Proportional Gradient by ThreeDee
ronviers
 |
Proportional Gradient by ThreeDee http://www.filterforge.com/filters/5311.html  How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 8, 2008 3:24 am | ||||
|
ronviers
|
Excellent! Going to use it right away. Thank you.  How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 8, 2008 3:25 am | ||||
uberzev
 |
http://www.filterforge.com/filters/5311.html
 |
|||
| Posted: April 8, 2008 3:29 am | ||||
ThreeDee
 |
Oh my god, Uberzev -- you've already done this before??? I better go see if you did it the same way. | |||
| Posted: April 8, 2008 4:19 am | ||||
|
ThreeDee
|
Man, how did I miss that? Your solution is simpler and faster. How did you ever discover that threshold oddity to find the center point, anyways?
Maybe we could just combine the automatic detection of horizontal vs. vertical and have a real easy plug-and-play snippet? |
|||
| Posted: April 8, 2008 4:33 am | ||||
|
uberzev
|
If you'd like to update your filter with my simplified logic that would be great. 
(I made it so long ago I had forgotten how it works.)
Your making some totally awesome filters, and if I wasn't as conceited as I am I'd spend more time praising your work around here. 
Also you may have overtaken me as having the coolest avatar.  |
|||
| Posted: April 8, 2008 5:06 am | ||||
| Kraellin |
lol. your avatar was never cool, uber. it just thought it was cool
now that 3d has spotted yours, uber, you might both explain how and when to use these. i tried once and got lost. Ya can't sell art to artists!
Craig |
|||
| Posted: April 8, 2008 2:40 pm | ||||
CFandM
 |
Good work Threedee
Try this Kraellin http://www.filterforge.com/forum/read.php?FID=10&TID=4923  |
|||
| Posted: April 9, 2008 4:59 pm | ||||
| Kraellin |
ah, thanks, CF Ya can't sell art to artists!
Craig |
|||
| Posted: April 9, 2008 10:52 pm | ||||
|
ronviers
|
Hi ThreeDee, here is a variation that makes life easier for filter makers wanting a simple snippet for updating their old filters - it bypasses the controls.
I changed it again, this time I just piled everything on top - less confusing and easier to work with. It would be nice if something could done to make it faster. Thanks for the great snippet. Proportional Gradient var b.ffxml How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 20, 2008 8:46 am | ||||
|
ThreeDee
|
Hi Ronviers,
Great, that looks very handy for inclusion in other filters! TD |
|||
| Posted: April 25, 2008 6:34 am | ||||
|
ronviers
|
Thanks for the new snippet! I will post back with some numbers a little later.
Sure thing. Snippets are for teaching and I like the way you lay them out and explain them with comments. How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 25, 2008 11:49 am | ||||
|
ronviers
|
I am not showing any significant difference in performance. In fact the new one took ten seconds longer than the old one. Here is my test. Tick the old to select the old version.
This probably a result of my test. Stretch by Corners jcompare old_newa.ffxml How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 25, 2008 1:50 pm | ||||
|
ronviers
|
Here is something I was thinking of. Here I shrank the frame down to tiny then expanded it back to normal after it does the calculation. It served only to make it much slower.  I'm really not good at this kind of thing. pg a.ffxml How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 25, 2008 1:55 pm | ||||
|
ThreeDee
|
If you were already using Uberzev's Proportional Gradient as the "old" one, the "new" one would not be faster, it just added the automatic detection of horizontal vs. vertical, which would result in some slowdown. It's a bit hard or me to tell from the stacked version what the "old" and "new" versions are, but I assume this is the case. |
|||
| Posted: April 25, 2008 5:26 pm | ||||
|
ronviers
|
Here they are spread out. Stretch by Corners scompare old_news a.ffxml How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 25, 2008 5:34 pm | ||||
|
ThreeDee
|
I took a look at the way the Stretch by Corners is put together and I have one (theoretical) idea: You may want to try the approach of making radial spokes first (that all originate from the center -- I believe uberzev or someone has one of those) and then, at the end, use offsets to place the center at the corners and stretch the image to the correct proportions. There is a limit to the maximum relative difference in the width and height of the image this way (either 2:1 or 4:1), but it may be faster. Or it may not. But it's another idea. | |||
| Posted: April 25, 2008 5:48 pm | ||||
|
ronviers
|
Thanks for the suggestion, and eventually I will look into it, but I am going to leave things just as they are for now. I am happy with what I have posted. How do I look? You look ready... [Kill Bill] |
|||
| Posted: April 25, 2008 5:57 pm | ||||
|
ThreeDee
|
Christ, I still don't recognize them due to the arrangement. Is the "old" one my first one and the "new" one the one that uses Uberzev's gradients with my automatic horizontal vs. vertical? In other words, are these my old and new versions? |
|||
| Posted: April 25, 2008 5:58 pm | ||||
|
ThreeDee
|
Okay, that's good. The filter is just fine as it is. I thought you might still be in the search for potential ways to improve speed. | |||
| Posted: April 25, 2008 6:01 pm | ||||