[fa.works] WORKS Digest V2 #71

works (08/09/82)

>From PLEASANT@RUTGERS Sun Aug  8 23:51:34 1982
Works Digest            Monday, 9 August 1982      Volume 2 : Issue 71

Today's Topics:
                  Menus vs. Commands vs...? (2 msgs)
            Detectable Resolution after Printing (2 msgs)
                    Display Resolution is Relative
                    Pointing Device Report Request
                 Wanted: Info on the Pixel 100/AP...


Date:  5 Aug 1982 1355-CDT
From: Jonathan Slocum <LRC.Slocum at UTEXAS-20>
Subject: menus vs. commands vs...?

Rather than arguing the merits of one method vs. another -- all of
which have their good points in some context -- why not center the
discussion on how the machine can "grow with its user"?  It seems
obvious to me (and the loosely anti-menu folks have pointed out one
instance) that one needs a lot of help getting used to a system, but
as that happens one needs less and less help.  N'est pas?

In a similar vein, it would surely be most winning to allow the user
to tailor the interface (e.g., command completion style) to his own
preferences -- and change them over time -- without writing programs/

This makes WRITING the interface more difficult, to be sure.  But only
ONE need be written -- not a different one in each style.  If the job
is done right, (almost) everyone should be happy.

So how does one integrate the styles in a reasonable manner?  It would
seem to me that the software basis would have to be object-oriented


Date:  8 Aug 1982 0219-EDT
Subject: menus
To:  minsky at MIT-OZ at MIT-AI

I have a copy of FinalWord editor by Mark of the Unicorn staff.  They
have a great, simple idea: for commands that require more than one
character, the menu appears only if you hesitate after  beginning a
command.  Thus the menus never bother you once you get good at using
the system.


Date: 6 Aug 1982 09:14 CDT
From: Johnston.DLOS at PARC-MAXC
Subject: Re: detectable resolution after printing.

I have seen, in the earlier days of computer-GENERATED typesetting
(i.e., four years ago), a printed flier from Mergenthaler advertising
their digital photocomp machines.  I don't know what resolution they
used, but the rough edges COULD be detected, even after printing.
They weren't really bad enough to prevent using the system, but close
examination (no magnification, just paper close to face) revealed the
roughness.  This was also from a relatively inexpensive machine (as
such equipment goes) and the fonts were recorded in only one point
size.  The rest were synthesized by software from that one master,
which was loaded from floppy disk.  This may have also had something
to do with the visible rough edges, more than the resolution.



Date: 6 Aug 1982 0713-PDT
Subject: Re: Display Resolution - PhotoTypesetting
From: NCrawford at DARCOM-KA

In response to Mike O'Dell's comments on what typographers and
printers have to say about images on photographic paper that are made
by computers - I would like to make a few comments.  First, I would
like to say it has been almost 3 years since I have done any
typography, however I do understand the problems of the quality of
type produced by computer typesetting.  The type of computer that sets
type a whole character at a time gives a better image than the faster
types that scan and set a line or page at a time as

- they set from the top of the characters to the base of the
characters and

- hence not giving the quality of the single image of the character
passing through the lense and onto the paper.

I can tell the difference with my eyeball without the aid of a "100
power magnifier" the laser or "jet ink" types' problem with the
fuzziness is brought about when you pick up the speed. To me, the slow
methods should be used for the "quality" applications and the faster
methods for "newspaper" quality.  Maybe my knowledge is not on the
level with you guys, speaking in terms of "inherent integration", but
I do know that the "smoother initial edges" do cause a "smoother final
product", especially when talking about ruling and graphics images.
Are there any computers now that can rule forms with the  quality that
an artist attains with a 0000 rapidograph?  I would like to see that
kind of quality and the ability to drop type in the forms centered or
flush within those lines.  Does anyone know anything about the Xerox
computer that claims to do this (saw an ad in newsweek about it)?

- Cheryl G.


Date: 6 Aug 1982 01:01:11-PDT
From: ihnss!houxp!dvorak at Berkeley
Subject: Display resolution is relative

All of this talk about resolution is interesting, but some notions are
not accurate--or at least not relevant.  At the risk of confounding
things . . .  One must be careful what types of images are used when
"resolution" is discussed; phototypesetters usually look at a greater
variety of images than (at least for the present) most of us using
workstations.  For example, setting thin, oblique lines and halftones
requires higher resolution than viewing 10-12 point text (normalized
for viewing distance) on a video display.  This brings up two points;
first, viewing distance must be taken into account; and second, film
and CRTs are drastically different display media.

