I see the problem you're describing and I agree it would be a good problem to solve. Here's the problem: The size tools compress to isn't arbitrary - it's based on the algorithm, how compressible the data is, and how much loss of quality the user is willing to accept. The algorithm is generally fixed - as a user you just use the algorithm of the tool you're using (or file type). The compressibility of the data is defined by the data you're using. A bland picture with few colors is more compressible than a very rich one, for example. Acceptable quality is a bit fuzzy, and this is where the problem is. For an image, is cropping better than losing color depth? Or resolution? Are you concerned about image scaling - how it looks on an iPad vs a big PC gaming monitor? So, the problem you want to solve requires lots of input from the user, about topics the user either cares not a whit about (until the picture comes out bad) OR the user cares passionately, and is already using something like Photoshop or GIMP and doesn't mind the extra work.