is alpha fixed to 8bit unsigned

[Boudewijn Rempt]

On Fri, 10 Jun 2005, Casper Boemann wrote:

> Have we  decided to fix alpha at Q_UINT8
>
> I don't think we should, but have no hard arguments, except it's never a good
> a idea to hardcode to much.
>
> On the other hand it may be very difficult to get rid of the assumption.
>

It's a hard decision indeed. I don't want to pull a Gates and say "8 bits
of alpha should be enough, nobody will ever need more", but on the other hand
I do have a hard time imaging why we'd need more. Alpha doesn't have the
same quantizing problems as colors have: alpha is not profiled, isn't subject
to lighting conditions like wide-range exr images.

But on the other hand, having a hard limit will hurt us some day. Maybe
some research is needed. Let's ask Thorsten whether he knows whether
higher resolution alpha is used today, or would be useful.

Boudewijn