Re: minutes, 29 August 2013 SVG WG telcon

On 2013-08-30 00:42, Cameron McCormack wrote:
> ...
>    krit: so I propose to remove kernelUnitLength="", either add a
>    new attribute to select between matching to CSS px or device
>    px, or use one of the existing properties
It's not necessarily a bad idea to use CSS px, and perhaps using "filter 
units" as in kernelUnitLength is not the greatest idea ever (I'm not a 
big fan of them anyway), but I /would/ allow specifying a value other 
than one (so not just a "toggle"). Especially given the increased 
availability of high resolution screens, it is not unreasonable to want 
a filter to act on a very specific scale. For example, for Gaussian blur 
we allow pretty much arbitrary sigma's, so why restrict a convolution 
specified by a matrix to act on "pixels" (whether they're CSS "pixels" 
or device pixels)?

I wouldn't worry too much about quality, there are some pretty decent 
ways of scaling a filter up or down. Also, not all classes of filters 
are equally affected (blurs tend to behave fairly well under scaling, 
differentiations less so). So if the author has some control over it (by 
specifying a value that is an integer number of pixels or not), then it 
should simply be fine in practice.

Received on Friday, 30 August 2013 07:00:09 UTC