Re: Stars in Stereo

[crowell@xxxxxxxxxxxxxxxxx (Jim Crowell)]

At 3:01 AM 2/26/96, kirk bender wrote:

>>There was some discussion recently about whether or not the stars
>>can be seen in stereo.
>...
>>Now the best stereoacuity I know of is 3" of arc.
>...
>
>Err... I don't know what you mean exactly by stereoacuity,
>but a book I have says the eye's optics are limited to 1 arc minute of
>resolution ( the diameter of a dime at 60m), and the human
>retina actually sets it to about 3 arc minutes. Maybe you meant
>arc minutes, rather than arc seconds.

As John B. already said, it really comes down to what exactly
you're measuring; "resolution" is a fairly vague term.  On the
one hand, you can measure "grating acuity", which is in fact a
measure of the optical quality of the eye.  You do this by
starting with a pattern of light & dark, very high-contrast
stripes on a gray background & decreasing the width of the
stripes until they can't be discriminated from a uniform gray.
Typically this gives you a value around 1 arcmin.  The acuity
your optometrist measures with a letter-chart is supposed to be
related to this, although in fact the relationship isn't all
that close.

On the other hand, you have the "hyperacuities", which give you
much smaller numbers & generally involve detecting differences
in position of two things that are simultaneously visible.
Vernier acuity is one example of this.  Typically it's measured
by taking a vertical line & shifting the bottom half of it
slightly to one side or the other--making a kink in the line--&
finding the smallest offset the subject can detect.
Stereoacuity's another example; you might measure it by
presenting a subject with pairs of vertical lines in each eye
with slightly different spacings between them--leading to a
perceived difference in depth--& finding the smallest difference
in spacing that can be detected.

Anyway, the basic idea is that in one case you're measuring how
well you can see something that's really small--the optical
quality of the system--& in the other case you're measuring how
well you can localize two things that are clearly visible.

At 1:21 PM 2/26/96, John Bercovitz wrote:
>
>One should also consider the way these acuity tests are done.  The
>subject is asked to guess and when the guesses statistically beat
>chance probability, the subject is said to have that acuity.  Not a
>very fair figure for average use.

Typically in a lab the subject has to reach some criterion level
of performance, typically 75%-correct because that's usually the
easiest value to measure accurately.  In a clinic, the biggest
problem is that you're reading that same set of letters over &
over, so of course after some point you _know_ what they are...

Actually, I think the most important differences between these
setttings & everyday life have to do with the displays
themselves; very high contrasts, sharp edges, that sort of
thing.

-Jim C.


------------------------------------------------
Jim Crowell
U.C. Berkeley School of Optometry
360 Minor Hall, Berkeley, CA 94720
(510) 642-7679
jim@xxxxxxxxxxxxxxxxx



------------------------------