Image to ASCII Art Converter — Photo to Text

Almost always because the source has too little contrast, too much fine detail, or both. ASCII art has roughly ten brightness levels to work with, so a busy photograph loses everything that depends on subtle tonal differences. Push the Contrast slider up, try a wider output, and prefer images with a clear subject against a plain background — logos, icons, portraits and high-contrast photos convert far better than a detailed landscape.

It depends where the art is going. 60-80 characters fits inside a standard code comment or a git commit message without wrapping. 100-120 suits a README or a terminal window. Above about 150 the detail is genuinely better but the result stops fitting anywhere text normally goes, so it is mostly useful for something you will view as an image or an HTML file.

Blocks is the default — it uses the Unicode shading characters ░▒▓█, which are solid rather than sparse, so the result reads as a picture immediately. Because they are Unicode rather than ASCII, they need a font and a destination that support them — anywhere you would actually paste this (a terminal, an editor, a README) handles them fine, but a legacy system might not. Standard is the classic ten-step ramp and is pure ASCII, so it pastes anywhere. Detailed uses seventy characters for much finer tonal gradation, which pays off on photos at wide settings but can look busy on small output. Silhouette uses just two levels, turning the image into a stencil.

You probably need the other background mode. "Dark on light" assumes the art will sit on a white page, so dark characters stand in for dark pixels. "Light on dark" assumes a terminal or dark editor, where the characters themselves are the light part, so the mapping flips. Switching modes fixes an image that looks like a photographic negative.

Yes, but not as plain text — plain text has no way to carry colour. Switch to Colour mode and each character is tinted with the colour of the pixel behind it; the "Download .html" button then saves a standalone HTML file with the colours baked in. The Copy button still gives you the plain characters, since that is what is useful to paste into code.

It measures the range of brightness your image actually uses and stretches that range across the whole character set before mapping. It matters far more here than in a photo editor, because a ramp only has ten to seventy steps — a photo whose tones all sit between, say, 60 and 190 would otherwise spend its whole life in the middle third of the ramp and come out as undifferentiated mush. A tiny fraction of the very darkest and very lightest pixels is ignored, so one blown-out highlight can't define the range. Turn it off if you want the original brightness values mapped literally.

No. The image is decoded, shrunk and converted entirely in your browser, and nothing is sent to a server. That is true of every tool on this site.

It should be — but only because the tool corrects for it. A monospace character is about twice as tall as it is wide, so mapping one character per pixel would render your image at half its true height. The row count is divided accordingly, and the preview and the downloaded HTML both pin a line-height that matches. If you paste the text into an editor with an unusually tight or loose line spacing, the proportions will shift a little.