A color space is the set of rules that turns the numbers in a photo into real colors. Every pixel is stored as three numbers for red, green and blue, and those numbers mean nothing until something says which red, which green and which blue they refer to. That something is the color space, and the color profile is the small label inside the file that names it. Get the label wrong or lose it, and the same photo looks dull, garish or simply different on the next screen it reaches.
This page is about that idea: numbers, the label that gives them meaning, and what happens when you change one without the other. It also sets the spaces photographers meet most, sRGB, Display P3, Adobe RGB and ProPhoto RGB, side by side, and says which to use for each job.
Numbers need a label: the idea behind every color space
Think of a pixel value like a distance written as “200”. Two hundred what? Metres and feet give very different answers. A color space is the unit. It fixes three things:
- The primaries. Exactly which red, green and blue a value of 255 means. A space with more saturated primaries can describe more saturated colors.
- The white point. Which color counts as white when all three channels are equal. The common photo and web spaces use D65, a standard daylight white; ProPhoto RGB uses D50, a slightly warmer white.
- The tone curve (often called gamma). How the numbers between 0 and 255 are spread from black to full brightness.
A color profile (an ICC profile) is a file, or a block of data embedded in a photo, that writes those rules down in a form software can use. It maps the device’s numbers to a device independent description of color, CIE XYZ or CIE Lab, which describes colors by how people see them. Color management software goes through that shared description every time it moves a color from one space to another: from your file to your screen, from your editor to a printer.
So “space” and “profile” are related but not the same. The space is the set of colors and the rules. The profile is the label that carries those rules with the file. A photo can be in Adobe RGB and still lose its label, and then nothing downstream knows it is Adobe RGB.
RGB versus sRGB
RGB is the general method: describe every color as a mix of red, green and blue. sRGB is one specific color space built on that method, with its own primaries, D65 white point and tone curve. Adobe RGB, Display P3 and ProPhoto RGB are other RGB spaces. “Is this file RGB or sRGB?” is like asking whether a measurement is in units or in metres: an RGB file only has a definite color once it names which RGB space it is in.
Same numbers, different colors
Here is one pixel, red 0, green 200, blue 180, read under four different labels. The Lab values show the color each label produces: L is lightness, and a* is the green to magenta axis, where a more negative number is a stronger green.
| Label on the file | Lab (L, a*, b*) | What you would see | Can sRGB show it? |
|---|---|---|---|
| sRGB | 72, -47, -2 | A clear teal | Yes |
| Display P3 | 72, -63, -4 | A noticeably stronger teal | No |
| Adobe RGB | 69, -76, -8 | A much more vivid blue green | No |
| ProPhoto RGB | 74, -148, -8 | A green beyond any screen or print | No |
Nothing in the file changed except the label, and the color moved from an ordinary teal to one no device can reproduce. That is why a profile matters, and why the next two sections matter more than any choice of space.
Gamut: how many colors a space can hold
The range of colors a space can describe is its gamut. On the standard CIE chart below, the horseshoe holds every color of light a person can see, and each space is a triangle whose corners are its three primaries. Any color inside the triangle can be written in that space with numbers between 0 and 255; anything outside it cannot.
The curved edge of the horseshoe is light of a single wavelength, the pure colors a prism or a rainbow splits sunlight into. Most of that edge lies outside sRGB, Display P3 and Adobe RGB, so none of them can hold most of the purest spectral colors.
Measured on this chart, from each space’s published primaries:
| Space | Area on the chart, compared with sRGB | Where it reaches beyond sRGB | Typical use |
|---|---|---|---|
| sRGB | 1.00 | Nowhere; it is the baseline | The web, most screens, anything with no label |
| Display P3 | 1.36 | Reds, oranges and greens | Many phones, tablets and laptop screens |
| Adobe RGB | 1.35 | Cyans and greens | Wide gamut monitors, some print workflows |
| ProPhoto RGB | 2.47 | Almost everywhere, and past the visible | Editing raw files in 16-bit |
Two cautions about those numbers. The chart stretches the greens and squeezes the reds and blues, so areas on it overstate how different the spaces look; treat them as a ranking, not a measure of how much more color you get. And Display P3 and Adobe RGB cover about the same area but a different shape, so neither contains the other.
