QBlit Unqueue width

[Nightshft]

I'm blitting a few shapes on a Bitmap (320x256x3).
Shape 0 has a size of 84x66 and is loaded with LoadShape.

Blitting in general and qBlit works well, but when I'm erasing the previous Blit with

unQueue #queue, #sourcebitmap

unQueue will overwrite an area of 112 width.

I don't want this because it overwrites a different blit on the right side that should stay there.

Expected width of the unQueue:

84/16=5.25 so I'd have expected 6x16=96 as width
Does anybody know why it is 7x16=112?

Is there any way to make unQueue overwrite a smaller area (less width)?
I guess no, which means I have to solve it otherwise, which is doable, but I just wanted to discuss this anyway.

PS: IDE is Amiblitz 3

[Daedalus]

You're on the right track with multiples of 16, as all blits are carried out on that basis, but I *think* the destination bitmap is also blitted in multiples of 16, which would explain what you're seeing. So for example, if you've a 10x10 shape and you blit it at 5, 0, it will erase up to 16x10. But, if you blit the same shape at 10, 0, it will erase up to 32x10.

Try blitting at a range of X values and see if the areas erased align with each other, which would confirm that this is the issue you're seeing.

[Nightshft]

Brilliant! That's it!
Just tried it.

When the shape is blitted to x-pos 63 instead of 64:
The unqueue block
- starts 16 px more to the left
- ends 16 px more to the left
Top explanation, thx!

Now I have peace of mind and can code something that works in spite of that (bBlit or maybe qBlit both shapes after the unQueue)
Thx

(PS: Another reason to start the thread was just for everybody else wondering now or in the future.)

[Daedalus]

If it's doable, what about keeping a separate "clean" bitmap and using that with UnQueue to replace the blit instead of simple erasing it? That will take more RAM, but is faster than using Buffered Blits. It won't solve the issue of overlapping blits though, which would have to both be blitted in the same frame every time...

[Nightshft]

l'm already doing this

unQueue #queue, #sourcebitmap

Btw, afterwards I tried bBlit and unBuffer to check out how good that works in this case, but it also overwrites adjacent area so I will stick to qBlit as - like you say - it's faster.