[sugar] icon assistance/validation

[Eben Eliason]

On Tue, Mar 18, 2008 at 6:09 PM, Gary C Martin <gary at garycmartin.com> wrote:

> On 18 Mar 2008, at 20:13, Eben Eliason wrote:
>
> > I can remove the stroke-opacity altogether.  As mentioned before, I
> > don't have any particular goal in mind with respect to that
> > property...I just implemented it at the same time as fill-opacity
> > because it was easy to add.  I do want to keep the fill-opacity so we
> > can take advantage of it in the future, though.  It is something that
> > may not be implemented immediately, but would make the code much
> > cleaner once we are able to support it.
>
> What would happen to overlapping fill areas if opacity starts getting
> fiddled with? My understanding is that the overlap areas will be
> additive and there for more opaque than non-overlapping areas leading
> to fuggly shapes appearing. Icon designers would need to always build
> all non trivial fill shapes out of little bits of non-overlapping
> primitives. That's quite a tall order.
>

Yeah, the more I've been playing with this, the more I'm realizing that it
might not be feasible.  The "hack", while semantically less clean and
requiring a few more lines of code, actually makes development of the icons
easier in general, and ensures that their look will be retained.


>
> If there really is some great feature that needs alpha, can the SVG
> not just be rendered in to a pixbuff, and then the pixbuff be
> composited into a widget with alpha. This way the SVG is 'flattened'
> so that you can dial up/down the transparency across the whole image
> without fuggly bits appearing.
>

Well, not exactly, as the goal is to render the "strokes" opaque gray and
the "fills" transparent (Those are in quotes since you can have both in
either color, technically).  I suppose, however, that you could easily use
the entities to paint strokes white and fills black, render it to a pixbuff,
and apply it as a mask to a solid of the desired color.  That would
certainly do the right thing.

Of course, in order for this to serve the purpose we're after, you have to
be able to create the pixbuff with the fill at various levels of gray, and
treat the result as a true alpha channel, and not as a bitmask.  If you want
to take a stab at this approach, feel free.

- Eben
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.laptop.org/pipermail/sugar/attachments/20080318/e8c1e120/attachment.htm