Gamut is not bit depth
Gamut is how far the range reaches. Bit depth is how many steps fill that range: 256 per channel in an 8-bit file, 65,536 in a 16-bit file. A wider space spreads the same number of steps over a bigger range, so each step is a larger jump in color. Computed from the definitions, one step at the top of the green channel moves the color nearly twice as far along the green axis in ProPhoto RGB as in sRGB. In 8-bit that coarseness can show up as banding in smooth skies once you edit hard, which is why ProPhoto RGB belongs in 16-bit files.
Seeing stripes in a blue sky after editing? Bit depth in photography explains the steps between tones and when 8-bit runs out.
Assign versus convert: the two ways to change a profile
Editing software offers two commands that sound alike and do opposite things. Mixing them up is one of the easiest ways to ruin a photo’s color.
- Assign a profile keeps the numbers and changes the label. The colors change. This is what the first table showed.
- Convert to a profile keeps the colors and changes the numbers. The software works out which numbers in the new space make the same color.
Converting from a small space to a larger one, sRGB to Adobe RGB for example, adds nothing. The colors were already inside sRGB, and they stay exactly where they were with new numbers. Color the file never held cannot be created by moving it into a bigger space, which is why the wide space has to come first, at the raw stage, and the small one last.
Here is the sRGB teal from earlier, converted into each space. The numbers change; the color you see does not.
| Converted to | New numbers (R, G, B) | Color |
|---|---|---|
| sRGB (the original) | 0, 200, 180 | The same teal |
| Display P3 | 90, 197, 180 | The same teal |
| Adobe RGB | 112, 199, 179 | The same teal |
| ProPhoto RGB | 119, 177, 166 | The same teal |
Notice that the red channel goes from 0 to well above 0. In the wider spaces this teal is nowhere near the edge of the gamut, so it needs some red to tone it down. That is the whole job of conversion.
When to convert: almost always. Converting a finished photo to sRGB for the web, or to a print lab’s profile, is a conversion.
When to assign: only to fix a wrong or missing label. If a file arrives with no profile and looks wrong, try assigning the space it was actually made in (sRGB is the usual guess, Display P3 for some phone images) and keep the one that looks right. Assigning to a file that was already labelled correctly breaks it.
What happens to colors that do not fit
Converting to a smaller space works perfectly for every color that fits inside it. Colors that fall outside have to be moved. Between the standard RGB spaces, software pulls each one in to the edge of what the smaller space can hold, which is a form of clipping: a range of very saturated reds can collapse into one flat red, and fine detail in a neon sign or a flower petal can go flat. How the move is made is called the rendering intent, and two of them matter for photographs:
- Relative colorimetric leaves every color that fits exactly as it is and moves only the ones that do not, to the nearest color the new space can hold. It is the usual default, often paired with black point compensation so the darkest tones map to the destination’s black instead of clipping. Choose it when only a few colors are out of range.
- Perceptual compresses the whole range a little so that differences between saturated colors survive, at the cost of shifting colors that would have fitted. Choose it when large saturated areas are out of range. It works with printer profiles; between the standard RGB spaces most software treats it like relative colorimetric.
The other two intents, absolute colorimetric and saturation, are meant for proofing and for graphics, and rarely suit a photograph. Soft proofing shows which colors will move before you print, and lets you compare the two intents on your own photo.
Most of the time none of this is visible. The bulk of real scenes, skin, foliage in shade, stone, sky away from the sun, overcast light, sits well inside sRGB. Only the most saturated colors are at risk: neon, stained glass, saturated flowers, tropical water, some sunsets.
Not sure whether a color is out of range or just too strong? Hue and saturation explains how far a color is from grey, which is what decides whether it fits.
Untagged files are read as sRGB
When a photo has no embedded profile, browsers, phones and most software assume it is sRGB. That one rule explains most of the “my colors look different online” problems:
- An Adobe RGB or ProPhoto RGB file that loses its profile is read as sRGB. Its numbers were made for a wider space, so every color is drawn less saturated than intended, and the photo looks dull and slightly flat.
