As a workaround, you could put a tiny offset to one of the RGB
channel values, so the Constant Color never outputs (0,0,0,0), or add
some logic to check for zero output and turn off the sprite
accordingly. I think you can check for null images using JavaScript,
and output a boolean that will turn the sprite or billboard on/off.
I must admit, I CAN see the logic of an image that will never be seen
being disabled. I mean, in a sense, there's no point using resources
trying to render an image that is 100% transparent, and will
therefore never be visible. Maybe I'm missing the point though.
a|x
On 27 Oct 2009, at 16:41, George Toledo wrote:
I thought I had been clear that the bug was in the Constant Color;
I didn't mean for anyone to wind up spending time looking at the
sprite. ("In short, if the color value of the Constant Color is
set to 0,0,0,0, it no longer outputs an image; it shuts down.)
It didn't really make sense to me to think the Sprite was broken,
since it can be fed a 0,0,0,0 from other sources (like a CI
Kernel), and works correctly.
So, I certainly didn't think it was in the Sprite. That's what I
was trying to say... pure alpha shuts off the Constant Color from
producing output image. This reveals the cause of a mystery bug
that has plagued me for awhile.
I just wanted to make sure it happened on other people's systems,
and that there wasn't any kind of valid reason for that to happen
that I'm overlooking. I would certainly be curious if this doesn't
happen outside of QC, which would isolate it to being a CI bug, as
you suggest. I'm glad to see it's not system specific, and I
appreciate others taking the time to give it a go and reconfirm
that for me.
A slight aside:
It's slightly unnerving to see bugs get deferred as being CI bugs,
or some other department's bug, etc., and then I have to go on the
CI list, or file it as a CI bug, or the driver department or whatever.
I understand that line of thought, but I would humbly say that it
is incorrect. If I'm using QC, and this happens, it's a QC bug, and
the QC dept. needs to handle this, or tell the Core Image dept., or
whoever it is that needs to fix the bug. It's a QC bug, because QC
uses CI. If it didn't use CI, then it wouldn't be a QC bug, because
I would never see the bug to begin with. They should get up, and go
walk over to the Core Image office, or whatever it is, instead of
deferring it back on the user, or waiting for people on-list to
figure it out for them.
I know that comes off as matter of fact, but I don't mean it to.
I'm just used to a mindset of people taking ownership of a problem
from cradle to grave. For example, If I called Apple up, and got
"Joe Shmo" on the phone, I would expect them to go report this to
whoever needs to know how to fix it, even if their job was totally
unrelated. In this same way, I don't have an expectation of having
to ask about this on the Core Image list, to have to isolate it to
being CI, or write it up as a CI bug. I will most definitely write
this up as a QC bug, since it happens in QC.
If someone in the Core Image department has to actually fix it
instead of someone who works on QC, this should be transparent to
me (pun intended).
-George Toledo
On Tue, Oct 27, 2009 at 10:41 AM, Christopher Wright
<[email protected]> wrote:
Is it not expected that the Sprite in the attached composition would
"disappear", when the Constant Color patch is set to pure Alpha black
Yeah, that's always been a real pain. I think it is the actual
CIFilter that causes the problem (i.e. I have had similar issues using
the CIFilter directly independent of QC).
I did some testing -- Constant Color produces no output image when
the input is 0,0,0,0. so the cropper produces nothing, and hands
off nothing to the sprite, who then operates as if no image is
attached at all -> default white sprite.
So Andrew's correct, it's the Constant Color / CIFilter that's
broken, not the sprite. (bug still stands though, just clarifying
where it lives)
Andrew, do you have a sample app that demonstrates the bug outside
of QC? if so, that's a CoreImage bug.
--
[ christopher wright ]
[email protected]
http://kineme.net/
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list (Quartzcomposer-
[email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/the_voder%
40yahoo.co.uk
This email sent to [email protected]
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/archive%40mail-archive.com
This email sent to [email protected]
