What does the blur value relate to?
I'm just curious... what does the blur value relate to? Is it a percentage of some max amount of blur?
- Tony
On Mon, Apr 02, 2007 at 06:46:35PM -0700, Tony Vigil wrote:
I'm just curious... what does the blur value relate to? Is it a percentage of some max amount of blur?
No, it's a non-linear (exponential?) relation that was arrived at through feedback from user testing and experimentation.
Bryce
Isn't it a little misleading to call it a % then?
I typed in 50 expecting to get something 50% blurred and got something so blurred and spread out it was basically invisible. The result from 0.5 is more like what I would expect to see as "50%" blurring.
--bb
On 4/3/07, Bryce Harrington <bryce@...983...> wrote:
On Mon, Apr 02, 2007 at 06:46:35PM -0700, Tony Vigil wrote:
I'm just curious... what does the blur value relate to? Is it a percentage of some max amount of blur?
No, it's a non-linear (exponential?) relation that was arrived at through feedback from user testing and experimentation.
Bryce
Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=D... _______________________________________________ Inkscape-user mailing list Inkscape-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/inkscape-user
On 4/2/07, Bryce Harrington <bryce@...983...> wrote:
No, it's a non-linear (exponential?) relation that was arrived at through feedback from user testing and experimentation.
No. There was only one user advocating the exponential scale, but his patch did not convince me. Anyone is welcome to try out the patch in our patch tracker; if the majority votes for it I'll go with it.
So currently, the scale is linear, and the range is explained in the 0.45 release notes:
The blur value is a percentage, with 100% corresponding to a blurring radius (standard deviation of Gaussian function) of 1/8 of the object's bounding box' perimeter (that is, for a square, a blur of 100% will have the radius equal to half a side, which turns any shape into an amorphous cloud).
http://wiki.inkscape.org/wiki/index.php/ReleaseNotes045#SVG_filters:_Gaussia...
On 4/3/07, bulia byak <buliabyak@...155...> wrote:
On 4/2/07, Bryce Harrington <bryce@...983...> wrote:
No, it's a non-linear (exponential?) relation that was arrived at through feedback from user testing and experimentation.
No. There was only one user advocating the exponential scale, but his patch did not convince me. Anyone is welcome to try out the patch in our patch tracker; if the majority votes for it I'll go with it.
So currently, the scale is linear, and the range is explained in the 0.45 release notes:
The blur value is a percentage, with 100% corresponding to a blurring radius (standard deviation of Gaussian function) of 1/8 of the object's bounding box' perimeter (that is, for a square, a blur of 100% will have the radius equal to half a side, which turns any shape into an amorphous cloud).
http://wiki.inkscape.org/wiki/index.php/ReleaseNotes045#SVG_filters:_Gaussia...
As far as explaining what the number means, I think the existing explanation you just gave of "a percentage of a fraction of the bounding box" is fairly simple and makes it easy to guess what a reasonable value would be. Making it instead into "a nonlinear mapping of a percentage of a fraction of the bounding box" would make it difficult to guess good values.
But it might be useful for the slider to behave in a logarithmic way, while still showing the % number with the same meaning it currently has. Values like 0.1 through 0.5 seem to be useful for adding different amounts of light fringe, whereas the visual difference between 20.1 and 20.5 is pretty much Nil. So the slider would be more useful if the left third mapped, say, from 0 to 1, middle third from 1 to 10, and last third about 10-100.
Very nice feature though, this blurring. Does it by any chance use this: http://incubator.quasimondo.com/processing/fast_blur_deluxe.php ?
Bill Baxter wrote:
... Very nice feature though, this blurring. Does it by any chance use this: http://incubator.quasimondo.com/processing/fast_blur_deluxe.php ?
No, the current release just uses a normal Gaussian blur filter using a straightforward convolution (in combination with subsampling when you select anything but Best as filter quality). In SVN there is also code for using an infinite-impulse response filter, which might be somewhat similar to what the page you linked to does, but as far as I could see it's definitely not the same. (If you're interested there are some references in the code to the papers on which the code is based, see nr-filter-gaussian.cpp.)
participants (5)
-
Bill Baxter -
Bryce Harrington -
bulia byak -
Jasper van de Gronde -
Tony Vigil