eugene@eos.UUCP (Eugene Miya) (11/27/89)
In article <3399@jhunix.HCF.JHU.EDU> bio_zwbb@jhunix.UUCP (Dr. William B. Busa) writes: > So, I pose the question: are these the reasons why >comp-visualization doesn't work? I further ask -- is my interpretation of >"scientific visualization" correct (lousy choice of words there...how >about "appropriate?"). Well, "visualization" can and does work, but you tend not to see it in this group. This group is largely concern with synthetic image synthesis. This is not the orientation of scientific graphics. Scientific graphics (analytic graphics to quote Carol Hunter at LLNL) has a somewhat different orientation than synthetic image generation. It is EXTREMELY important to make a distinction between the researchy, prototypical, quick view versus the flashy presentation graphics which come out from certain supercomputer centers. "Scientific visualization" has come off looking like a solution looking for problems. This is partially the "fault" of artists, partially the fault of some eager researchers, partially the fault of a conservative scientific community. The interest in Mandlebrot sets and "Chaos" by a some what naive and over enthusiastic public (but can you blame them when you consider warm superconductors and cold fusion?) The Report on SV (I was give a case of these) tried to emphasis the non-quantitative aspects of SV. Alvy Ray Smith when he talks about this subject has to emphasize the non-geometric aspects. BUT the general orientation in science IS numbers. It's not just numbers but mathematics which are the primary predictive tool in the sciences. It's hard to make a prediction from a picture. Well it turns there are people who do use imagery quantitatively: photogrammetrists, cartographers, etc. Billions are spent in areas like these. There are other aspects. Existing synthetic imagery systems introduce problems such as perspective, lighting models, anti-aliasing, etc. Problems? You say? Yes, the quantatitive imagery people spend those billions for machines to remove perspective. Lighting models clash with the display of data values with coloration. Also needed are frames of reference: scales, norms, etc. See, these are things independent of specific disciplines like x-ray crystallography or fluid dynamics. The scientist wants to know abouyt things like lengths, areas, volumes, temperatures, pressures, etc. etc. It has been said that the artist likes a blank canvas, well scientists tend to like graph paper or tabular papers. And so on and so on. Anyways, I can go on, but this is what I told IBM when they came by to ask whether they should get into SV. You must also remember, that graphics to these end users is a tool. Just a tool. They don't ponder their crescent wrenches (another less powerful tool). A lot of scientists have been somewhat offended by the entertainment like nature of some presentations. That should be toned down, but we do have to get more young people interested in the sciences. But remember to make the distinct between education and research. Another gross generalization from --eugene miya, NASA Ames Research Center, eugene@aurora.arc.nasa.gov resident cynic at the Rock of Ages Home for Retired Hackers: "You trust the `reply' command with all those different mailers out there?" "If my mail does not reach you, please accept my apology." {ncar,decwrl,hplabs,uunet}!ames!eugene Support the Free Software Foundation (FSF)
mcdonald@aries.uiuc.edu (Doug McDonald) (11/27/89)
In article <5646@eos.UUCP> eugene@eos.UUCP (Eugene Miya) writes: >It is EXTREMELY >important to make a distinction between the researchy, prototypical, >quick view versus the flashy presentation graphics which come out from >certain supercomputer centers. >Another gross generalization from > >--eugene miya, NASA Ames Research Center, eugene@aurora.arc.nasa.gov Well, I have a specific scientific graphics problem that left our NCSA helpless - they have nothing to help me. (In actual fact this arises from my hobby - but that matters not at all.) I have a pile of negatives of a couple of astronomical objects that I have made with my telescope. The present ones are made identically, except that a few have been made through clear glass plates. In the future I want to use filters. I want to combine these on a computer. The ultimate goal is to make pictures by combining photos made through narrow band filters (where I observe there is no power for a CCD, so I will use a nice image tube I already have.) So I have these 35 millimeter negatives (black and white usually, but some might be color). How do I get the data into a computer? I need a minimum of 256 gray levels, 1024 would be more than enough. There is enough resolution to need 2048x2048 pixels. Nobody around here seems to have any device to do this. I can, and have, printed the negatives and run them through an Apple scanner on a Mac II, but the software only uses 16 gray levels, though it has enough resolution. But actually using prints is a bad idea due to image distortion - I need to be able to add together various frames for signal to noise enhancement. Can anyone recommend something that will do this digitization, preferably something reasonably cheap? I would prefer to connect the thing to an IBM PC though a Silicon Graphics Iris 4 would do (I have to pay through the nose for the Iris). Doug McDonald (mcdonald@aries.scs.uiuc.edu)
