Thank you very much for your feedback.

As long as the background can be easily removed when using the existing filters, I am now figuring out that my proposed chromatic aberration filters do not bring something fundamentally new (except from the behavior with the alpha channel, which is different).

However, I see 2 issues with the existing filters that can be improved:

1) Aliazing issue with the 'nudge RGB' filter:

The 'nudge CMY' and the 'nudge RGB' filters do not have a totally symmetrical behavior. While the 'nudge CMY' filter is perfectly fine, by careful examination, one will notice that the 'nudge RGB' filter suffers from a subtle aliasing effect at the border of the 3 decomposed layers (present in all browsers rendering and Inkscape itself). However, I found a nice (not so easy to figure out) solution to this. Replace the 3 color matrices:

 red:

  0 0 0 0 1
  0 0 0 0 0
  0 0 0 0 0
  1 0 0 0 0

 green:
  0 0 0 0 0
  0 0 0 0 1
  0 0 0 0 0
  0 1 0 0 0

 blue:
  0 0 0 0 0
  0 0 0 0 0
  0 0 0 0 1
  0 0 1 0 0

by the following new ones (in short: add the 1 -1 at the end of the last lines):
 blue:
  0 0 0 0 0
  0 0 0 0 0
  0 0 0 0 1
  0 0 1 1 -1

and you will see that the weird aliasing effect just vanishes. Moreover, that way, the 'nudge RGB' filter recovers a totally symmetrical behavior compared to the 'nudge CMY' filter and everything is fine.

2) Offset issue with the 'nudge RGB' and 'nudge CMY' filters:
The 2 first offsets are (logically) applied to the source graphic while the third offset is (surprisingly) applied to the blended results of the first 2 layers. This is rather counter-intuitive. It seems more logical that the third offset also applies to the source graphic (my filters just do that).

And, voilà: if those 2 issues are corrected, one will not really need my filters anymore. As long as the existing filters are actually built-in filters, I'm going to open a bug report (with solution) to <https://bugs.launchpad.net/inkscape/>.

Warm regards,

Fabien