Re: one remark about SVG

At 12:51 AM 3/7/2004 +0300, 
=?koi8-r?Q?=22=E1=CE=C4=D2=C5=CA=20=ED=C1=C3=C5=C5=D7=D3=CB=C9= wrote:

>I think, the perfect vector format must support any gradient filling, not 
>only linear and radial. Such restrictions make SVG less flexible, than it 
>could be. When one wants to paint something, using strictly limited set of 
>tools, he needs to compensate absent tools in some left-handed way. It 
>means, in this case, creation additional contours, which aren't a must. 
>There are at least two potentially useful methods for 2-D interpolation, 
>which are never used in gradient filling. Look at WinAPI's procedure 
>GradientFill(). It requires a mesh of triangles as one of its arguments, 
>but this fact means only, that boys from Microsoft know math not very 
>well. There is a way to realize gradient filling without meshes at all. 
>One the other hand, to enclose every area, which is to be paint, isn't a 
>must too. There is a way to use open contours and therefore don't create 
>contours, except existing in an image. I think, sooner or later, such 
>potentially useful options will be included in some
>   vector format. There is only one question- who will be the first. 
> mats54@mail.ru.

We had it on our list of things to do for long time, but never got around 
to actually doing it.
One problem is that triangular mesh is an easy way to implement but not 
particularly semantically
meaningful way of expressing it (as you mentioned it), so it does not seem 
like it is any better
solution than just embedding an image. If you have a pointers on how you 
think it can be done on
"higher" semantic level with contours, please point us at it (Russian - 
guessing by your name and
address - is OK, as well as math). Note that in the end it's almost 
certainly will have to be broken
up by the implementation into triangular mesh that approximates shaded fill.

Peter

Received on Sunday, 7 March 2004 19:19:22 UTC