eugene@eos.UUCP (Eugene Miya) (11/28/89)
In article <1989Nov27.024857.9480@ux1.cso.uiuc.edu> mcdonald@aries.scs.uiuc.edu (Doug McDonald) writes: >Well, I have a specific scientific graphics problem that left our >NCSA helpless - they have nothing to help me. (In actual fact this >arises from my hobby - but that matters not at all.) Unfortunately, this tends to kind of matter with bureaucratic organizations which control the type of equipment you need below. I can only sympathize. >I have a pile of negatives of a couple of astronomical objects > >So I have these 35 millimeter negatives (black and white usually, >but some might be color). How do I get the data into a computer? >I need a minimum of 256 gray levels, 1024 would be more than enough. What you need is a Perkin-Elmer Microdensitometer (we had one in the Radar Science and Engineering Group at JPL). You could not get 1K levels, nor are they cheap, but you might contact a P-E salesperson to see if there is one in your area which you might be able to borrow one some where. Another gross generalization from --eugene miya, NASA Ames Research Center, eugene@aurora.arc.nasa.gov resident cynic at the Rock of Ages Home for Retired Hackers: "You trust the `reply' command with all those different mailers out there?" "If my mail does not reach you, please accept my apology." {ncar,decwrl,hplabs,uunet}!ames!eugene Support the Free Software Foundation (FSF)
jim@baroque.Stanford.EDU (unknown) (11/28/89)
In article <5646@eos.UUCP> eugene@eos.uucp writes:
But remember to make the distinct between education and research.
Actually, I think there are three distinct areas in which computer
aided "visualization" is a useful tool. Although, each has very
different needs, those needs are very well served by the large amount
of information that can be conveyed visually. I call the three areas
NerdVis, EdVis, and ShowVis.
I. NerdVis. This flavor of "scientific visualization" gives a
scientist or engineer who is trying to solve a problem a better
understanding of the system he is studying. Ideally, the
visualization would not be separated from the simulated or physical
process, so that one could interactively tweak code or parameters and
immediately "see," i.e. analyze, measure, compare, etc., the results.
But currently, because of the different capabilities of the simulating
(supercomputers) and visualizing (workstations) systems, the processes
are separated. A similar separation exists between most experimental
measurement and display systems. This separation may always exist,
but conceptually, the "visualization" should not be separated from the
problem being studied.
NerdVis is NOT an end in itself. It is a means to understand a
problem. The goal is to give each and every scientist the tools that
will allow her to "visualize" and interact with her own data to
understand her specific problem. To be useful, a visualization can't
take weeks, or even hours, to produce. There is NO need for graphics
artists or renaissance teams. This is the scientist's ball game.
Ideally, a graphics programmer shouldn't be necessary. These
"visualizations" should be as easy to produce as X-Y plots. I think
Stardent's AVS and Artifact are early prototypical examples of tools
which will empower scientists to visualize their own data without
kowtowing to graphics gurus! Power to the scientific masses!
II. EdVis. Educational "visualization" needs to be simple,
interactive and enticing. This means simultaneously manipulating the
parameters of a physical model and "visualizing" the behavior of that
model. Unlike research level visualization, educational models are
usually simple enough that pedagogically useful and graphically
compelling simulations can be made on almost any PC or workstation.
To date, the software to do this is sadly lacking. After all, who
develops it? A high school teacher here, a professor there. Each
doing his own thing for his own class with little cooperation or
sharing.
The tremendous potential that EdVis has to make math and science more
interesting and accessible has not been exploited. Unfortunately,
there is little financial incentive for commercial development of such
software. I also know of little governmental support.
III. ShowVis. Now comes the fancy "gee whiz" presentation quality
visualization. It isn't necessarily directly helpful to scientific
understanding, but indirectly it is indispensable. In order to
survive in a society full of slick presentations, science needs to
show a glitzy face too. ShowVis is the type of visualization for
which hundreds of person hours might be spent to produce a single
video segment. (Not really practical if you're generating results
from ten Cray runs a week.) ShowVis needs the services of
professional design and production teams: programmers, graphics
artists, editors, narrators, etc. The primary purpose is marketing,
i.e. convincing management, Congress, agency officials, and even the
general public that certain projects are worthwhile and should be
funded. As an unavoidable side effect, some education occurs, indeed
this provides perhaps the only information the TV bound public will
get about Mars, Saturn, etc.
