Editing an ImageLesson 8 of 78

Contents

Editing an Image

Lesson 8 of 78. Sign in to keep your place.

Module 1 · What an image is made of

Some colours have no number

9 min read

How much of what a person can see each space enclosessRGB: the web, phones,anything unlabelled35Display P3: current phoneand laptop screens45Adobe RGB (1998): extendedwhere a press can print50Rec.2020: the videostandard; no consumerscreen covers it76ProPhoto RGB: larger thanvision, and 16-bit only90per cent of the visible chromaticitiesAreas on a chromaticity diagram, which is a statement about which colours, never about how bright. Awide space spreads the same 256 codes over more colour, so every step is a bigger jump: ProPhoto in 8bits bands on ordinary skin, and its primaries lie outside human vision entirely.
Areas on a chromaticity diagram, which is a statement about which colours, never about how bright. A wide space spreads the same 256 codes over more colour, so every step is a bigger jump: ProPhoto in 8 bits bands on ordinary skin, and its primaries lie outside human vision entirely.

Figure

What a colour space actually specifies

Saying a file is “sRGB” is saying three things at once:

  1. Primaries: precisely which physical red, green and blue the values 255,0,0 / 0,255,0 / 0,0,255 refer to.
  2. White point: what all three at maximum means. Almost everything uses D65, a standardised daylight around 6500 K.
  3. Transfer function: the gamma curve from the earlier lesson.

Change the primaries and the same numbers mean different colours. That is the whole mechanism behind every “it looked different on their screen” conversation.

The space’s gamut is the triangle those three primaries enclose on the chromaticity diagram: every colour you can make by mixing them. Anything outside the triangle has no representation in that space at all. Not a rounded one: none.

The spaces you will meet, and how much bigger they are

  • sRGB (1996): the default for the web, for phones, for anything unlabelled. It covers roughly a third of the chromaticities a person can see. Its weak corner is saturated cyans, greens and oranges.
  • Display P3: the space of modern phone and laptop screens, about 25 per cent larger than sRGB, mostly in reds and greens. Apple devices have used it since 2015 and current Android flagships do too.
  • Adobe RGB (1998): similar size to P3 but extended in a different direction, notably cyan-greens. It exists because those are the colours a CMYK press can print and sRGB cannot, which is why print workflows still use it.
  • ProPhoto RGB: enormous, and deliberately so: its primaries lie outside the range of human vision, so some representable values correspond to no real colour. Lightroom works in a variant of it internally.
  • Rec.2020: the video standard for wide-gamut HDR. No consumer display covers it fully.

A concrete example: a saturated traffic-cone orange, or the cyan of a swimming pool, or a pure LED green sign. Photograph any of those and the sensor recorded them fine. Convert to sRGB and they hit the wall of the triangle and flatten into a single muddy tone with no internal detail: a shape that was a shaded cone becomes a flat orange blob.

What happens at the wall: rendering intents

When you convert from a larger space to a smaller one, the software has to decide what to do with the colours that do not fit. That decision is the rendering intent, and you choose it in the Convert to Profile dialogue.

  • Relative colorimetric: keeps every in-gamut colour exactly where it is and clips the out-of-gamut ones to the nearest point on the boundary. Accurate for most of the picture, and it destroys detail in the saturated areas, because several distinct colours all clip to the same edge point. Turn on black point compensation with it, or your shadows will block up.
  • Perceptual: compresses the whole gamut inwards so the relationships between colours survive. Every colour shifts slightly, including ones that would have fit. Better for an image with large saturated areas; unnecessary for one without.
  • Saturation: keeps things vivid at the expense of accuracy. For charts and business graphics, not photographs.
  • Absolute colorimetric: additionally reproduces the source white point, so paper white shifts. For proofing one press on another, not for delivery.

The practical rule: relative colorimetric with black point compensation by default; switch to perceptual when a soft proof shows you a large area going flat.

Soft proofing: seeing the loss before it happens

You can preview the conversion. In Photoshop, View > Proof Setup and View > Gamut Warning paints out-of-gamut pixels grey. GIMP has View > Colour Management > Soft-proofing with an out-of-gamut warning colour you can set. Both let you load an ICC profile for the destination: a printer’s paper profile, or plain sRGB.

Do this before you promise a client a colour. Ten seconds of gamut warning prevents the conversation where a printed brochure’s brand green is visibly duller than the website’s.

ProPhoto and the 8-bit trap

A wide space spreads the same 256 codes over a much larger range of colour, so each step is a bigger jump. ProPhoto RGB in 8 bits is a known way to produce banding in ordinary skin tones, because the steps between adjacent codes are large enough to see.

The rule follows directly: wide gamut requires more bits. If you work in ProPhoto or Adobe RGB, work in 16-bit. sRGB in 8-bit is fine because the space is small enough that the steps stay under the visible threshold for most content.

Free tools, and the honest limits

GIMP has full ICC colour management, soft proofing and profile conversion. darktable and RawTherapee both let you pick a working profile and an output profile. DisplayCAL with ArgyllCMS drives calibration hardware and builds profiles, free. littleCMS is the engine underneath most of it.

The gap worth knowing: Photopea’s colour management is thin. It largely assumes sRGB. For web work that is usually harmless; for print work, or for handling a client’s Adobe RGB files, it will silently get things wrong. This is the clearest case in the course where the free browser tool is not a substitute and GIMP is.

And two limits on the whole idea. First, a wide-gamut file on a wide-gamut monitor still looks wrong if the application is not colour-managed: most browsers are now, most games and many desktop utilities are not, so the same file can look oversaturated in one window and correct in another on the same screen. Second, gamut is about which colours, not how bright. It says nothing about dynamic range, and a wide-gamut standard-range image is a completely different thing from an HDR one.

Every lesson here is open. Your progress is recorded once you pass the Design, before Any Software exam.

Check your understanding

A photograph of a bright cyan swimming pool, edited in Adobe RGB, is converted to sRGB with relative colorimetric intent. The pool comes out as a flat patch with no visible variation in the water. What happened?

Pick the one you would defend. Your answer stays private.

Next lesson · 8 minThe monitor is a measuring instrument nobody calibrated

Stuck on something?Ask Neema, your AI tutor

How to cite: Addaly (2026) ‘Some colours have no number’, in Editing an Image. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/image-editing/colour-spaces-and-gamut (Accessed: 13 October 2026).