Re: [stella] single line kernels and horizontal position calculation

Subject: Re: [stella] single line kernels and horizontal position calculation
From: Manuel Polik <manuel.polik@xxxxxxxxxxx>
Date: Mon, 12 Feb 2001 10:27:18 +0100
Glenn Saunders wrote:
> It looks like no matter how fast the horizontal calculation routine that it
> winds up consuming too much time on the scanline to scan out the results on
> the same line, hence the traditional 2-line kernel utilizing VDEL.

Yes, you'll always loose lines for positioning. At the moment you're
hitting RESP0,X, it's already too late to display anything on that
position, in that line.
> Take a game like Atlantis, for instance.  It only draws one sprite going
> across the screen on any given scanline, right?   And it enforces vertical
> separation so that there is time to calculate new horizontal positions
> between ships.  But other games have more free-form displays which would
> have objects moving vertically through the "compute scanlines".

Uhm... maybe, but this vertical moving sprite is then set at the very
beginning of the screen and doesn't have multiple horizontal positions
too, or? (At least not in the same frame, I'd think)
> I am thinking that the only way to solve this is to use tables to
> precalculate everything.
> Does anyone have any tables handy for this purpose?

> I may need to do something because I am going to have to adjust the missile
> positions on every line AND rewrite the playfield.  

That won't work. Maybe you can shift it +- 7/8 pixels from one line to
another, but not totally reposition it every line.

> I could precalculate
> the missile offsets during VBLANK but I may be short on the RAM needed to
> store these values.

Uhm... now you're confusing me :-)
Could you be just a bit more precise, what exactly you're going to do? 
(A missile offset for every line would of course be up to 192 bytes RAM)


Archives (includes files) at
Unsub & more at

Current Thread