You might want to pick up a copy of the OpenGL extensions guide,
which discusses ARB_fragment_program in depth.
I'm sure there are other references available - but this is a good
one and discusses semantics and limitations of fragment
program behavior. For those kernels executed on the GPU, these docs
are relevant. For those kernels executed on Altivec,
I'm sure they are simply using vector float as the native data type,
which would imply uniform behavior across all the channels.
On Jun 24, 2005, at 8:35 AM, Nick Nallick wrote:
I'm building a multi-pass CoreImage filter that invokes two kernels
in series. The second kernel depends on the fact that the output
from the first kernel may contain floating-point pixel values
greater than 1.0. This seems to work fine for the RGB components
but alpha seems to behave differently. It appears the alpha
component is being clamped to a smaller value than the others
(perhaps 1.0). Is alpha indeed treated differently than the RGB
components, and if so is there anything I can do to stop this from
happening?
Thanks,
Nick
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartz-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/quartz-dev/gk0001%
40digigami.com
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartz-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartz-dev/email@hidden
