Bulk Color Extractor: Colors of Many Images at Once

About the Bulk Color Extractor

The Bulk Color Extractor reads a whole folder of pictures and gives you one row for each. A row shows the file name, a small thumbnail, the five most common colors with the share of the picture each one covers, and the average color of the picture. You can add up to 20 images at once, watch each one move from queued to done, and then save the whole table as a CSV file for a spreadsheet or as JSON for a script.

It reads PNG, JPEG, GIF, BMP, WebP and AVIF files up to 50 MB each and up to 200 MB in all. Your images never leave your device. Nothing is uploaded, there is no account, and every picture is read and counted by your own browser, one after another.

How to use it

If one file cannot be read, only its row shows the reason, such as an unsupported type or a file over the size limit. The other rows carry on. A fully transparent picture finishes as Done and says "No colors found."

How each row is worked out

Every picture is read at up to 1,000,000 pixels, so a very large photo is scaled down first. Pixels with an alpha value below 25 (almost fully transparent) never count. From the counted pixels the tool works out two things.

The average color is the plain mean of each sRGB channel, rounded to the nearest whole number:

$\bar{c} = \operatorname{round}\left(\frac{1}{N}\sum_{i=1}^{N} c_i\right)$

where $c$ is the red, green or blue value and $N$ is the number of counted pixels. A picture that is half rgb(255, 0, 0) and half rgb(0, 0, 255) averages to $\left(\frac{255 + 0}{2}, 0, \frac{0 + 255}{2}\right) = (127.5, 0, 127.5)$, which rounds to #800080.

The top colors come from median cut on a histogram that keeps 5 bits of each channel. The tool splits the counted pixels into up to five boxes of similar colors, reports the mean color of each box, and gives each its share:

$\text{share} = 100 \times \frac{\text{pixels in the box}}{\text{counted pixels}}$

In the same red and blue picture the two colors are #ff0000 at 50.0 percent and #0000ff at 50.0 percent. Shares are rounded to one decimal place, so a row can add up to 99.9 or 100.1 percent. A picture with fewer than five distinct groups of colors lists fewer than five.

What the files contain

The CSV file is named colors.csv. Its first line is the header

file,average,colour_1,share_1,colour_2,share_2,colour_3,share_3,colour_4,share_4,colour_5,share_5

and each finished picture adds one line. The colors are hex codes and the shares are percentages with one decimal. Missing colors are left empty so every line has the same number of columns. A byte order mark is added at the start so Excel and other spreadsheets read file names with accents correctly.

A file name that contains a comma or a quote is wrapped in quotes with inner quotes doubled. A name that starts with =, +, - or @ gets a leading apostrophe, which stops a spreadsheet from running it as a formula.

The JSON file is named colors.json. It is a list with one object per picture, holding file, average and colors, where each color has a hex and a share.

Limits worth knowing

• Up to 20 images, 50 MB each and 200 MB together. If you add more than fit, the page says how many were added and how many were skipped.
• Pictures are processed one at a time, so a long list takes a little while but the page stays responsive.
• Only finished rows are exported. Rows that failed are left out of both files.
• Colors are numbers in sRGB as your browser shows them. The tool does not name colors, remove backgrounds or read embedded color profiles.

Related tools

To see more than five colors of a single picture, with sorting and ignore options, use the Color Extractor. To read one exact pixel, use the Image Color Picker. To resize or change the format of the same pictures first, try the Image Converter.

Frequently asked questions