Another option to consider (although not for my first prototype run), is to upgrade to a larger MCU.
I was thinking of either the AT90USB1286 or the ATmega1284P.
The AT90USB1286 is basically a larger version of the ATmega32U4 and would retain the USB functionality while upgrading the flash to 128 kB, the RAM to 8 kB and the EEPROM to 4 kB. It also provides a bunch more pins that would provide for more expansion options. It is roughly double the price of the ATmega32U4.
The ATmega1284P is only slightly higher cost than the ATmega32U4 but doesn’t have USB functionality, meaning an extra (inexpensive) USB-to-serial (CH340G). In addition to upgrading the flash to 128 kB, and the EEPROM to 4 kB, and giving some extra expansion pins, it upgrades the RAM all the way to 16 kB!
Obviously, either of these options means it would no longer be binary compatible with the official Arduboy .hex files.
Also, it would probably require a board option that sets a pre-processor variable so people could still keep their games compatible with the official Arduboy (with reduced functionality) while still targeting the larger device with extra graphics, sounds, and/or levels.
Thoughts? Is it worth it?