On Fri, Apr 07, 2000 at 04:08:51PM +0200, Sven Neumann <[EMAIL PROTECTED]>
wrote:
> I have made the changes you proposed (only in my tree) and I think they
> should go into CVS. However the current system makes it difficult to
Me too! ;)
IMHO, the problem is just the symptom of the current alpha-handlign, which
might be intuitive for programmers, but is definitely not intuituve for
users.
For example, it was intuituve that you couldn't raise the background layer
when it had no alpha channel - to programmers.
At the moment, alpha handling is very clumsy. Adding and removing alpha is
difficult, and IMHO should rarely be explicitly necessary.
The only occasion I can think of where adding/removing alpha manually
makes sense is when saving an image - and only because the current save
architecture does not auto-optimize that.
I mean, everybody remembers the times when gif images were saved with the
number of colours you told the program, not the number of colours actually
in the image. Today you would be astonished if saving a 16 colour gif
would write the image with 8-bits.
The same could be said for alpha: If you save an opaque image with an
alpha channel you should be suprised if the resulting image would save
this superfluous alpha information.
> The export facility was never meant to produce the optimal result, but
> should give people a way to save to any format quickly without loosing
> too much information (like whole layers...).
This is true. However, it would be a cool thing to support save plug-ins with
better functionality, like fucntions to query the actual number of colours, a
function to query wether layers are really necessary, a function to tell
wether alpha is necessary and so on.
As I see it, providing this (at some stage) would save the plug-in
implementors from re-implementing the wheel again and again, or failing to
do so, and saving suboptimal images.
--
-----==- |
----==-- _ |
---==---(_)__ __ ____ __ Marc Lehmann +--
--==---/ / _ \/ // /\ \/ / [EMAIL PROTECTED] |e|
-=====/_/_//_/\_,_/ /_/\_\ XX11-RIPE --+
The choice of a GNU generation |
|
