8/21/2001 5:53:37 AM, cybergoth@xxxxxxxx wrote:
>NOW we'd have enough RAM to display 6 balls with Bob's
>multiplexing algorythms, so we could have 6
>almost_non_flickering Balls displayed using only one object.
>
>...but then all the repositioning needed within the
>multiplexer killed the idea again before I finished typing it...
This brings to mind an idea I've had for a while but haven't actually
tried implementing. Is it possible to use a multiple-branching kernel
via an indirect jump to time RESPx writes without having to use an
entire scanline to position an object?
What I mean is something like this. Offscreen, you do some calculation
and store into SpritePosition (value 0 to 10) and HmoveValue. You have
a table set up in ROM for BranchTable (values for the addresses for the
Kernel# labels). Then you do this during the kernel:
LDY SpritePosition
LDA BranchTable,Y
STA KernelBranch
JMP (KernelBranch)
Kernel0 STA RESP0
do other processing like six playfield writes
LDA HmoveValue
STA HMP0
JMP EndLine
Kernel1 do playfield write #1
STA RESP0
do playfield writes 2-6
LDA HmoveValue
STA HMP0
JMP EndLine
Kernel2 do playfield writes #1-2
STA RESP0
do playfield writes #3-6
LDA HmoveValue
STA HMP0
JMP EndLine
[etc]
Kernel10 do playfield writes #1-6
LDA HmoveValue
STA HMP0
STA RESP0
EndLine STA WSYNC
STA HMOVE
The Kernel# labels vary the position of the STA RESP0 by equal to
or less than 15 pixels between them, so you can use HMOVE to get
the object to any position on the screen.
So you've positioned your object by using lots and lots of ROM, but
only 28 cycles all told (including the last HMOVE). Much less than using
a full 76 cycle scanline.
Is this feasible? Has it already been done anywhere?
-
Archives (includes files) at http://www.biglist.com/lists/stella/archives/
Unsub & more at http://www.biglist.com/lists/stella/
