ASCII Art: How Text Pictures Are Made
ASCII art is a picture assembled from ordinary printable characters, with each glyph's weight standing in for a pixel's brightness. The style grew out of typewriters and bulletin boards and never left, because plain text renders everywhere. How the mapping works and which controls shape it is below; to draw a banner from a word rather than an image, the ASCII Text Generator on this site renders it in a classic font.
Free and instant, with ten classic fonts compiled in, and the result copies as plain characters that stay readable in a terminal, a comment or a readme.
draw a word as a banner in your browser →A picture that is also just text
ASCII art is an image made out of characters people can type. There is no file format and no decoding step: the picture is the text, and any program that displays characters displays the picture, more or less correctly depending on the font it uses. The name comes from the ASCII character set, the 128 codes early terminals shared, though modern pieces borrow symbols from beyond it, such as the shaded blocks used to fake a smooth gradient. What makes the approach work is density. A period covers less ink than the letter A, and an at sign or a hash covers more, so a row of symbols ordered from light to heavy becomes a row of shading if each character is matched to the tone it should represent. Done well, the eye stops reading letters and starts seeing a shape.
Where the style came from
Long before home computers, typists were assembling pictures from punctuation on paper; typewriter portraits date to the late nineteenth century, built line by line with backing spaces and overstrikes. Computing inherited the habit because early printers and screens could do letters and little else. By the 1980s, bulletin board systems were trading artwork made of characters, and an entire aesthetic formed around text-mode screens, where the character set was the only paint available. Banner generators followed: the FIGlet program, written in the early 1990s, read font files of hand-drawn letterforms and turned a typed word into a large one, and those fonts are still in use. The scene that produced them has mostly moved on, but the files did not, and text-only art remains one of the few formats from that era that opens correctly on everything built since.
Two kinds of ASCII art: letters and pixels
Tools carrying an ASCII name usually belong to one of two families, and they share nothing but the label. The first family draws text: you type a word, the program looks up a hand-drawn shape for each letter and stacks those shapes into a banner. The second family derives a picture from an image: you supply a photograph or a logo, and the program samples it and chooses a character for each spot. The ASCII Text Generator, the banner tool this guide hands off to, is the first kind. It takes up to about eighty characters, offers ten classic font faces — among them Standard, Slant, Doom and Star Wars — and includes a compact spacing option that tightens wide gaps so a banner fits a narrow pane. The output is plain text, copied with one click. If you expected a photograph turned into symbols and got a clean word instead, you opened a tool from the other family.
How brightness becomes a character
An image-to-text converter reduces each pixel to a single number: its luminance, computed as a weighted mix of red, green and blue that matches how the eye weighs them, with green counting most. That number runs from dark to bright, and the converter keeps a ramp — around ten characters sorted by how much ink each one puts on the page, from a space through punctuation and letters up to a solid symbol. Every pixel's luminance is matched to a position on the ramp, and the character sitting at that position is written out. Adjusting the contrast control changes where those tones land by stretching or squeezing the range before the lookup, which is often the difference between a gray wash where everything looks alike and a portrait whose features separate. Color is not carried through at all; the output is a shape drawn in ink density, which is why subjects with strong light and dark areas survive the process and flat, hazy ones do not.
Converting an image, step by step
With the image to ASCII converter linked from this guide, the flow is short. Drop an image onto the page; the file is read locally, so nothing leaves the device. The converter draws it to a canvas at the width you chose — the slider runs from 40 to 200 characters and starts at 80, a good middle for a terminal or a text file. The height is worked out from the image's proportions and then halved, because text glyphs are taller than they are wide. A contrast slider spanning 32 to 256 reshapes the tonal range before the character lookup, starting at 128. A dropdown offers four ramps — Standard, Detailed, Blocks and Minimal — trading detail for size, from a ten-step wash up to a dense ramp with far more intermediate tones. The settings are applied as the image is read, so the way to compare two looks is to change a control and drop the image again. The finished piece appears in a monospace block, ready to copy.
Why monospace decides everything
A character grid lines up only if every character occupies the same width. In a monospace font — the kind terminals, code editors and plain-text files use — an i and an M take identical space, so column one hundred sits directly under column one. Switch to a proportional font, where widths vary from glyph to glyph, and the rows slide out of alignment: what was a circle becomes a fan. That single requirement explains most of the places the style still appears, and most of the places it breaks. The art survives in terminals, in code comments, in markdown fenced blocks and in plain text files. It fails in word processors, in chat windows that reflow text, and anywhere trailing spaces are trimmed or line breaks are rewrapped. When a piece arrives mangled, the font it landed in is the first suspect, not the art.
Tips that separate a smudge from a portrait
Choose subjects with contrast. A face against a plain background, a logo with clean edges, a scene with strong shadows — each gives the ramp real work to do, while a hazy landscape turns into an even field of punctuation. Start at the default width and go wider only when more detail is needed, or narrower when the output has to fit a fixed pane; a 200 character rendering is nearly impossible to read on a phone. For photographs, the Detailed ramp rewards the extra width; for logos and large shapes, Blocks produces bolder silhouettes; for a small signature-sized result, Minimal stays legible when only a handful of characters fit across. Mind the surface the piece will live on: the ramp sends the darkest pixels to the sparsest characters, so the tones read correctly where characters are light on a dark ground, as in a terminal, and invert when the same text is pasted as dark glyphs on white. When in doubt, render two widths and keep the one that still reads at a glance.