Image Histogram
Inspect RGB channel distribution for an uploaded image.
Overview
The Image Histogram tool charts the distribution of red, green, blue and luminance values across every pixel of an uploaded picture. Drop in a photo and the tool returns four bar charts with 256 bins each, so you can see at a glance whether the image is under-exposed, blown out, or sitting comfortably across the tonal range.
Photographers use this for the same reason they trust the histogram on the back of their camera — it is the most honest exposure indicator available. Print designers reach for it when checking that a scan has clean black and white points, and developers use it when verifying that an image-processing pipeline preserved tonal range.
How it works
The decoder walks the image once and counts how many pixels fall into each of the 256 possible 8-bit values for each channel. The luminance histogram uses Rec. 709 weighting — 0.2126 R + 0.7152 G + 0.0722 B — so its peak position matches the impression of overall brightness rather than a naive RGB average.
The output is normalised so the tallest bar reaches the top of the chart, which keeps the shape readable regardless of resolution. For 16-bit-per-channel inputs the values are downsampled to 8 bits before binning so all images use a comparable horizontal axis.
Examples
Backlit portrait → luminance histogram is bimodal: a tall cluster at
the dark end (subject silhouette) and another at the bright end (sky).
Studio product shot on white → luminance is heavily skewed right with
a thin spike at 255 (the seamless background).
Foggy landscape → narrow histogram centred around mid-grey with no
shadows below 60 or highlights above 200.
FAQ
Why are the channels offset from each other?
The relative position of the red, green and blue peaks reflects the colour cast of the scene. Warm sunset images push red right and blue left, for example.
Does a peak at 255 mean blown-out highlights?
A tall bar at exactly 255 means many pixels were clipped — detail there is unrecoverable. A short bar is harmless and just reflects pure-white objects in the scene.
Is the histogram per-channel or combined?
Both. The tool draws individual red, green and blue charts plus a separate luminance chart so you can spot per-channel clipping that the luminance view hides.
Can it analyse 16-bit raw scans?
Yes — the input is downsampled to 8-bit for display but the count is taken before downsampling, so accurate distributions still surface.
How big can the input be?
The counter is O(n) in pixel count and runs comfortably on multi-megapixel files. Very large panoramas are downsampled before binning to keep the chart responsive.