Ask a tool to pull colors from a photo and it will happily hand you two, eight, or twenty. The number itself isn't the interesting question — what you're going to do with the palette is. A duotone and a full extraction solve different problems, and reaching for the wrong one is the most common reason a "color-matched" design still looks off.
What a duotone actually captures
A duotone reduces a photo to two tones: a dark and a light, mapped across the image's shadows and highlights. It throws away almost everything about the original color — hue variation, mid-tones, saturation shifts — and keeps only the contrast structure. That sounds like a loss, but for certain jobs it's exactly the point. A duotone treatment reads as a single, deliberate mood rather than "here are this photo's colors." Editorial photography, poster treatments, and app icons often lean on two-tone extraction because it flattens a busy image into something a logo or a headline can sit on top of without competing for attention.
The trade-off is obvious once you say it out loud: a duotone can't tell you what a product's actual packaging color is, because it was never trying to. If someone asks "what's the exact color of that dress in the photo," a duotone gives them a stylized answer, not a factual one.
What a full palette is for
An eight- or twelve-color extraction is closer to an inventory. It's trying to answer "what colors are actually present, and how much of the frame does each one cover." That's the version you want when the goal is accuracy: matching a website's accent color to a product photo, building a brand palette from a reference image, or making sure a resized thumbnail doesn't clash with the colors already in the picture.
The cost of a full palette is noise. Real photos are full of near-duplicate tones — a dozen slightly different greens in a houseplant, or three near-identical grays in an overcast sky. A good extractor has to actively merge similar tones together (a "minimum distance" between chosen colors, in practical terms) or the palette ends up cluttered with colors that are technically different but visually redundant.
Choosing between them
A rough rule that holds up in practice: reach for two-to-three tones when the palette needs to become a mood or a stylistic filter, and reach for eight-plus when the palette needs to be usable data — something you'll paste into a stylesheet, hand to a print shop, or reference in a design spec. Anything in between, five or six tones, is a reasonable middle ground when you want real accuracy but also want the result to stay skimmable at a glance, which is why it's a common default for brand work.
It's also worth matching the grain to the photo itself. A single-subject product shot on a plain background rarely needs more than five or six colors — most of the frame is one or two dominant tones plus the product. A busy scene, a crowd, a landscape at golden hour, benefits from a coarser extraction simply because there's more actually happening in the frame that a tight duotone or five-color palette would flatten away.
A quick way to test it
If you're not sure which grain to use, extract twice: once fine, once coarse, and look at what changes. If the coarse version just adds near-duplicates of the fine one — three shades of the same beige — the photo didn't need the extra colors, and the tighter palette is the more useful, more portable result. If the coarse version reveals a genuinely different color that the fine pass missed entirely, that's a sign the image has more range than a tight palette can represent, and the larger set is worth keeping.
Either way, the number of colors is a decision, not a default. Pick it based on what the palette needs to do next, not on habit.