Image to ASCII Art Generator

This tool turns a still image into ASCII art: a block of monospaced text that, read from far enough away, looks like the picture. Your browser draws the image small, samples the brightness of each cell, and replaces it with a character from a ramp — a space where the picture is lightest and the densest character where it is darkest. It is for making ASCII art from a photo or a logo, and for turning an image into text you can paste into a README, a code comment, a chat message, a terminal banner or a plain-text email. The file is never uploaded to a server.

The output is a rectangle of characters: as many columns as you asked for, and as many rows as keep the picture's shape. A monospaced character is taller than it is wide — roughly twice as tall — so the tool draws about half as many rows as a same-width block of pixels would need, and the art does not come out stretched. Lines are separated by a plain newline character. A space is a real character rather than an empty cell, so a line that looks blank is usually a light part of the picture, not a mistake.

The character ramp is the ordered list of characters from lightest to darkest. The default, ` .:-=+*#%@`, spends ten steps from a space to an at-sign; a longer ramp keeps more shading and a two-character ramp gives a stark black-and-white stencil. The first character is the lightest on purpose, so a normal photo on a white background reads correctly. Tick invert for a dark terminal — the ramp flips, so a space becomes the darkest character and bright pixels carry the weight. Choose Custom to type your own ramp, including block characters like ░▒▓█.

Colour turns each character into 24-bit ANSI: the tool prints an escape sequence before every character that sets its foreground to that cell's colour, and resets at the end of each row. That is the format a terminal reads, not plain ASCII — it shows as ordinary text in an editor and as colour in a terminal. This page previews the characters without the colour codes, because escape sequences are invisible in a web page, so copy and download give you the ANSI text. Leave colour off unless you are pasting into a terminal; it is much longer than the plain version.

Be clear about what this is: a downsampled approximation. Almost all of the picture's detail is thrown away and only one brightness value per character cell survives, so fine features disappear. The proportions you see depend on your viewer's font — the tool assumes a cell about twice as tall as it is wide, and a different font or line height will stretch or squash the result slightly. Gradients look banded, because the ramp has only as many steps as it has characters.

How it works: your browser decodes the image (a phone photo is turned upright using its EXIF orientation), draws it onto a small canvas — one pixel per output column — reads those pixels back, and this tool does the rest in JavaScript on the raw pixel data. Each cell's colour is averaged, composited over white where it was transparent, and reduced to one brightness value with the Rec. 709 luma coefficients. Limits: up to 50 images at a time, 50 MB each, and 1 to 500 columns. Your images stay on your device: the Network tab in your browser's developer tools shows no request carrying them.

1. The file's size is checked (50 MB) before it is read, and its first bytes confirm it is PNG, JPG, WebP, GIF, BMP or AVIF; the browser decodes it upright. 2. The image is drawn onto a small canvas: one pixel wide per output column, and twice as many pixels tall as the output has rows. A 100-column request therefore samples a canvas 100 px wide. 3. Each output cell averages the pixels that fall inside it, compositing each pixel over white using its alpha first (a canvas leaves transparent areas as colourless black, which would otherwise read as dark). 4. The averaged colour becomes one brightness value with the Rec. 709 luma coefficients: luminance = 0.2126 R + 0.7152 G + 0.0722 B. 5. The brightness is scaled onto the ramp: index 0 is the lightest character (a space) for the brightest cells and the last index is the densest for the darkest. A 10-character ramp has 10 steps, each covering about one ninth of the range; invert reverses the direction. 6. The characters are joined into rows separated by newlines. With colour on, each character is preceded by a 24-bit ANSI foreground code (ESC[38;2;R;G;Bm) and each row is closed with ESC[0m. 7. The text is offered for copying and saved as a .txt file named like photo-ascii.txt.