Re: [stella] Jumpman: Follow the leader - No good!

Subject: Re: [stella] Jumpman: Follow the leader - No good!
From: Manuel Rotschkar <cybergoth@xxxxxxxx>
Date: Mon, 11 Oct 2004 22:12:58 +0200
Hi Christopher!

Assuming Left/Right/Up/Down/Jump that's 5 bits + 1 byte for # frames (max 256 frames at a shot). But you can cheat, you'll never have Up+Down+Right+Left so 1111 can be jump and you're down to 1 nibble for the directions + 1 byte for # of frames.

Hm... Jumpman doesn't recognise the diagonals, so with


nothing
up
down
left
right
up+fire
left+fire
right+fire

I could probably decode a frames input in 3 bits.

The problem with storing the # of frames extra is, is that you're totally "user" dependent then, unlike more linear storage. Still it might work.

So you can store 2 directions changes in 3 bytes. If you can set aside even 20-30 bytes you can get a pretty complex set of movements.

Spinning your idea further, I could use the the first three bits of a byte for decoding the direction/action and the other 5 as counter. If a direction lasts longer than 32 (movement-)frames, it'd just start a "new" byte.


I dunno if that's enough, or you have that much RAM but I thought I'd mention it....

Might work. The level already requires 30 RAM bytes for structure buffering. Since I'll try to get the regular framework running with less than half of the RAM, there might be some 30 bytes left. Still if someone wiggles the Joystick like a Decathlon player I'm busted in a second or two...


Greetings,
	Manuel

Current Thread