Re: [e-users] e16.8 in CVS

[Kim W. <ki...@wo...>]

I'll favor "look good" over "physically correct" in this case :)

The code removal as suggested doesn't fix the problem entirely, as you
may have seen, but can speed up shadow image calculations if the region
isn't used anyway.
The proper fix involves properly repainting the window stack taking into
account desired clipping regions, like in clipping away the shadow image
part that lies under the window.
I think that should be fixed now.

/Kim

Viktor Kojouharov wrote:
> Actually, if you want it to be physically correct, the shadow should
> be seen through a transparent window. The question here is, would it
> look pretier without the square part of the shadow. Then again, if it
> should be removed, then it should be totally removed, since the center
> square part will never be seen when the window is opaque. So the
> fastest way is to delete lines 746-752 in ecompmgr.c and see what
> happens :)
> 
> On Tue, 04 Jan 2005 20:52:27 +0100, Kim Woelders <ki...@wo...> wrote:
> 
>>Viktor Kojouharov wrote:
>>
>>>kwo, this is a patch that someone created for xcompmgr, that's
>>>supposed to de-uglify shadows for translucent windows. basically, the
>>>shadow opacity is modified by a coefficient, dependant on the window
>>>opacity. that way translucent windows look nice, not like dirty snow
>>>:). Can you implement it into ecomp?
>>>
>>
>>Done. However, shadowed windows don't look quite as they should yet
>>(the shadow is seen through transparent windows) because I have messed
>>up some things along the way :)
>