W3C home > Mailing lists > Public > www-style@w3.org > April 2012

Re: [css3-transforms] Relation between mathematical description and effect of a transformation

From: Dirk Schulze <dschulze@adobe.com>
Date: Sat, 21 Apr 2012 06:55:18 -0700
To: "Dr. Olaf Hoffmann" <Dr.O.Hoffmann@gmx.de>
CC: "www-style@w3.org" <www-style@w3.org>, "public-fx@w3.org" <public-fx@w3.org>
Message-ID: <1FFC88AF-B869-415D-81B7-95899205FDF7@adobe.com>
Hi Olaf,

On Apr 21, 2012, at 6:40 AM, Dr. Olaf Hoffmann wrote:

> Hello,  
> 
> this is about:
> http://www.w3.org/TR/2012/WD-css3-transforms-20120403/#mathematical-description
> 
> The draft shows some examples, what can be assumed as the effect of several 
> types of transformations and in this chapter 17. it provides a 'Mathematical 
> Description of Transform Functions' - well, better it provides only matrices,
> no direct relation to the effect of such a matrix on the presentation.
> 
> For 2D-Transforms it is sufficiently described in SVG already what the 
> effect for a point r = (x, y, 1) for a matrix M is, r_p representation in the 
> previous coordinate system, r_c in the current coordinate system
> (respectively r=(x, y, z, ?) for three dimensions?): 
> 
> r_p = M r_c

A bug report [1] on W3C also points out that we should be more clear and adapt more from SVG 1.1[2] as well as SVG Transform [3]. And I agree.

> 
> 
> I think this should be noted in this draft as well - and this is even more
> important for the matrices related to 3D-transformations, because it is not
> obvious, what the relation is.
> The old SVG transform draft http://www.w3.org/TR/SVG-Transforms/
> has slightly more advanced descriptions. Well, even with these formulas
> I do not get something similar to for example 5 of the current draft.
I agree that SVG Transforms is more descriptive, but also slightly different from CSS3 Transforms.

> 
> Due to my experience with perspective transforms, for a central projection,
> what seems to be intended here in examples like 4,5, one needs an
> additional transformation like (index _p here for projected)
> 
> (x_p, y_p) = (x_c, y_c) * l/z_c with l a length.
> 
> Obviously the fourth dimension of the matrices is intended for this,
> but the relation to such a transformation is not decribed.
> 
> The parallel projection as intended in example 3 is simpler,
> one just has to use a simple 3x2 (respectively 4x2) matrix, to
> extract only the x and y components.  
> 
> 
> 
> 
> 
> I suggest to decribe/define the effect of such matrices
> in detail as a functional relation between the representation of an arbitrary
> point r_c in the current coordinate system to the representation of
> this projected point r_p.
> Other solutions for the problem are possible as well of course, but
> without a precise description at least the effect of the 3D transforms
> are undefined and those of the 2D transforms are applicable only for SVG,
> that has already a precise description for 2D.
> 
> 
> Best wishes
> 
> Olaf
> 
> 
> 
> 
> PS: Is it really useful to change the preferred mailing list for this draft
> to the www-style list instead of the public-fx, as for the previous draft?
> Because the draft applies still to SVG as well, I added the public-fx
> 
Mails should always go to public-fx. I am not sure why this changed in the ED.

Greetings,
Dirk

[1] https://www.w3.org/Bugs/Public/show_bug.cgi?id=15605
[2] http://www.w3.org/TR/SVG/coords.html
[3] http://www.w3.org/TR/SVG-Transforms/
Received on Saturday, 21 April 2012 13:55:49 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:52 GMT