Today I was wondering about how to make the about screen more easily translatable. I figured out that the main problem is that text and effects applied to it are stored as curves. Maybe if we could use raw text and apply filters to it, then translators would have an easier task. This made me try to implement the mirrored reflex effect that we can see in the 
0.45 about screen using filters. This is what I came up with:<br><br>(1) feImage xlink:href=#(gradientrect)<br>(2) feDisplacementMap input=SourceGraphic and displacement map=(1)&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; &lt;-- makes the mirrored reflex<br>(3) feComposite input1=(2) input2=SourceGraphic &lt;-- adds the original image to it
<br><br><br>apply a linear gradient to the rect whose id is gradientrect<br>This gradient must be setup so that it mirrors the source graphic when applied to feDisplMap<br>We probably would need to have an additional primitive on this stack, in order to alter opacity in the reflex
<br><br>Unfortunately, we don&#39;t have feImage working yet. Maybe there is another way of implementing this.<br>And it also remains the problem of having to depend on system available fonts<br>Anyway, this is just brainstorm :-D
<br><br>This also brings to discussion the general question about tools to make easily translatable drawings using SVG.<br><br>Juca<br>