)
- Spiridonov et al.: Study of the colour gamut volumes of electrophotographical digital printing systems, GRID 2022 (2022)
- Stokes, Anderson, Chandrasekar and Motta: A Standard Default Color Space for the Internet, sRGB (1996)
- W3C: CSS Color Module Level 4 (2026 draft)
- Wikipedia: Rec. 2020 (CIE 1931 coverage figures)
- Wikipedia: Jacob Christoph Le Blon (three-colour mezzotint printing, 1710 to 1725; source for the story)
- International Color Consortium
- Graphic Design Stack Exchange: What is the difference between CMYK and RGB?
- Graphic Design Stack Exchange: RGB and CMYK: why both?
- Graphic Design Stack Exchange: What kind of black should I use when designing for CMYK print? (rich black recipes)
- web-features explorer: color() function (2025)
- WebKit: Add wide gamut P3 and alpha transparency to your color picker in HTML (2025)
