Questions

Printscii redraws a photo in the characters of a retro computer font (UNSCII, IBM VGA, Amiga Topaz or C64) and can blend the result with the original photo.

Profile pictures, prints and posters, images for articles and newsletters, and album covers. The profile picture maker and the embed cover the first and third.

Yes. Photos are converted in your browser and are not uploaded to a server. The last photo you opened is kept in your browser's storage so a reload can restore it. Photos sent from a terminal are the exception: they are converted on the server, then discarded.

Yes. curl printscii.com/fox plays a running fox in characters (curl printscii.com lists the others), and curl -F [email protected] printscii.com turns a JPEG or PNG into coloured text. Add ?cols=120 to set the width.

The photo is divided into a grid of cells. For each cell, the graphics card compares the cell with every character in the font and picks the character and two colours that reproduce it most closely.

It does not use generative AI, and there are no prompts. Characters are chosen by a fixed matching method, so the same photo and settings always give the same result. The optional subject isolation uses a small segmentation model (MODNet), run in your browser, to find the person in a photo.

It reads JPEG, PNG, WebP, GIF and HEIC (iPhone photos). The free download is a JPEG up to 1200 pixels wide with a small printscii.com credit in the corner. A paid unlock gives full-resolution JPEG and PDF.

Images you make are yours to use. For licensing, integrations or bulk use, email [email protected].

Not yet. For licensing, white-label use or an API, email [email protected].

Email [email protected].

Chrome, Edge, Firefox and Safari, on desktop and mobile. Printscii needs WebGL 2. Large images convert more slowly on phones, and subject isolation is slower in Safari.