Re: order of transform/motion animation application

On Tue, 01 Feb 2011 03:10:54 +0100, Cameron McCormack <>  

> Dirk, while reviewing the Safari/WebKit results, claims that
> animate-elem-24-t is incorrect since it gets the relative order of
> application of motion animation and the transform attribute the wrong
> way around.
> I made a little test:
> Firefox and Opera move the rectangle to position A (motion applied after
> rotate transform), while Batik and WebKit move it to position B (motion
> applied before rotate transform).
> This is what the spec has to say:
>   The various (x,y) points of the shape provide a supplemental
>   transformation matrix onto the CTM for the referenced object which
>   causes a translation along the x- and y-axes of the current user
>   coordinate system by the (x,y) values of the shape computed over time.
>   Thus, the referenced object is translated over time by the offset of
>   the motion path relative to the origin of the current user coordinate
>   system. The supplemental transformation is applied on top of any
>   transformations due to the target element's ‘transform’ attribute or
>   any animations on that attribute due to ‘animateTransform’ elements on
>   the target element.
> This seems unclear to me.  When it talks about translating along the
> axes of the user coordinate system, it sounds like it is describing
> option A.  But then stating that the transformation is supplemental and
> is applied on top of the transform makes it sound like option B.
> Which is it?  (Either way, clarification with an example is needed.)

The legacy behaviour seems to be option A. We have some examples from  
Ikivo/zoomon that seem to rely on option A behaviour, so I'm guessing that  
their animation authoring tool outputs content like that.

I have no problem with changing what Opera does in this case if it  
provides a more useful behaviour to the content authors, as it would  
simplify our implementation slightly. It would break some existing content  
though, and as Dirk notes would require changes to the testsuite(s).


Erik Dahlstrom, Core Technology Developer, Opera Software
Co-Chair, W3C SVG Working Group
Personal blog:

Received on Wednesday, 2 February 2011 09:30:05 UTC