Working with normalized coordinates is convenient when you want to make your PixelMath expressions independent on specific image properties.

You can also work in pixels. Excerpted from the function documentation provided in the Expression Editor dialog:

`w( [image=$T] )`

Width in pixels of an *image* (invariant subexpression).

`h( [image=$T] )`

Height in pixels of an *image* (invariant subexpression).

`x()`

Current horizontal (X-axis) coordinate in integer pixels.

`y()`

Current vertical (Y-axis) coordinate in integer pixels.

Note that PixelMath is case-sensitive, so x() and X() are different functions (the latter evaluates to the current X-axis coordinate normalized to [0,1]). Note also that in formal documentation descriptions (metalanguage), square brackets denote optional items. So you can say just w() and it will be equivalent to w( $T ). Or you can say w( foo ) to get the width in pixels of an image with the foo identifier.

For completeness, here are the corresponding functions for normalized coordinates:

`X()`

Current normalized horizontal (X-axis) coordinate, in the [0,1] range.

`Y()`

Current normalized vertical (Y-axis) coordinate, in the [0,1] range.