Resizing Pipeline for GPT-4 Vision

[blue-j]

ImageMagick version

7.1.1-29 Q16-HDRI aarch64

Operating system, version and so on

macOS 13.6.6

Description

I'm trying to pre-process images for use with GPT-4 Vision. The documentation states:

Images are first scaled to fit within a 2048 x 2048 square, maintaining their aspect ratio. Then, they are scaled such that the shortest side of the image is 768px long.

Here are some examples demonstrating the above.

A 1024 x 1024 square image
1024 is less than 2048, so there is no initial resize.
The shortest side is 1024, so we scale the image down to 768 x 768.

A 2048 x 4096 image
We scale down the image to 1024 x 2048 to fit within the 2048 square.
The shortest side is 1024, so we further scale down to 768 x 1536.

I can't seem to crack this. Presumably we pipe one resize into another? Any pointers would be gratefully accepted. : )

• J

[fmw42]

The following should do what you want.

magick image.suffix -resize "2048x2048>" miff:- | magick - -resize "768x768^" result.suffix

Quotes (or \ escapes) seem to be needed when I paste into my Terminal. But in a script, they may not be needed.

See geometry for > and ^ at https://imagemagick.org/script/command-line-processing.php#geometry

[snibgo]

Presumably we pipe one resize into another?

There is no need for piping. That just adds unneeded complications. Windows syntax:

magick in.png -resize "2048x2048^>" -resize "768x768^^" out.png

Bash syntax:

magick in.png -resize "2048x2048>" -resize "768x768^" out.png

[fmw42]

OOPS! Sorry, I forgot the pipe was not needed -- just following what he said. Dumb!. Thanks.

[blue-j]

thank you so much!! you're the best. : )