sRGB, CMYK, and which one your photos need.

A color space is a set of instructions.

A photo file is a grid of numbers. On its own, a number like 30, 80, 200 does not name a color. It names a recipe: this much red, this much green, this much blue. The color space is the part that says how strong this much is, and which exact red, green and blue are meant. Feed the same numbers to two devices working from different instructions and you get two different colors, both of them behaving correctly.

That is why a good file carries a profile, a small tag riding inside it that names the space the numbers belong to. With the tag attached the file is unambiguous. Without it the numbers are a guess, and the guess almost every browser makes is sRGB. A surprising share of color disasters are not broken files at all. They are correct files whose tag went missing on the way, being read in the wrong language.

Color space belongs to the same family of quiet decisions as depth of field: invisible when it is right, and the first thing anyone notices when it is wrong.

Screens add light. Paper takes it away.

Your screen is a dense field of tiny emitters in red, green and blue. Every color you have ever seen on it was made by lighting those three in some proportion. All three at full strength give you white. None of them lit gives you black. This is additive color: you begin in darkness and add light, and the screen manufactures every bit of that light itself.

A print starts at the opposite end. The paper is already white, or as near white as paper gets, and all it does is bounce back the light in your room. Every drop of ink removes something from what comes back. Cyan absorbs red light, magenta absorbs green, yellow absorbs blue. You begin with everything and subtract. Lay all three down together and in theory you get black; in practice you get a heavy brown mess, which is why there is a fourth ink, black, the K in CMYK.

Everything else follows from that. A screen can make a color brighter than the room it sits in. A print can only hand back part of the light falling on it.

Why ink cannot reach the blue on your screen.

Every device can produce only so many colors, and the full set it can produce is called its gamut. Picture it as a shape rather than a list. The screen's shape and the ink's shape overlap generously through the middle, where most of the colors in a portrait live, and part company at the edges, where the saturated colors are.

Screen blue is the clearest case. A display emits blue directly, as a narrow band of intense light. To make blue on paper you lay cyan and magenta over each other and keep whatever the two together fail to absorb, and what comes back is always calmer than the emitted version. The same holds for electric greens, hot oranges and saturated violets. Ink is not defective; it is a different instrument with a smaller range at the top.

This matters far less for a face than people fear. Skin, hair, gray and white backdrops, navy and charcoal clothing all sit well inside both shapes, which is why a good portrait converts to ink honestly. What shifts is the saturated material around you: a brand blue, a strong red garment, a backdrop pushed hard. If your logo carries an electric blue, it will read quieter on paper than on your monitor, and no file you send will change that.

When a color cannot be printed, the conversion moves it to the nearest one that can be. If several neighboring shades land on the same neighbor, the difference between them vanishes and a smooth gradient turns into visible steps. A careful conversion compresses the whole range gently instead of clipping, which is one more reason to let the press do it.

sRGB is the common denominator.

sRGB was defined as the color an ordinary screen could be relied on to show, and it stuck. Browsers assume it, phones assume it, social platforms assume it, email assumes it. That makes it the safe currency of the internet rather than the richest one, which is exactly right for a picture that must look like itself in places you will never inspect. Anything bound for a screen should leave as sRGB, in an eight bit JPEG, with the profile attached.

Wider working spaces exist for editing, and they describe colors sRGB has no numbers for. They earn their keep inside the edit and become a liability the moment the file leaves. A wide gamut file that loses its profile is read as sRGB, every color is interpreted as a tamer version of itself, and the picture arrives flat and faintly gray. That is the usual cause of a photo looking washed out on a website when it looked fine on the computer.

Color space and resolution get confused constantly, so separate them once: the space decides what the numbers mean, and the pixel count decides how large the picture can be shown before it falls apart. How much resolution a given use needs is its own guide, and neither question answers the other.

CMYK is not one space. It is a press and a paper.

sRGB is a single defined standard, which is what makes it dependable. CMYK is not one thing at all. Its numbers describe how much of each ink goes down, and what that produces depends on the press, the ink set, the paper and how thirstily the paper drinks. Coated stock holds ink near the surface and keeps color crisp. Uncoated paper absorbs it, every dot spreads slightly, and the image lands darker and softer.

Which is why converting to CMYK yourself, to be helpful, is usually the least helpful thing you can do. Unless someone has handed you the exact profile for the press and paper the job is running on, you are converting against a guess, and in the process you have permanently discarded every color that guess could not hold. Give the printer a good RGB file and let the people running the machine convert against their own profile. The exception is a written specification: then follow it exactly, down to the profile name and the format, and ask rather than improvise if a line of it is unclear.

The trip is also not a round one. Once colors have been squeezed into what ink can do, converting back does not bring them home. Keep your original as the master and treat every conversion as a copy made for one job.

What actually goes wrong, and how to recognize it.

Four failures cover almost everything, and each has a signature you can learn to spot in a second.

• A CMYK file uploaded to a website. Some systems refuse it, some convert it on the fly with no care at all, and you get a muted version with the life gone out of the darks.
• A wide gamut file with the profile stripped. Everything looks desaturated and slightly gray, skin worst of all, because saturated numbers are being read as though they meant something milder.
• An sRGB file sent to a commercial printer with nothing said about it. Usually fine, since the printer converts it. Occasionally it meets an automatic conversion built for something else and the deep colors block up into flat shapes.
• A screen capture of a photo, sent in place of the photo. It carries whatever the screen was doing at that moment and a fraction of the original pixels.

Learn two signatures and you can diagnose most of it from across the room. Dull and gray is a profile problem, and the fix is in the file. Dark and heavy is a print problem, and the fix is elsewhere: why a screen and a print disagree even when the file is correct is a separate question with separate answers.

The questions to ask before anything prints.

You do not need to become a color engineer. You need four questions, asked once, of whoever is doing the printing.

• What color profile do you want, and would you rather convert the file yourself?
• What paper is this running on, coated or uncoated?
• What format and pixel dimensions do you want, and should it be flattened?
• Will I see a proof, and will it be printed on the same paper as the job?

Ask about shape while you are there, because a page rarely matches the proportions a camera produces. Print sizes and cropping covers how much of you disappears when a frame changes shape. For a piece you are hanging rather than publishing, the surface moves color as much as the ink does, which is where canvas, metal and paper stop being interchangeable.

For portraiture the working answer stays reassuringly boring. A finished file, retouched by hand and delivered in sRGB, is the normal currency of the craft, because it behaves everywhere a headshot gets used. Hand retouching and a private gallery within the week are part of what the investment covers, and albums, fine prints and gallery canvases can be purchased as well. When color is the last item on a list that began with needing a new picture, the complete guide to professional headshots covers the rest.