> here is a general stretchblit routine
As you again use floats and/or take the burden to used fixedpoint, I think I
should explain the concept of the bresenham algorithm.
The basic problem that it solves in a very fast and elegant way is:
You want to do n times task A (like reading srcwidth times from the bob)
and m times task B (like writing destwidth times to the destination)
and you want to distribute that tasks evenly.
The simple idea to solve it follows from the fact, that n*m=m*n.
Now you take a register which is initialized to 0.
If you add m to it, whenever you do Task A, you will end up with a total
of m*n when you have done A n times.
Whenever you do Task B, you subtract n. As you do B m times, that subtracts
m*n total.
So when you have done both tasks as often as you wanted, you are back at 0.
Now for the clever idea Bresenham had: If you do A whenever the register
is negative and B when it is positive, this will result in an even
distribution.
Example: Say we want to blit a 3 pixel wide source to a 10 pixel wide
destination.
That means in the above terminology:
n=3, A=readpixel
m=10, B=writepixel
As we want to read first we define "negative" als <=0
Reg: neg? Action regchange
0 y A: readpixel reg+=m (reg+=10)
10 n B: writepixel reg-=n (reg-=3)
7 n B: writepixel reg-=3
4 n B: writepixel reg-=3
1 n B: writepixel reg-=3
-2 y A: readpixel reg+=10
8 n B: writepixel reg-=3
5 n B: writepixel reg-=3
2 n B: writepixel reg-=3
-1 y A: readpixel reg+=10
9 n B: writepixel reg-=3
6 n B: writepixel reg-=3
3 n B: writepixel reg-=3
0 y Done.
CU, Andy
--
= Andreas Beck | Email : <[EMAIL PROTECTED]> =
