Re: [stella] Gunfight 2600: Descissions had to be made

Subject: Re: [stella] Gunfight 2600: Descissions had to be made
From: Manuel Polik <manuel.polik@xxxxxxxxxxx>
Date: Tue, 06 Mar 2001 10:03:52 +0100
Erik Mooney wrote:

> >The next step I'll work on is the weapons system. Continuing with the
> >playfield right now doesn't make sense, since I just don't now how many
> >RAM is left after I'm done with the weapons. (I need at least need 2
> >bytes of X/Y coordinates, plus one byte direction + one byte lifetime
> >per bullet, so handling 12 bullets for example would eat up a minimum of
> >48 bytes RAM...)
> You don't need 8 bits each for direction and lifetime.  The direction is
> only 8-way right?  

Uhm... well, normally yes. But I'm thinking about 2- and/or 3-way guns,
which would introduce some more angles to take care about.

> That's only 3 bits for direction.  I'm not sure how
> long the bullets last, but 6 bits gives you over one second at 60 fps.

That's something I have in mind. The trick I'll do is: Decrementing the
bullet life-time not every frame, but only every 8th or so.
> Can't quite compress it to one byte per bullet which would be convenient,
> unless you have half-second bullet lifetimes, or run the bullet processing
> only every 2 or 3 frames 

With the little trick above, I can process the bullets every frame and
still have a big enough lifetime value. But I can display only three
bullets per frame, so I have to flicker them obviously. I just have to
test to which degree this is acceptable, 12 bullets might already be too
much, so I might come down to a total of 8 or 6 (which would then render
3-way shots impossible... at least for both players at once :-))

> I'm not quite sure of the behavior of the bullets - do they bounce or
> anything?  If not, you're storing redundant information somewhere in the
> X/Y/lifetime bytes.

Yes, they are intended to bounce, from obstacles as well as from screen
> >Please comment on the new demo. I hope it works fine for both players, I
> >just had no second player to test if the players can interfere with each
> >other during some weird Joystick movements :-)
> Not sure what comments you want to see, but it does look good.  

Exactly the comments you just sent. Thanks!

> Hope
> there's time to load some playfield data instead of storing what looks
> like the scanline counter.

Currently there's still 24 NOPs in the second scanline, which should be
enough for displaying some fine playfield obstacles. At least I hope so.
I just need some 20-30 bytes of RAM left in the end to make 'em
shootable. We'll see how it works out :-)


Archives (includes files) at
Unsub & more at

Current Thread