Newsgroups: comp.graphics
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!hellgate.utah.edu!cs.utah.edu!thomson
From: thomson@cs.utah.edu (Rich Thomson)
Subject: Re: Wanted: Members for Scientific Visualization Mailing List
Date: 26 Nov 89 15:37:05 MST
Message-ID: <1989Nov26.153705.12935@hellgate.utah.edu>
Organization: Oasis Technologies
References: <11726@phoenix.Princeton.EDU> <1989Nov25.142552.1702@hellgate.utah.edu> <3399@jhunix.HCF.JHU.EDU>

In article <3399@jhunix.HCF.JHU.EDU> bio_zwbb@jhunix.UUCP
    (Dr. William B. Busa) writes:
>[Raises the question of] *why* comp-visualization is so demonstrably
>unsuccessful.  [...] a few possibilities:

I agree that the mailing list has, so far, proven to be somewhat of a
failure (in the sense that it should be communicating some information and
isn't).

>	* Its a mailing list, rather than a newsgroup.

While being a mailing list means that it isn't sponsored world-wide by
usenet (except that it is listed in the list of mailing lists posted
regularly to news.lists), it does mean that contributions arrive in one's
personal mailbox as opposed to a common feeding trough.  This can decrease
response time to a particular query, depending on how the particulars of
the mailing list mechanism.  Some mailing lists are setup to automatically
forward contributions to all subscribers.  The mailing lists I've managed
have always been of the form of re-formatted (I hesitate to say 'edited',
because I never remove any contributed information) bundles.

>	* Its subject matter is *too broad*. By striving to interest many,
>it ends up interesting none.

This could be a valid criticism given the current description for comp-vis.
comp-vis was intended to bring together people from different areas to have
them share information as opposed to a bunch of specialists in computer
graphics talking about Phong shading.  I had hoped that we would see things
like: bibliographies, pointers to algorithms for volume (really,
n-dimensional) rendering, etc.

>The proposal [...] to create a group interested specifically in "scientific
>visualization" -- which I, at least, take to mean digital image analysis
>and enhancement of images created via techniques such as microscopy, 3D
>NMR, X-ray crystallography, etc.-- would interest me directly, and I'm
>sure I'd participate in such a group (especially if it was an unmoderated
>newsgroup).

    One of the problems is that the term "scientific visualization" has so
many (too many?) meanings.  My interpretation is that sci-vis refers to the
problem of displaying data (gathered from an experiment or simulation of
some phenomenon of interest) in a more "meaningful" form than tables of
numbers.  The trick is coming up with the "meaningful" form.  For some,
this may simply be a 2D/3D graph.  For others, contour plots would be more
meaningful.  While there are some general approaches, there is the problem
of mapping the data into a visual form that _inherintly_ highlights some
particular relationship in the data.  The problem is further made difficult
by the fact that the researcher sometimes doesn't know a priori what the
relationship -*should*- look like.

    Furthermore, for just about every technique of coming up with an
image, there seems to be a specific technique for improving that particular
image.  These are techniques related to the process of imaging, i.e. what
happens between the crystal and what you see on the screen for X-ray
crystallography.  These techniques tend to be unique to scientific
visualization and the particular phenomenon under investigation.  These
methods tend to be of extreme interest to a select few, mildly interesting
to a larger group, and not of any interest to the unwashed masses.

    There are general-purpose "data" visualization techniques: density
plots, false color contour plots, isocontours.  Then, there are
general-purpose image processing techniques that can be used to filter
images (low-pass -- "smoothing", high-pass, edge-sharpening, edge detection,
contrast enhancement, histogram equalization, etc). These techniques are
not unique to scientific visualization and have many common roots with
signal processing theory and robotics.  They are of general interest to
those working in computer graphics and IP related work.  While they may
not have been specifically developed for scientific visualization, they
have a role to play in sci-vis.

    Another aspect that must be considerd is the intended audience.
This was discussed in a recent issue of _Computer Graphics World_, where
they were interviewing scientists at several research labs.  The scientists
in the article declared a need for visualization tools at both ends of what
I would call a "quality spectrum".  They wanted fast, interactive, relatively
low quality (pixel aliasing (NOT data aliasing) was acceptable, wire-frames
instead of smooth-shaded objects, etc) images while they were probing their
data for hidden relationships and verifying postulations about the
phenomenon under investigation.  Then, they wanted to be able to process
the same data to produce high-quality images (smooth shading, little or no
pixel aliasing, 3D perspective views, titles, etc) for reports and
conferences.

    Another problem is what I call the "knowledge gap" between researchers
and the computer scientists helping the researcher visualize results.  How
much knowledge of computer graphics does the researcher need to understand
the limits and accuracies of the visual form?  How much knowledge of the
phenomenon under investigation does a programmer need to ensure that the
visual form doesn't introduce "phantom effects"?  For me, there seems to be
no general answer to these questions.  Personally, when I've talked with
scientists about specific applications I have experienced both the former
and the latter.


    I would like to see comp-vis be a place where people exchange
algorithms and techniques both accessible and usable by a large community
of users (of which programmers and scientists are a part).  I think that
discussions of particular techniques for XYZ application is not going to be
of interest to most of the readers.  On the other hand, some may expect
that 'discussion of technique' means a place where they can get particular
solutions for their particular problems.

>Dr. William Busa, Dept. of Biology, The Johns Hopkins University, Charles
>& 34th Sts., Baltimore, MD 21218              (301) 338-8207
>bio_zwbb@jhunix.hcf.jhu.edu                 uunet!mimsy!jhunix!bio_zwbb

[ BTW, Dr. Busa are you familiar with Bill Thomson of Thomson Instrument
    Co.?  My dad does alot of business at Johns Hopkins. ]

						-- Rich
Rich Thomson	thomson@cs.utah.edu  {bellcore,hplabs,uunet}!utah-cs!thomson
"Tyranny, like hell, is not easily conquered; yet we have this consolation with
us, that the harder the conflict, the more glorious the triumph. What we obtain
too cheap, we esteem too lightly." Thomas Paine, _The Crisis_, Dec. 23rd, 1776
