W3C home > Mailing lists > Public > www-style@w3.org > December 1998

Re: the whole pixel size dilema

From: Chris Lilley <chris@w3.org>
Date: Tue, 29 Dec 1998 23:57:22 +0100
Message-ID: <36895E52.43F5A2D8@w3.org>
To: benjaminh@epic.co.uk
CC: www-style@w3.org

Benjamin Hardcastle wrote:
> > that IE3,4, or 5 will not scale pixels in raster artwork and HTML elements
> > such as tables when the logical resolution of the system is set to 120dpi
> > or "large fonts". It should scale by 25%).
> >
> Graphics only scale well when you have a scaling factor of multiples of 100%. 

That is not true. It is true that if a particular briowser implements
scaling badly, then it is better not to scale. However, that is not the
same thing.

> If
> you were to scale by 125% do you interpolate the extra pixels? 

Yes, using either bilinear or (preferably) bicubic interpolation.

> Do you do nearest-neighbour? 

Not if you want to look at the result and retain your lunch.

> Think of how that would affect a simple shape such as a
> circle. You would either get a blurred image (interpolation) or discontinuities
> in the image (nearest-neighbour).

It isn't that bad. Photoshop users scale images all the time.

>  Of course, this is all fine if the image is a
> vector one rather than raster. 

Exactly, which is why W3C is working on an XML vector graphics format.

> How much burden can a UA take? After all, it
> would have to do some form of conversion on the image to scale it. For
> interpolation, the UA would have to convert to 16 or 24 bit to scale properly.

Yes it would, good point. If there is inadequate color availability (ie
8 bitindexed, or, worse, the dreaded "web-safe" fixed palette) then
scaling images shows up this lack.

For an example of a browser where the images and text can be made to
scale together, try Opera 3.51 (you need 3.51 rather than 3.5 to get the
PNG support).

Received on Tuesday, 29 December 1998 17:54:56 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:53:56 GMT