W3C home > Mailing lists > Public > www-svg@w3.org > June 1999

Re: Additional Filter Effects?

From: Peter Graffagnino <pgraff@apple.com>
Date: Wed, 02 Jun 1999 10:30:59 -0700
Message-Id: <37556A49.834295B7@apple.com>
To: www-svg@w3.org

Gaussian Blur is specified rather than a general convolution filter for
2 reasons: 

1) it is by far the most useful convolution for artistic effect.
2) it can be implemented is such a way that the performance is N^2 the
resolution (i.e. same as any other rendering).  General convolutions are
 N^4 (assuming the kernel scales with the image), and general separable
convolutions are N^3. 

We strongly desire a filter mechanism that has minimal performance
"surprises" at high resolutions.

Peter

Message-ID: <B38DE02578EBD111AC4200A0C95D19EE01CAE149@orsmsx30.jf.intel.com>
From: "Macri, Dean P" <dean.p.macri@intel.com>
To: "'www-svg@w3.org'" <www-svg@w3.org>
Date: Mon, 3 May 1999 16:29:13 -0700 
Subject: Additional Filter Effects?

I noticed while reviewing the SVG specifications regarding filter effects
that there isn't any general purpose convolution proposed.  Some of the
effects (e.g. Gaussian Blur) are easily implemented as a convolution.
Additionally, by having a generic convolution, effects that haven't been
explicitly taken into account could still be achieved.

Just a thought,

Dean Macri
Internet Software Engineer
Intel Corporation
Received on Wednesday, 2 June 1999 13:28:15 GMT

This archive was generated by hypermail 2.3.1 : Friday, 8 March 2013 15:54:17 GMT