W3C home > Mailing lists > Public > www-svg@w3.org > June 2011

Re: which version is correct?

From: Alex Danilo <alex@abbra.com>
Date: Thu, 16 Jun 2011 12:48:14 +1000
Message-Id: <ES3VML.95FNQ90LIAEF@abbra.com>
To: "David Dailey" <ddailey@zoominternet.net>
Cc: "'www-svg'" <www-svg@w3.org>
Hi David,

--Original Message--:
>In the SVG WG’s familiar example of composite filters at http://www.w3.org/TR/SVGFilterPrimer12/examples/filters01.svg ...
>I recently had occasion to look a bit more closely. Four browsers make reasonable attempts at pulling it off: Chrome, Opera, IE/ASV and FF4.
>ASV and Chrome are fairly close to one another, with Opera’s version being darker and more “tubefy” like and FF’s being considerably lighter. The differences become much more apparent as the graphic is scaled to fill a larger part of the screen (by, for example simply removing the part of the code that specifies the size of the SVG as width="7.5cm" height="5cm" ). 
>My question :  which version is closest to what the spec says? 

A very good question - and also very hard to be sure of. I guess a clean room implementation from
just the spec. might be able to answer that question, but it's not likely to be simple. There are a number
of things that can affect the example - like was the correct colour space used for filtering (linearRGB
as default) - and how many bits per channel were used, were overshoots clamped before the end of
the compositing pipe or at each stage, etc.

>A meta-question: is the spec intended to nail such issues to such accuracy that these browser differences will one day cease to exist, or is it intended to give each implementer such freedom as to interpret the spec with her own corporate flair?

I don't think the differences you're seeing are kosher. The implementations should be close enough
that a trained eye would be needed to pick any difference. In that example the differences are far
too large to tolerate.

Of course, more content on the web using filters might push the implementations to clean up any
differences, as would more comprehesive tests so that the implementers have something to
compare with (chicken and egg again).

>Ultimately, if SVG is to be used for things like trademarks (or other things requiring high precision reproducibility) , then a  trademark applicant may need to cover a lot more bases in specifying the range of art that is sought to be protected.

Totally agree with you here.


>Thanks for sharing your opinions, as I’m not sure all will agree.
Received on Thursday, 16 June 2011 02:48:48 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 March 2017 09:47:25 UTC