[CREATE] OpenRaster and filters

[Øyvind Kolås]

On Tue, Jun 17, 2008 at 7:43 AM, Kai-Uwe Behrmann <ku.b at gmx.de> wrote:
> Am 17.06.08, 00:01 +0200 schrieb Cyrille Berger:
>> On Friday 13 June 2008, Kai-Uwe Behrmann wrote:
>> > b) to make options a non pixel thing, e.g. move to a precentage
>> >    relationship
>> How would you define a percentage for a radius ?
>
> Well, there are some logical answeres. I would choose the OpenRaster image
> width as the reference.

The width of the image is not good for OpenRaster as an OpenRaster
document essentially can be copied and pasted as a layer into another
OpenRaster document. For this to work you would have to work on the
extent of the input buffer not the extent of the image (a final crop).

Besides the whole premise of this discussion is wrong as we most
probably shouldn't be discussing an ill-defined radius for a gaussian
but the standard deviation to be used (in pixels or not is still
relveant though).

There are filters that only take parameters making sense in pixels
though, the way the SVG 1.2 filter specification solves this is to
specify a specific resolution that the filter should be interpreted
at.

> With pixel independency the filters can easily be applied to several image
> resolutions without any recalculations.

This is not likely to happen in this manner, since both the filters
and the images involved form part of a composition. We haven't been
discussing resolution independence in OpenRaster at any length yet,
and I think it is premature to do so. Both Krita and GIMP are mainly
focused on layers that remain at fixed scale and measture their
positioning using pixel. It is important that we get this working
before we start worrying about positioning the upper left of the
auxiliary buffer in the over operation based on relations to the input
buffer, as well as build scaling of the layer according to it's parent
layer).

If/when we go down the route of incorporating relative units for
geometry related properties I do hope
that we stick with numbers in the 0.0-1.0 range instead of 0.0-100.0.

/Øyvind K.
-- 
«The future is already here. It's just not very evenly distributed»
 -- William Gibson
http://pippin.gimp.org/ http://ffii.org/