Can a photo actually read your undertone?

Undertone gets talked about like a hidden layer under your skin, as if there were a true color buried somewhere and the job is to dig it out. There isn't one. What exists is the color your skin actually is, and every read of it – yours, a stranger's, a camera's – is a read of light bouncing off you and arriving somewhere else.

That matters because it tells you what can go wrong. Nothing has to be hiding for a read to fail. The light just has to be doing something on the way in.

A photo has two real advantages over you in a mirror. It holds still, so the same face can be compared against a fixed reference instead of against your memory of ten minutes ago. And it doesn't have your eyes, which quietly adapt to whatever room you're in and then insist that room was neutral. That second one is why your foundation looked fine in the shop.

It also has one disadvantage you should know about. If you're wearing base, the camera reads the base. Foundation, tinted moisturiser, concealer and bronzer all sit between the lens and you, so on a full-coverage face the honest description is the complexion you built rather than the skin underneath it. siOsi reads undertone off the least-covered skin it can find – jaw, upper neck, temple, hairline – and it says so when there isn't enough of that to go on.

Ordinary indoor bulbs sit around 2700K, which is a long way down the warm end. Overcast daylight through a window lands somewhere near 6500K. That gap is bigger than the difference between a warm and a cool complexion, which is the whole problem in one sentence.

Under a warm bulb, everybody drifts warm. Cool skin looks neutral, neutral skin looks warm, and warm skin looks like it has been basted. Then you do the gold-versus-silver test and gold wins, because gold was always going to win in that room.

This is the failure that quietly wrecks most at-home undertone tests, and it isn't a flaw in you or in the tests. It's a mismatch between where the test assumes you are standing and where you actually are. Ten minutes near a window fixes more of it than any amount of squinting.

It's also the reason siOsi measures the light itself on every photo instead of trusting its own impression of it. When the bright parts of your frame come back tinted, the temperature read is treated as unproven and the confidence drops. The undertone it reports does not get quietly swapped to suit the lighting – that would be inventing a person. It reports what it sees, and tells you how much to lean on it.

Most color tools will give you a season no matter what you upload. Feed one a black-and-white photo and it will still pick a side, because the answer field has to be filled in. That's a design decision, and it isn't a flattering one.

siOsi is allowed to come back with a low number, and it is allowed to come back with no undertone at all. That happens when there's nothing to read it from – heavy base over the jaw and neck, a strong color cast off a wall or a bulb, a greyscale or heavily filtered image, or so little bare skin in frame that any answer would be a coin toss dressed up as a measurement.

A low confidence score is a result, not a failure to produce one. It's the page telling you which of the two things went wrong: the photo, or the ambiguity. Those have different fixes, and knowing which one you're looking at saves you from redoing a test that was never going to work.

And a genuinely undecided answer is information too. Some people really do sit close to a boundary, and being told so beats being handed a confident label you then spend a year fighting half your wardrobe over.