It may and should be an optional alternative mode.

I suggest to input numerator and denominator of the ratio, like we would width and height, in a field each.

Currently v1.99 the ratio is input as a two-decimal-digits decimal number.

I argue that it is too inaccurate.

But changing the length of the mantissa is only shifting the problem to larger pictures especially rational non-decimal ratio such as 2/3.

Three decimal digits would be quite an upgrade for 1000 pixels scale, but i would rather not assume the user only handles small pictures.

I think numerator and denominator fields are more convenient because :

- Computing the ratio from them is still easy for the program.
- The program can compute a multiplication then a division instead of expecting a very rounded, quite imperfect ratio from an early division.
- When it is not already well known, i think it is more convenient for the user to input two integers than computing the ratio themselves.

Compare with inputting 1.33 or 1.78 on a 2000 pixel image. The rounding becomes about 6 pixels in the updated dimension. Not everyone will neglect it in every case, even if it sounds pedantic to most.