interleaved bitmaps and blitting

[h0ffman]

Quick question.. I'm thinking of using an interleaved bitmap instead of the usual bitmap method of each bitmap one after the other.

The idea is that I should be able to blit all 5 planes in one blit rather than having to add plane size to the pointer on each blit.

Anyone had any experience with this? Also is there an iff converter which will output the raw image this way?

By my calculations the plane pointers in this case would be.. (for a 320 pixel width screen)

plane 1 = source
plane 2 = source+40
plane 3 = source+80
plane 4 = source+120
plane 5 = source+160

and the modulo would be 120.

[h0ffman]

Seems to be the day for it today!! I found that PicCon can save interleaved raw bitmaps, chucked in the new values above and all went it perfick

[mc6809e]

Blitting to multiple planes in with one blit is wonderful, especially if the dimensions of the pixel block are small. A case where blitting is only slightly faster than the CPU becomes a huge win with interleaved bitplanes.

[Codetapper]

If you can spare the extra chip memory to store the mask once for each plane - yes!

[nandius_c]

I'm using interleaved bitplanes for my game. I create the interlaved data with IFFConverter which is very easy to use and can also generate palette and sprite data. As Codetapper said, I'm storing the mask once for every bitplane.

[mc6809e]

Originally Posted by Codetapper

If you can spare the extra chip memory to store the mask once for each plane - yes!

That's true. Not a big problem if only small bobs use the technique, of course.

I've been wondering, though: is there any extra use that can be gotten out of what is normally the mask in this case? Some special effect, possibly using some unusual LF? That way the extra copies of the mask aren't wasted.

I'm thinking something like semi-transparency.

If the palette is setup correctly, then semi-transparency can be done with something like LF=$B2.

With that LF, whenever the bob is written on top of pixels not masked out, the underlying pixel is inverted. With the right palette, some colors can become the shaded versions of the others when the background images has its bits flipped when merged with the bob.

[Mrs Beanbag]

even with only one mask and separate blits for each bitplane to draw a bob, a single blit could still be used for restoring background graphics.

[DiabloV]

Hello

After a lot of headaches , I've finally managed to come up with a mask blending routine.

Here's an example.

Don't hesitate to tell me if I'm doing something wrong.

Now, that's all well and good, but I need to know what I do when I move the bob to refresh what was behind it?

What's the best technique?

https://mega.nz/file/cLlgVaLb#uG8iZ3...ZCPgbGV8MF6dIc

[buzzybee]

I use triple buffering for Reshoot R and Reshoot Proxima 3, as this allows me to blit interleaved objects without having to store background bitmap data for each blitted objects, each frame.

The digest of how it works:

• FrameBufferA: This bitmap contains background data with blitted objects combined. This buffer is on display

• FrameBufferB: At the start of the frame, this contains background data and blitted object. First step: Restore background to empty state by copying clean background data from BufferC. For efficiency you would only partially restore background where blitted objects were drawn, using stored object blit feed data from the former blitter object process. After restoring, BufferB contains a clean background image. BufferC is now available for blitting objects. While blitting objects to BufferC, you would store object blit feed data (source adress, target adress, size, modulus) for efficient background restore in the next frame.

• FrameBufferC: This contains empty background state only / no blitted objects at the start of the frame

You need to rotate these three buffers with each frame: BufferA becomes BufferB, Buffer B becomes BufferC, BufferC becomes BufferA and so on. Rotation is achieved non-expensive by swapping memory pointers.

This works with static and scrolling environments. It's fast but heavy on memory as you need three framebuffers. Therefore it's well suited to AGA-chipsets with 2 MB Chipmem.

If speed is not so much a concern, you can streamline the background restore process by simply copying the complete bitmap instead of partially restoring it.