Around 15 o'clock on Nov 26, Dr Andrew C Aitchison wrote:
> Come to that, there is no good reason why the average application
> should be using fonts where it matters. If I have two 1024x768 displays,
> one a 14" laptop and the other a 50" plasma display, I usually want the
> text to be the same number of pixels high in them both, not the same
> point-size.
This argument starts with the assumption that the display is being used in
a similar manner in both environments, and that users will position
themselves at an appropriate distance so that the angular size of the
display is essentially constant, independent of the linear size of the
screen -- the angular size will fill the "comfortable" area where the
entire screen can be easily scanned by both eyes without head movement.
An informal study of various screen sizes around your own environments may
well demonstrate this -- larger monitors are often pushed further from
the user than smaller ones. Those with televisions may also notice this
effect; a 40" tube cannot be comfortably viewed from 5 feet away. I'd
love to perform a more formal experiment...
I'll note that in 1990, the typical monitor was 1024 pixels wide and that
in 2000, a typical monitor was 1280 pixels wide -- the shift from using
75dpi fonts by default to 100dpi fonts mirrors this display size migration.
Of course, there are limits -- the comfortable focal range of the user
often limits how close the monitor can be placed, and larger monitors are
often designed for a number of people to view simultaneously. On
projection monitors, you often need to *increase* the size of the text so
that audience members further away get a readable angular size.
Once you start with this postulate, the logical consequence is that the
effective angular size of the pixel depends solely on the number of pixels
on the screen. I don't know whether this depends on the horizontal or
vertical size, but the ever-increasing aspect ratio of movie presentations
leads me to suggest that width dominates height, at least in that venue.
It also means that larger monitors serve two purposes -- to increase the
effective angular resolution of the monitor and to allow those of us with
apparently decreasing arm length to continue to read the screen.
For the former, the same effect can be had by purchasing a smaller monitor
with finer dot pitch, or by taking advantage of the sub-pixel resolution
available from digital LCD monitors. The latter can be solved with reading
glasses (admitedly, not an ideal solution).
This argues rather strongly for a decoupling of physical monitor resolution
from effective DPI used in calculating presentation sizes. In this world,
we might talk about a "logical" DPI, calculated and stored separately from
the physical size presented by the X server.
Xft already uses a separately configurable value (Xft.dpi) that can be used
to adjust the size of text on the screen.
I suggest a simple calculation for this logical DPI:
logical_dpi = max (width_in_pixels / (13 2/3), 75);
width logical_dpi
640 75
800 75
1024 75
1280 94
1600 117
1920 140
2048 150
Limiting the value at the low end is designed to reflect that at some point
text at normal point sizes must occupy a reasonable number of pixels to be
readable.
This calculation only works for desktop environments; other environments
have other assumptions. Palmtops are expected to occupy a smaller fraction
of the visual field while Omnimax theaters and other immersive environments
are designed to fill a significantly larger fraction.
[EMAIL PROTECTED] XFree86 Core Team SuSE, Inc.
_______________________________________________
Fonts mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/fonts