- A Display P3 file from a phone that loses its profile looks a little muted in the same way.
- An sRGB file that loses its profile usually looks fine, because the guess is right. This is the strongest reason to export for screens in sRGB.
Profiles go missing when an export setting strips metadata, when an upload service throws it away to save bytes, or when a file is saved from software that does not write one. Current browsers do read embedded profiles; the usual culprit is a file that arrived without one, or one that was never converted out of a wide space.
Ready to save a finished photo for a website, a client or a lab? Exporting photos covers color space, size, format and sharpening for each destination.
Where color spaces show up in your workflow
In the camera
Most cameras offer an sRGB or Adobe RGB setting. It applies to JPEG files, the preview on the rear screen and the histogram. It does nothing to a raw file. A raw file holds the sensor’s own readings, which have no color space yet; the space is chosen later, when the raw file is developed. If you shoot raw only, the setting changes the preview and the histogram, not the data. If you shoot JPEG and post straight to the web, sRGB is the safer choice because the file will be read correctly everywhere.
On your phone
Many phones capture and display in Display P3, and tag their photos that way. Wide gamut is now an everyday thing: a photo of a red jacket or a turquoise pool may hold colors that a plain sRGB screen cannot show. On a P3 screen those photos look as captured; on an sRGB screen color management fits them in. Trouble starts only when the label is lost along the way.
While editing
Raw editors develop into a wide working space so that no color the camera recorded is thrown away while you work. If you hand a photo to a pixel editor, a 16-bit TIFF in Adobe RGB or ProPhoto RGB keeps that room (image file formats compares TIFF with the alternatives). Your screen still shows only what it can, so a wide working space protects the data, not the view; a calibrated wide gamut monitor shows more of it.
On the way out
Convert, do not assign, and embed the profile. For screens and social media, convert to sRGB. For a print lab, follow the lab’s instructions: many labs ask for sRGB, and some publish profiles for their printer and paper so you can soft proof. For your own printer, the printer and paper profile does the final conversion.
Want the full path from camera to screen to paper, calibration included? Photography color management walks through each step.
Which color space should you use?
The choice depends on where the photo is in its life, not on which space is “best”. Wide while it is being made, matched to the destination when it leaves.
- sRGB: Use it for: Anything for screens: web, social media, email, client galleries; many print labs. Bit depth: 8-bit is fine. Read correctly by: Everything, and assumed when a file has no label.
- Display P3: Use it for: Photos made and viewed on phones and tablets with P3 screens. Bit depth: 8-bit is fine. Read correctly by: Color managed devices and current browsers.
- Adobe RGB: Use it for: Printing on your own printer or at a lab that accepts it; a working space for JPEG shooters. Bit depth: 8-bit works, 16-bit is safer for heavy edits. Read correctly by: Color managed software only; looks dull if read as sRGB.
- ProPhoto RGB: Use it for: Editing raw files and keeping 16-bit master files. Bit depth: 16-bit only. Read correctly by: Color managed editors; never for delivery.
In practice that comes down to one workflow:
- Capture raw if you can. The camera’s color space setting then changes only the preview.
- Edit wide. Let the raw editor work in its wide space, and send files on to a pixel editor as 16-bit Adobe RGB or ProPhoto RGB.
- Keep a master. Save the raw file and its edits, or a 16-bit TIFF, as the copy you return to.
- Convert for each destination. sRGB for screens; the lab’s requested space, or your printer and paper profile, for prints. Embed the profile every time.
For printing, the right answer depends on who does the final conversion. If the lab asks for sRGB, or says nothing, send sRGB: its machines expect it. Some labs accept Adobe RGB, and some publish profiles for their printers and papers so you can soft proof first. If you print yourself, keep the file in Adobe RGB or ProPhoto RGB at 16 bits and let the printer and paper profile convert it, because good inkjet papers reach some cyans, greens and yellows that sRGB cannot hold.
Black and white photos care much less. A neutral grey sits in the middle of every space, so gamut makes no difference to it. What still differs is the tone curve each space uses to spread its levels, so convert rather than assign as with any photo. For a black and white photo going to a screen, sRGB is all it needs.
