----- Original Message -----
Sent: Wednesday, April 12, 2000 7:05
PM
Subject: 8-bit alpha in 32-bit
frame-buffer?????
>> Are the 8-bits of alpha in a 32-bit frame-buffer
used by game developers for anything?
32-bit video modes are the same as 24-bit video
modes. So why do they exist? Simple: they're faster, because it's easier for
the CPU to send out or read in 4 bytes (dwords) each time, instead of 3 bytes
(read one word, than one byte). So think of the alpha byte as a
padding byte - standard (without hardware acceleration) VESA modes don't use
them for anything! Of course, they waste more memory.
>> If they are used, then how are their
absence dealt with in a 16-bit (565) display?
Lets just say, that in 16-bit modes, there's no support for
alpha channels, the obvious advantage is that they're faster - each pixel only
uses 2 bytes. 15-bit color modes (not used anymore?) are ARGB, with one alpha
bit, and 5 bits for each color (1555). Just like 32-bit modes,
the
padding bit isn't used.
>> Do most modern games simply ignore this storage
facility?
>> Can they be used to do sort-independent
translucency?
>> How can this be achieved when combining alpha
info with an opaque texture completely obliterates the alpha channel
information?
I think that the 3D 'alpha' acceleration features of modern
cards do use the information stored in the alpha channel, but I'm not sure.
What I do know, is that most cards don't mind what
value the alpha channel is given (usually zero) and that you can use it
however you wish.
Thanks :-)
Tom
No problem,
Habil