Free tool · nothing is uploaded
Histogram viewer
The RGB and luminance distribution of any photograph, and exactly how much of the frame is crushed to black or blown to white — the two failures no amount of editing can undo.
A histogram is a description, not a verdict
The most common thing said about histograms is that a good one is an even spread from edge to edge. That is not true, and following it will flatten your photographs. A low-key portrait belongs on the left. A snowfield belongs on the right. The shape describes the picture you took; it has no opinion about whether that picture is any good.
What a histogram genuinely diagnoses is the edges. Tones stacked against either end have been recorded as pure black or pure white, and those values are not dark or bright — they are absent. That is the reading worth acting on, and it is why the clipping percentages sit above the plot here rather than below it.
Why the channels matter separately
Overexposure rarely arrives evenly. Shoot a sunset and the red channel hits its ceiling while green and blue are still comfortable — the sky loses its gradient and goes to a flat orange mass, long before the frame as a whole looks overexposed. A luminance histogram averages that away and shows nothing wrong.
Saturated colour does the same at the other end: deep blues in shadow can pin the blue channel to zero while the picture still reads as merely moody. Watching the three channels apart is what makes those visible in time to do something about them.
Frequently asked questions
How do I read a photo histogram?
Left is black, right is white, and the height at any point is how many pixels sit at that brightness. A hump on the left means a dark photograph, not a wrong one. There is no correct shape — a histogram describes the picture, it does not grade it. The one thing it genuinely diagnoses is whether tones are stacked against either end, which is clipping.
What is clipping and why does it matter?
Clipping is where pixels have hit pure black or pure white. Those values carry no information at all, so lifting shadows or pulling highlights afterwards recovers nothing — the data was never recorded. It is the one histogram reading that is unambiguously a problem, and the only fix is at capture, by changing exposure or by bracketing.
Is some clipping acceptable?
Usually, yes. A specular highlight on chrome or water, or a genuinely black night sky, is meant to be at the limit. What matters is whether clipping has eaten something you wanted — a face, a cloud, texture in a dress. A percentage tells you how much of the frame is gone; only you can say whether that part mattered.
Why do the RGB channels separate at one end?
Because a single channel has clipped before the others. A sunset often blows the red channel while green and blue still hold detail, which is why the sky goes flat and orange before the whole frame looks overexposed. Watching the channels separately catches that well before a luminance histogram would.
Does the luminance histogram use the same scale as RGB?
No, and deliberately. Luminance is normalised against its own peak rather than the shared RGB one — a strongly tinted image has a much taller luma peak than any single channel, and sharing a divisor would squash the colour curves into the floor of the plot.
Is my photo uploaded?
No. It is decoded and sampled on your own device with a canvas. Nothing is transmitted, so there is no copy anywhere but your machine.