It's probably worth mentioning that, for flipping sprites during program execution, you don't have to implement those algorithms yourself. Just use the built-in flipH() and flipV() commands.

The only restriction is that you can't flip it twice without a copy, but the main advantage of pre-flipping the sprite before compiling is the speed gain because flipping a sprite takes some time.  It also makes sprite animation easier since you can reference sprites by a simple offset instead of extra conditionals for which direction to flip.

Of course the disadvantage is that you will have 2-4 times as many sprites that needed flipping than before, but when you only have a few sprites that need flipping (like a main character for instance) its probably smaller than including the flip routine in code.

There is also a hybrid compromise approach when you have a ton of sprites that need flipping, but you still want the fast speed without the extra size.  Just use the flip command to flip all the sprites in the beginning and copy them to a free ram location when the program loads.  Then, you can use that location with the speed of the preflip method, but with the program size of the runtime flip (minus the decompression routine).