You could count non-zero pixels in rows and columns with something like:

cols, rows = (x[1] > 128).ifthenelse(1, 0).project()

That's taking the green band, finding pixels greater than 128, and setting them to 1, with other pixels as 0. Then you project that to get the number of non-zero pixels in each row and column.

Yes. project is outputting only 0 or 1 (like the input). That won't work for me.

It should work. I see:

$ python3
Python 3.12.3 (main, Sep 11 2024, 14:17:37) [GCC 13.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import pyvips
>>> x = pyvips.Image.new_from_file("/home/john/pics/k2.jpg")
>>> cols, rows = (x[1] > 128).ifthenelse(1, 0).project()
>>> cols.max()
1304.0
>>> 

So one column has 1304 green pixels which are greater than 128.

libvips uses 0 for false and 255 for true, so you could also write:

>>> cols, rows = ((x[1] > 128) >> 7).project()
>>> cols.max()
1304.0
>>> 

Which is a bit quicker, though maybe harder to read heh.

x.ifthenelse(1, 0).project()

seems to have given some useful output. Thanks for the help, and the excellent mindbending library.