I dunno what you're doing. :-)
When I draw my graphics, I clear the backbuffer then redraw the whole thing, then I flip.
When I flip, the backbuffer becomes the frontbuffer. The backbuffer() pointer now points to the buffer that used to be the front buffer. It is now the buffer I should write to.
I also do triple buffering as suggested above, but only because drawing to an offscreen image is much faster than drawing to the screen. If I am doing alpha blending and other blending operations in 2D I need this speed. I lose some speed overall because of the copy I have to do at the end, but because I can draw pixels many times more it makes up for it when I am doing lots of blending.
As for 3D, I can't see what you could be doing wrong there, unless you are doing 3D over a blank background, and do not have cameraclsmode set to clear the color and zbuffer. But if that were the case you'd have 50 copies of the cube onscreen and not just two flipping back and forth.
Unless you were flipping and not calling renderworld every time you flip. Then you would be flipping an older image back to the visible screen.
I didn't see anyhting in your 3D demo which looked like you were doing any of that though. It looked normal to me, other than the afterimages normally caused by low framerates and high contrast.
When I draw my graphics, I clear the backbuffer then redraw the whole thing, then I flip.
When I flip, the backbuffer becomes the frontbuffer. The backbuffer() pointer now points to the buffer that used to be the front buffer. It is now the buffer I should write to.
I also do triple buffering as suggested above, but only because drawing to an offscreen image is much faster than drawing to the screen. If I am doing alpha blending and other blending operations in 2D I need this speed. I lose some speed overall because of the copy I have to do at the end, but because I can draw pixels many times more it makes up for it when I am doing lots of blending.
As for 3D, I can't see what you could be doing wrong there, unless you are doing 3D over a blank background, and do not have cameraclsmode set to clear the color and zbuffer. But if that were the case you'd have 50 copies of the cube onscreen and not just two flipping back and forth.
Unless you were flipping and not calling renderworld every time you flip. Then you would be flipping an older image back to the visible screen.
I didn't see anyhting in your 3D demo which looked like you were doing any of that though. It looked normal to me, other than the afterimages normally caused by low framerates and high contrast.