Bouncing Square Clipping

[tda0626]

Thanks, Steve. I will have to read that book.

Here's the link to that chapter (Bit-Plane Animation).

Thank you!

[_japheth]

Thank you. Is the technique you are describing called double buffering?

No - I guess double buffering is closely related to "page flipping" ( changing the start address of the "visible" video memory ).

When you allocate memory with DOS function AH=48, you specify the number of pages of memory needed as 1000h. Why did you specify that and not 4000 decimal, which is 64k/16?

You're right, 4000 would have been the exact value - 1000h wastes 1536 bytes.

Btw., I made some timing tests: your code, with the vsync wait deactivated, needs 86 ms for 600 loops - so < 150 microsecs for one loop - for my understanding that's far too fast to cause "flickering" troubles. Would be interesting to see what's displayed with a true CRT monitor attached - regrettably I don't have one.

[tda0626]

Thank you. Is the technique you are describing called double buffering?

No - I guess double buffering is closely related to "page flipping" ( changing the start address of the "visible" video memory ).

When you allocate memory with DOS function AH=48, you specify the number of pages of memory needed as 1000h. Why did you specify that and not 4000 decimal, which is 64k/16?

You're right, 4000 would have been the exact value - 1000h wastes 1536 bytes.

Btw., I made some timing tests: your code, with the vsync wait deactivated, needs 86 ms for 600 loops - so < 150 microsecs for one loop - for my understanding that's far too fast to cause "flickering" troubles. Would be interesting to see what's displayed with a true CRT monitor attached - regrettably I don't have one.

Is that good for a line draw routine? Never benchmarked it. I am sure it could use a little bit of optimization.

When drawing a filled in rectangle or square, is there a preferred method? The way I draw it now is line by line, top to bottom.

[NoCforMe]

When drawing a filled in rectangle or square, is there a preferred method? The way I draw it now is line by line, top to bottom.

Well, assuming you've drawn a contiguous border, you could use a flood-fill routine, I guess, but that might actually be slower, as well as requiring recursion.

The advantage of using flood-fill is that it will fill any fully-enclosed shape, regular or not.

[_japheth]

Is that good for a line draw routine? Never benchmarked it. I am sure it could use a little bit of optimization.

Optimization is usually good, but:

• 86 ms for 600 loops without vsync synchronization
• 10,000 ms for 600 loops with vsync synchronization
• conclusion: your prog spends more than 99% of its time waiting for vsync

So it's perhaps a bit moot here...