Compute an image histogram for every pixel channel

Supply one or more named channels, choose how many equal-width bins you need, and receive an ordered count array for each channel. It is useful when image bytes have already been decoded and you need reproducible statistics without uploading or re-encoding the original file. Every value is checked against the 8-bit range, every supplied pixel contributes to exactly one bin, and an empty pixel list is rejected instead of producing a misleading all-zero result.

Prepare named channel values

Provide the decoded pixels as a channels array. Each entry has a unique channel name and a values array containing integer intensities from 0 through 255. Names such as red, green, blue, alpha, and gray are conventional, but the calculator does not impose a fixed color model; a channel may represent any component that uses the same 8-bit domain. Keep the values for a channel in their original order if that is convenient, although order does not affect a histogram because only frequency matters. Every channel must contain at least one pixel value. The request is rejected when the outer channel list is empty, when an individual values list is empty, when a name is missing or duplicated, or when a value is fractional or outside the allowed range. These checks prevent incomplete source data from looking like a valid distribution. The capability accepts up to sixteen channels and up to one million values per channel, keeping browser and API execution bounded.

Understand bin assignment and boundaries

Choose an integer bin count from 1 through 256, or omit it to use 16 bins. The full 8-bit intensity domain is divided into equal-width intervals, starting at zero and ending at 255. For 16 bins, each bin covers sixteen consecutive values: the first counts 0 through 15, the second counts 16 through 31, and the last counts 240 through 255. More generally, the bin index is the floor of the pixel value multiplied by the bin count and divided by 256. The upper endpoint is explicitly kept in the final bin, so a value of 255 is never lost to an out-of-range index. This definition is deterministic even when the selected bin count does not divide 256 evenly. Each pixel increments exactly one counter, and the counters for a channel therefore sum to its pixel_count. Use fewer bins for a compact overview and more bins when subtle intensity differences matter.

Read, compare, and validate the result

The response reports the selected bins, the fixed value_min and value_max, and one histogram object for each input channel. Each histogram preserves the channel name, states pixel_count, and returns counts in ascending intensity order. Position zero is always the darkest interval and the final position is always the brightest interval, which makes the arrays straightforward to chart or compare. Before using a result downstream, sum each counts array and confirm that it equals pixel_count; the algorithm guarantees this invariant, and checking it can expose accidental truncation elsewhere in a pipeline. Histograms summarize frequency, not spatial arrangement, so two images with very different shapes may have identical results. They also do not decode image files, infer color spaces, normalize exposure, or combine channels. Decode the source first, retain the channel interpretation needed by your application, and use this capability for the focused counting step. Browser execution keeps the supplied lists local, while API automation costs $0.002 per request.