The alternative is to remember the format the image was loaded from, track all of the features of each individual format, note when the user changes something that can't be saved in the original format, then before the user "saves" in yet another format decide when to pop up a dialog that describes how these featuresets overlap and alternate ways to "save" that might preserve more or fewer of them.

All of this to avoid saying: "You're converting your working copy into a lossy format. You might lose things."