How do I read a histogram?

The horizontal axis is levels 0 to 255, dark on the left, bright on the right; the vertical axis is how many pixels sit at that level. All bunched left means underexposed, all bunched right means overexposed, bunched into a narrow middle band means flat. A bar hitting the ceiling at either end usually means crushed blacks or blown highlights — that detail is gone, and pulling it back only yields gray.

Why does the Pixels number not match my image?

Two reasons. First, the statistics run on a downsampled preview: an image longer than 256 px on its long edge is scaled to 256 before counting, so that number is the sampled count, not the document total; at or under 256 nothing is scaled. Second, transparent pixels are not counted, so an image on a transparent background comes out lower still. The shape of the distribution is unaffected, so judging exposure from it is perfectly sound — just do not read that number as your document size.

What do Mean and Std Dev tell me?

Mean is overall brightness — around 128 is neutral, and clearly lower means the picture is dark overall. Std Dev is contrast: a small value means pixels are crowded into a narrow band of levels, which reads as flat and gray; a large value means the tones are spread out. Making a photo look "clear" usually amounts to pushing the std dev up, with levels or curves.

Is it free? Are my images uploaded?

Free, no sign-up, no usage limit. It all runs in your own browser; nothing is uploaded while you're signed out, and once you sign in your work is saved automatically to your own cloud projects.

Histogram · read exposure and color distribution

Five channels, mean / std dev / median. Processed locally; nothing uploaded while signed out.

or start a blank document

A histogram is the photo's blood test: the horizontal axis runs black to white, the vertical says how many pixels sit at each brightness. Piled to the left means underexposed, piled right means overexposed, and stacked hard against either end means crushed black or blown white — where there is no detail left to recover.

Open Window > Histogram. The horizontal axis is 256 levels from 0 to 255, dark on the left and bright on the right, and a tall bar means many pixels sit at that level. Five channels are available — Luminosity, R, G, B, and Color — where "Color" overlays the three curves so a cast shows up at once; and two sources, Entire Image and Active Layer, the latter being what you want while grading one layer. Four readouts sit underneath: Mean (overall brightness), Std Dev (contrast — the smaller, the flatter), Median, and Pixels. Hover a bar and it also shows that Level, its Count, and its Percentile — the share of pixels no brighter than that point, which is exactly what you want when setting black and white points. Two things to be clear about: transparent pixels are left out of the statistics entirely, so an image on a transparent background only counts the parts with content; and the statistics run on a downsampled preview — an image longer than 256 px on its long edge is scaled to 256 before counting. So Pixels reports the sampled count of non-transparent pixels, not the document total; at or under 256 nothing is scaled and the readouts are exact. The shape of the distribution is unaffected, but do not read that pixel count as your document size. It all runs in your own browser (once you sign in, your work is saved automatically to your own cloud projects).

Key features

Five channels

Luminosity, R, G, B, and Color — the last overlays all three so a cast shows at once.

Whole image or one layer

Switch the source to Active Layer while you are grading a single layer.

Four readouts plus hover detail

Mean, Std Dev, Median, Pixels; hover a bar for level, count, and percentile.

Runs locally

Processed in your browser. While you're signed out nothing is uploaded — only an automatic draft stays in this browser (clearing browser data deletes it); once you sign in, your work is saved automatically to your own cloud projects.

How to use

1
Open the image

Drag it straight in.
2
Open the histogram panel

Window > Histogram.
3
Pick a channel and source

Luminosity for exposure, Color for casts; switch the source when grading one layer.
4
Correct accordingly

Bunched left is underexposed, bunched right is overexposed; open it up with levels or curves.

FAQ

How do I read a histogram?: The horizontal axis is levels 0 to 255, dark on the left, bright on the right; the vertical axis is how many pixels sit at that level. All bunched left means underexposed, all bunched right means overexposed, bunched into a narrow middle band means flat. A bar hitting the ceiling at either end usually means crushed blacks or blown highlights — that detail is gone, and pulling it back only yields gray.
Why does the Pixels number not match my image?: Two reasons. First, the statistics run on a downsampled preview: an image longer than 256 px on its long edge is scaled to 256 before counting, so that number is the sampled count, not the document total; at or under 256 nothing is scaled. Second, transparent pixels are not counted, so an image on a transparent background comes out lower still. The shape of the distribution is unaffected, so judging exposure from it is perfectly sound — just do not read that number as your document size.
What do Mean and Std Dev tell me?: Mean is overall brightness — around 128 is neutral, and clearly lower means the picture is dark overall. Std Dev is contrast: a small value means pixels are crowded into a narrow band of levels, which reads as flat and gray; a large value means the tones are spread out. Making a photo look "clear" usually amounts to pushing the std dev up, with levels or curves.
Is it free? Are my images uploaded?: Free, no sign-up, no usage limit. It all runs in your own browser; nothing is uploaded while you're signed out, and once you sign in your work is saved automatically to your own cloud projects.

Try Histogram · read exposure and color distribution now

No install · no sign-up · local stays local · free AI credits every day