Hey, I hope I´m posting in the right forum.

So I´m making a card-driven roguelike game and I got most of the elemental mechanics to work and now I´m about to create the cards, but before I setup the system I wanted to talk with people, who can provide me some help regarding performance. The game is targeted for Android devices, so efficient memory usage is my priority. I come up with two ways to make the card system. (The maximum of visible cards is eight.)

1. Each card get its own sprite and will be included in a family with several booleans, numbers, and so on.

2. I make one sprite with frames for all cards. Information changes will be run through events.

Pros and Cons

1. Changing information on the fly. Better overview during development, but the game has to load many objects at the beginning.

2. Less objects to handle, way worse overview, much more work in the event sheets.

So, my question is, which way is the better one in the end? Would it be so bad, if I load my planned 300 cards at the beginning of the game, destroy them and then create them when I need them (when they´re drawn), or should start just with one object, so the memory just has to load the frames. Or do exist other methods I didn´t think about? From my understanding, the second way should be the more efficient one, but how much worse is the first one?

Hopefully you get my point, English isn´t my first language. I´ve read the FAQs about performance, but before I get to the meat of my game and just wanted to be sure, how my next steps should look like.