W3C home > Mailing lists > Public > www-style@w3.org > January 2005

Re: Intrinsic size of jpeg images

From: David Woolley <david@djwhome.demon.co.uk>
Date: Sat, 29 Jan 2005 10:53:49 +0000 (GMT)
Message-Id: <200501291053.j0TArnF03449@djwhome.demon.co.uk>
To: www-style@w3.org

> This implies (to me anyway) the png and gif formats are REQUIRED for
> graphical UA conformance, or at least highly RECOMMENDED.

I suspect that this has more to do with GIF having simple transparency
and PNG having alpha channel transparency and being the only common 
formats with these, whereas JPG is a simple image with no features that
interact with the rest of CSS (once you ignore scaling).

More generally though, from before CSS, web designers have only really
ever targetted visual displays (it is quite rare to find web pages
that will print without being clipped, especially in portrait format,
and side panels waste a lot of space and that space can't always be
eliminated by print|selection) and have always designed in pixels,
expecting those to be display pixels.

The current definition of pixels has been generalised to be angular, but
with strong expectations that:

- an HTML/CSS pixel will rounded to an integral number of device pixels;

- for common screen resolutions, and certainly to beyond 1024x768, there
  will be one device pixel per PX.

As a result images are normally constructed to pixel sizes and any
intended rendering size an artifact of the image processing that was
done, and the expectation is that px units are device pixels and the
same as image pixels.

If I've got the maths right for the original question, I suspect that
the user had an unusually high resolution display, IE and Opera were
making the standard compromise assumption of 800x600 = 14 inch diagonal,
whereas Mozilla was compensating for the actual display resolution
(or at least the total pixels part of it) and using two device pixels
per PX.

Incidentally, as JPEG is best for natural images, which are normally
of 3D scenes, there is no intrinsic 2D size for a JPEG image (other
than in its final stages of conversion to print media) and an angular
size would be more meaningful (i.e. film diagonal/focal length)
Received on Saturday, 29 January 2005 10:59:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:35 GMT