W3C home > Mailing lists > Public > www-svg@w3.org > February 2005

Re: rendering of boundaries of touching objects

From: Chris Lilley <chris@w3.org>
Date: Mon, 28 Feb 2005 18:58:49 +0100
Message-ID: <246084747.20050228185849@w3.org>
To: Craig Northway <craign@cisra.canon.com.au>
Cc: Piers Titus van der Torren <pierstitus@gmail.com>, www-svg@w3.org

On Monday, February 28, 2005, 6:22:12 PM, Craig wrote:

CN> Hi,

CN> This is a known issue with the method of anti-aliasing that these 
CN> renders use. Other anti-aliasing methods can be used but are more 
CN> expensive, i.e. sub-pixel rendering uses more memory etc... The SVG 
CN> working group have decided not to specify how anti-aliasing is done, and
CN> I don't think they will in the future. This artefact is not wanted, but
CN> this is a comprise made for performance.

An example of an SVG implementation that avoids this is the demo SVG
viewer used to show the capabilities of the antigrain library.


CN> Regards,
CN> Craig Northway

CN> Piers Titus van der Torren wrote:

>>When objects touch on a boundary (both object have the same nodes on
>>the boundary) and it's rendered antialiased, with the objects filled
>>without stroke, the boundary line is visible, it's a bit transparent.
>>At least in most renderers it is (Adobe SVG viewer, Batik, KSVG,
>>The question is, is this behavior wanted? It seems to me that if
>>there's no space between regions there shouldn't be a transparenent
>>This behavior is caused by the way transparency works, two shapes of
>>50% opacity don't make 100% opacity, so the antialiased edges keep
>>I can't think of an easy way to get rid of those lines, but if there's
>>need for maybe we should think harder.
>>For example this is a problem with autotracers like autotrace or the
>>potrace based tracer of inkscape, where are transparent lines between
>>every shape.
>>see the attached file for an example.

 Chris Lilley                    mailto:chris@w3.org
 Chair, W3C SVG Working Group
 W3C Graphics Activity Lead
Received on Monday, 28 February 2005 17:58:49 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 8 March 2017 09:47:02 UTC