Where your monitor fits
Your monitor has a profile too, which describes how that particular screen shows color. Monitor calibration measures the screen and writes an accurate one. That makes your own view trustworthy, but it does not change what is in your files, and it cannot fix a photo that lost its label. Calibration and correct tagging solve different problems, and you need both.
Common mistakes with color spaces
- Thinking a bigger space is always better. A wide space only helps if every step after it is color managed and the file has the bit depth to fill it. An untagged ProPhoto JPEG is worse than a plain sRGB one.
- Treating assign and convert as the same thing. Assign changes colors; convert keeps them. To move a finished photo into sRGB, convert.
- Setting the camera to Adobe RGB to get more color in raw. Raw files are unaffected. Only the JPEG, the preview and the histogram change.
- Exporting without the profile. A wide space file with no profile looks dull everywhere. Leave “embed color profile” on.
- Blaming the browser. Current browsers read embedded profiles. When a photo looks washed out online, check whether it was converted to sRGB and whether the profile survived the upload.
- Fearing the conversion to sRGB. Most scenes fit inside sRGB, and those that do not lose only their most saturated corners. Soft proof if those colors matter.
- Editing ProPhoto RGB in 8-bit. The steps are coarse enough to band in skies and gradients. Use 16-bit until the final export.
- Expecting calibration to fix color everywhere. It corrects your monitor only. Other people’s screens and missing profiles are separate problems.
Try this: follow one photo’s label
This takes about ten minutes with a phone or a camera and a computer.
- Take a photo with a saturated color in it: a red or orange object, a bright flower, a blue or turquoise sign.
- Open it on a computer and find the color profile in the file’s information panel (on most systems, “Get Info”, “Properties” or the metadata panel of your editor). Note what it says: sRGB, Display P3, Adobe RGB, or nothing.
- Send the photo somewhere and back: upload it to a social or messaging app, then download it again. Check the profile on the downloaded copy.
- If your editor allows it, make two copies of the original. On one, assign sRGB. On the other, convert to sRGB. Put them side by side.
How to judge it. If the downloaded copy changed from Display P3 or Adobe RGB to sRGB and still looks the same, the service converted it properly. If the profile vanished and the strong color looks duller, the label was stripped. In step 4, the converted copy should look like the original and the assigned copy should look different, most of all in the saturated color. If the assigned copy looks the same, the photo was already sRGB.
Want to check you can keep assign, convert and sRGB straight? Try the color management quiz.
Questions about color spaces and profiles
Should I set my camera to sRGB or Adobe RGB?
If you shoot raw, it barely matters, because the setting affects only the JPEG preview and histogram. If you shoot JPEG for screens, choose sRGB. Choose Adobe RGB for JPEGs only if you will edit them in color managed software and convert to sRGB at the end.
Why does my photo look washed out after uploading it?
Usually because it was saved in a wide space such as Adobe RGB and either never converted to sRGB or lost its profile on the way, so it is being read as sRGB. Convert to sRGB on export, embed the profile, and upload that copy.
Does converting to sRGB lose color?
Only colors that sRGB cannot hold, which in most photos is a small number of very saturated areas, if any. Everything inside sRGB converts with no visible change. Keep your full edit as the original and export an sRGB copy, so nothing is lost for good. Converting the other way, from sRGB to a wider space, gains nothing.
What is the drawback of working in a large color space?
Its colors can be represented poorly when the file goes out to a smaller space or loses its label: out of range colors have to be moved, and an untagged wide space file looks dull. A wide space also spreads its levels further apart, so it needs 16-bit editing to avoid banding.
Is a color profile the same as a camera profile in my raw editor?
No. A camera profile (or look) in a raw editor decides how the sensor’s readings are turned into colors and contrast, a creative and technical choice made during development. An ICC color profile labels which color space finished pixel values belong to. They work at different stages.
Does white balance change the color space?
No. White balance sets what the camera treats as neutral in the scene’s light. The color space’s white point is a fixed part of its definition, used to describe colors after the photo is developed.