With respect to viewing distance, humans typically view text at a
more-or-less constant subtended angle; if the text is too small, you
want to get closer; if too large, you want to move back.  It all has
to do with how our eyes move when we read (the assumption is that lots
of text is to be read--a few isolated words here or there do not
matter).  This is why text in  books is typically 9 to 11 points (1
point = approx. 1/72 inch)--people view it from somewhere around 35-40
cm, depending upon the quality of their vision (which depends on age
even if no refractive errors are present).  At 40 cm, 10 pt. type
subtends approx. one-half degree of visual angle.  Stating it this way
makes things independent of the medium being viewed, which is
important because CRTs are viewed at different distances.  For sake of
argument, let's say CRTs are viewed at twice the distance we read
hardcopy.  Then we only need half the resolution on the screen than we
need on hardcopy.  BUT this assumes the physical properties of
light-emitting phosphors scanned by a raster beam are the same as the
light-reflecting/absorbing properties of paper and ink, which is
clearly not the case.

This brings up the issue of image quality and its dependence upon the
medium used.  Here I would like to argue against those that contend
that phototypesetters are not "rational "people because they want 5000
LPI on film, when that resolution is not achieved by the actual
process of printing.  The graphics arts people know a LOT more about
what it takes to make high quality images than most of us; i.e., there
is a method to their  madness.  I won't run on about this, but given
the wide variety of images they deal with (text to halftones) and the
extreme difficulty of reproducing some of them (halftones), their
demand for >1000 LPI is reasonable.  This is especially true in light
of the fact that a formal knowledge about halftone reproduction by
digital means, and our perception thereof, does not exist.  This does
not justify the phototypesetters' demands for high resol., but it
makes it hard for us less-experienced to criticize (or at least it
should).  And don't forget that the detail on a 2500 LPI film image is
somewhat degraded in making the master, again in the offset process,
and again in the ink-paper interaction; thus to get 500 LPI on
calendared paper might well take much more on the original film.  Yet
another consideration is that they work at various % of original.  So
let's withhold at least some of the slander; those guys are in
business, so their bucks are where their LPI are.

Getting back to the generic issue of 'resolution' -- I have visually
scaled a variety of text images at various resolutions, with final
reproductions on high-resolution photo paper.  Images above about 600
per inch were essentially indistinguishable from  images at about
1200.  Remember--these are photographic comparisons--and other display
media (CRT, laser printing, etc.) will not be as high quality as
photography--so that on a CRT, 300 per inch might well be
indistinguishable from an "analog" image on a phosphor at "infinite
resolution," and 200 per inch will look pretty darned good.  Once
again, the qualifier is that the displayed image is text of various
point sizes, styles and faces.

By the way, up until recently, phototypesetters laid out pages using a
display that was not a bit-mapped representation of the final page,
but a crude symbolic representation thereof.  While 200 per inch on a
CRT wouldn't hold a candle to images on film, any typesetter would
like to have a terminal that could display 200, thereby affording a
more realistic "preview" of what the page was really going to look

In summary, resolution is relative, and what resolution is "necessary"
is dependent upon the image being displayed, the medium used to
display it, the observer viewing it, and the application.  This is why
200 per inch would suffice as high quality for most CRT-based
workstations, but 800 per inch is insufficient for phototypesetting
pages with pictures.

Chuck Dvorak (houxp!dvorak)
Bell Labs


Date:  7 Aug 1982 at 0027-PDT
From: hampton at ACC
Subject: Pointing Device Report Request
cc: UNIX-Wizards at SRI-UNIX

Can anybody out there remember a (Bell Telephone Labs, I think) report
contrasting the various pointing devices ( Trackballs, light-pens,
mice, etc.)?  A copy of the report or a pointer to it would be


Date: 5 Aug 82 16:37:13-PDT (Thu)
From: decvax!harpo!npoiv!npois!cbosgd!nscs!jpj at Ucb-C70
Subject: Wanted: Info on the Pixel 100/AP...

Greetings!  I just received some information about a UNIX/System III
Multi-user configuration built around a machine called the Pixel
100/AP.  This is what the system looks like:

        100/AP (sic) Supermicrocomputer (68K/8 Mhz)
        1 mByte 150ns dynamic RAM
        128k I/O processor memory
        40 mByte Winchester disk w/controller
        2 x 630k Diskette drives
        8 x Pixel terminals (!)
        8 x RS232 ports
        2 x parallel ports
        UNIX System III
        Choice of programming language (includes 'C', UCSD Pascal...)

(And the bottom line ...)       Price = $19,900.00

Well!  Has anyone heard of these folks?  Better still, does anyone
have any experience with this stuff?  To get 8 users on a UNIX system
for ~ $20K seems hard to believe.  Comments please - I'm in a bit of a

Thanks in advance...

Jim Jenal
BTL/CB  (but not for long...)


End of WorkS Digest