Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site mit-eddie.UUCP
Path: utzoo!linus!genrad!mit-eddie!smh
From: smh@mit-eddie.UUCP (Steven M. Haflich)
Newsgroups: net.micro
Subject: Re: Question on the mouse, comments abou - (nf)
Message-ID: <389@mit-eddie.UUCP>
Date: Mon, 4-Jul-83 11:22:39 EDT
Article-I.D.: mit-eddi.389
Posted: Mon Jul  4 11:22:39 1983
Date-Received: Mon, 4-Jul-83 15:48:41 EDT
References: <1690@yale-com.UUCP>
Organization: MIT, Cambridge, MA
Lines: 31


	A more complete explication of the `crosshair' method for following
a lightpen, cited a while ago by Jerry Leichter:
	The solution involves displaying a right-angle cross comfortably
larger than the aperture of the pen.  The cross point displays the cursor
position.  Following consists of keeping track how far from the cross point
in opposite directions the cross is visible to the detector.  The tracking
algorithm continually adjusts its idea of the cursor location to keep the
visible portion of the cross centered.
	That's all in theory, for there are several practical problems.
For some applications, it is necessary to have an acquisition mode in
which the display fills the screen with hash in order to pick up the
initial position of the pen.  (Otherwise, the user must point to the
displayed cursor to `pick it up,' analagous to picking up a mouse.)
Some sort of smoothing algorithm is necessary on tracking, for detection
at the extremes of the aperture tends to be noisy (or rather, statistical).
All this, of course, burns a lot of processor time (possibly all of it),
although the problem can be moved to a dedicated microprocessor.  Remember,
a mouse generates no output while stationary.  Also, this technique awkward
to implement on some lightpen/display hardware.
	My experience (which is admittedly ancient) is that lightpens work
a lot better for pointing to discrete objects, such as menus, than they do
for detailed tracing.
	Like Jerry I cannot provide a reference for first use of the
crosshair method.  However, it was the common penfollow technique used
at the MIT RLE PDP1 in the mid '60s.  (If my sources are any good, this
this machine with its old point-plotting display, but without the exotic
timesharing hardware, is now on display and running at the DEC museum
in Marlborough MA.)
					Steve Haflich
					genrad!mit-eddie!smh