imgcurl

inspired by this tweet.

See any image on the web with a single curl

Quickstart:

Run this in the terminal:

curl -v 'imgcurl.samwheating.com?url=https://shortnur.pro/beeg'

You'll need to have curl installed.

Usage:

curl -v imgcurl.samwheating.com?url=<ANY_DIRECT_IMAGE_URL>

You can also specify the width (in # of characters) with a width=<number> query param. The default is set to 150.

ex: curl -v imgcurl.samwheating.com?url=<ANY_DIRECT_IMAGE_URL>&width=100

works with most image formats (transparent PNG is ideal).

These headers get blocked by most production servers (gunicorn etc), and also broke Google Cloud Run.

How does this work?

Most modern terminals allow the use of ANSI escape codes to modify the colour of text. By breaking down an image into an array of pixels and converting each one into the nearest colour in the 216-colour ANSI palette.

ANSI escape codes allow for 256 different colours:

  0-  7:  standard colours (as in ESC [ 30–37 m)
  8- 15:  high intensity colours (as in ESC [ 90–97 m)
 16-231:  6 × 6 × 6 cube (216 colours): 16 + 36 × r + 6 × g + b (0 ≤ r, g, b ≤ 5)
232-255:  grayscale from black to white in 24 steps

So by converting each RGB colour values from a range of 0-255 to 0-5, we can get a reasonable approximation of the true colour to the pixel from the 216-colour ANSI colour cube.

Each "pixel" in the image is thus represented as \u001b[48;5;<COLOR NUMBER FROM 16-231>m \u001b[0m.

From here, each pixel in a row is converted into an ANSI control character and written into a string. Each row is then added to the response as a header.