Yep, I don't think __alpha is defined when building on more modern machines
(didn't know the history behind it, though. Thanks Jonathan)
Ben, thanks for the credit, but this is not what I meant in the original
post.
What I was trying to say was that the original issue Ron posted was due to
Merry Xmas its fixed. The weird thing is that fixing was to turn off crop
to format in the blur11 then turn it back on. ? So I don't know.
strange. maybe it was holding onto the format from pre crop some how for
whatever reason.
set cut_paste_input [stack 0]
version 6.2 v3
push
Oh I also hooked and unhooked a constant at one point. Maybe sparkle needs
an input size?
Randy S. Little
http://www.rslittle.com http://reel.rslittle.com
On Thu, Jul 21, 2011 at 01:12, Randy Little randyslit...@gmail.com wrote:
Merry Xmas its fixed. The weird thing is that fixing was to
Hi Randy,
Still wrong on my end... :-(
Played with Blur11 - didn't seem to make any difference.
Thanks for trying.
Ron Ganbar
email: ron...@gmail.com
tel: +44 (0)7968 007 309 [UK]
+972 (0)54 255 9765 [Israel]
url: http://ronganbar.wordpress.com/
On 20 July 2011 18:12, Randy Little
The Constant node doesn't seem to help me either.
The problem started with some other image, not the Sparkle. I just used that
so I'll have something to post here.
Ron Ganbar
email: ron...@gmail.com
tel: +44 (0)7968 007 309 [UK]
+972 (0)54 255 9765 [Israel]
url:
did you try hooking and unhooking or leaving hooked a a black constant.
Its 2 am I don't know what buttons I pushed to make it work. SORRY. BUT
Randy S. Little
http://www.rslittle.com http://reel.rslittle.com
On Thu, Jul 21, 2011 at 01:16, Ron Ganbar ron...@gmail.com wrote:
Hi Randy,
Is the gamma (curve) all the way up on 5 and size on 2 and you're not
getting a square?
Ron Ganbar
email: ron...@gmail.com
tel: +44 (0)7968 007 309 [UK]
+972 (0)54 255 9765 [Israel]
url: http://ronganbar.wordpress.com/
On 20 July 2011 18:19, Randy Little randyslit...@gmail.com wrote:
Its like the bbox is not resizing even though it shows it bigger right? it
was working for a bit but now I can make it work again. I get what looks
like the image hitting the bbox that exist if you turn off the crop
Randy S. Little
http://www.rslittle.com http://reel.rslittle.com
On Thu,
Hi Ron,
I think you're hitting a float precision limit there, or rather a
limit introduced by the way most (all?) computer graphics software do
filtering operations by sampling a square area. But I'll put a big
question mark on the second assumption since I don't know the
specifics.
If you take
The thing is ivan if you look at it it hits exactly the edge of what the
bbox was before the crop. Is that just coincidence?
Randy S. Little
http://www.rslittle.com http://reel.rslittle.com
On Thu, Jul 21, 2011 at 01:50, Ivan Busquets ivanbusqu...@gmail.com wrote:
Hi Ron,
I think you're
yeah, strange.. the effect happens actually on any simple
blur... looks like the gaussian blur is not really spreading
uniformly, but more boxy. boosting the quality level
doesn't help either. ironically, if you switch the filter to
box it gets more like a circle.
also, if i replicate the same
got ivan's mail late, but i think he's spot on..
you can try with bigger bbox and ridiculous high quality
level and it still is the same problem.
if you really need such a high gamma, you could try blurring
the image at different rotation, then rotate back. something
like:
set cut_paste_input
Actually, I think the reason it looks correct in 8 and 16 bit is
because values hit 0 before they reach the edge of the square sampled
area.
Which would be the same as clamping some of the very low end in a float image.
The problem is, in the case of a radial gradient like this one, that
the
13 matches
Mail list logo