The reason why it's too slow for realtime use is that readpixelfast and writepixelfast are slow simply because they have to transverse the videocard bandwidth bottleneck to return/set anything...
If you are wanting to do this in realtime then forget about the backbuffer() as it won't do you any good...instead create a image in system memory for this, and store your sprites in banks or arrays...writepixelfast is very fast when used within system memory...and you can then use the built in Blitz draw image routines (NOT writepixelfast) to put your system memory image "buffer" onto the frontbuffer() for the user to see...
When you writepixelfast you are not only sending a 16/32 bit color across to the video card but you are also sending out a message telling the video card you are sending a single pixel and screen cords for it's location...so in effect you are trying to transfer around 4 times the data just to write one pixel to the screen...but when useing the Blitz image drawing commands this can be done faster because the system tells the videocard how much is being transfered at once, so each pixel doesn't need the added "header" memory transfer overhead that writepixelfast does...
If you are wanting to do this in realtime then forget about the backbuffer() as it won't do you any good...instead create a image in system memory for this, and store your sprites in banks or arrays...writepixelfast is very fast when used within system memory...and you can then use the built in Blitz draw image routines (NOT writepixelfast) to put your system memory image "buffer" onto the frontbuffer() for the user to see...
When you writepixelfast you are not only sending a 16/32 bit color across to the video card but you are also sending out a message telling the video card you are sending a single pixel and screen cords for it's location...so in effect you are trying to transfer around 4 times the data just to write one pixel to the screen...but when useing the Blitz image drawing commands this can be done faster because the system tells the videocard how much is being transfered at once, so each pixel doesn't need the added "header" memory transfer overhead that writepixelfast does...