Jim Helman
Department of Applied Physics P.O. Box 10494
Stanford University Stanford, CA 94309
(jim@thrush.stanford.edu) (415) 723-4940 eugene@eos.UUCP (Eugene Miya) (11/29/89)
In article <JIM.89Nov27133735@baroque.Stanford.EDU> jim@baroque.Stanford.EDU (unknown) writes: >In article <5646@eos.UUCP> I wrote: > > But remember to make the distinct between education and research. > >Actually, I think there are three distinct areas in which computer >aided "visualization" is a useful tool. >NerdVis, EdVis, and ShowVis. I gave a couple of days of background thought: no problem with NerdVis, so the distinct is between EdVis and ShowVis. EdVis is important, too, but I have a problem separating it from ShowVis. It's not clear that you need to have such a big difference between the kid learning in school and the funding CongressCritter. It's learning, and it's also possible to have too much flash. Perhaps it was because I grew up in Southern California not far from Hollywood. I've seen too many films recently which have taken historical liberties with real events (The Right Stuff, Capone, etc. and also liberties with science [name your own]). There are also the excellent exceptions. I have nothing against Show Business as a source of enjoyment, but truth must NEVER be compromised. EdVis must be inexpensive, but it does not mean it has to be low quality. ShowVis can also be seen as an insult to intelligent viewers as well. Let's just make it a function of education (a semantic difference). Education requires the use of progression and background info leading to THE result. Anyways, for the next meeting of BA/ACM/SIGGRAPH/TIGSV, we will have some "real" scientists. I'll get some physicists (peaceful and less than 8), a chemist, maybe Langridge will come down, otherwise I'll get a botanist or two, etc. etc. (non-computer people of course). I believe there are very common themes between all those disparate sciences. Another gross generalization from --eugene miya, NASA Ames Research Center, eugene@aurora.arc.nasa.gov resident cynic at the Rock of Ages Home for Retired Hackers: "You trust the `reply' command with all those different mailers out there?" "If my mail does not reach you, please accept my apology." {ncar,decwrl,hplabs,uunet}!ames!eugene Support the Free Software Foundation (FSF)
rsc@altair.uucp (R. S. Cunningham) (12/01/89)
jim@baroque.Stanford.EDU (Jim Helman) says in 89Nov27133735@baroque.Stanford.EDU > II. EdVis. Educational "visualization" needs to be simple, > interactive and enticing. This means simultaneously manipulating the > parameters of a physical model and "visualizing" the behavior of that > model. Unlike research level visualization, educational models are > usually simple enough that pedagogically useful and graphically > compelling simulations can be made on almost any PC or workstation. > To date, the software to do this is sadly lacking. After all, who > develops it? A high school teacher here, a professor there. Each > doing his own thing for his own class with little cooperation or > sharing. > > The tremendous potential that EdVis has to make math and science more > interesting and accessible has not been exploited. Unfortunately, > there is little financial incentive for commercial development of such > software. I also know of little governmental support. This is an excellent statement of one of the overlooked opportunities of computer graphics. Lots of people have written lots of "his (or her) own thing for his (or her) own class with little cooperation or sharing" and it shows -- we really aren't making much progress in the area. I know SIGGRAPH has an interest in this area and has just supported a project which is developing a volume for the Mathematical Association of America on Visualization in Mathematics with a strong educational emphasis. Dr. Walt Zimmermann (math, U. of the Pacific) and I are editing and collecting this volume. I am sure that much more needs to be done here and encourage anyone in any discipline to contact me with ideas. I'm especially interested in persons who want to make something happen, but plain discussions are still welcome. And if we might get together at SIGGRAPH '90 or somewhere else, we just may be able to get something to happen here. Steve Cunningham Computer Science Department CSU, Stanislaus Turlock, CA 95380 (209) 667-3176 (ans machine)
rick@hanauma.stanford.edu (Richard Ottolini) (12/01/89)
Jim Helman at Stanford divides SciVis into NerdVis and EdVis. I think this distinction is mainly one of quality. In our laboratory our better research graphics software migrates into courseware. Better is defined as generalized to many datasets and an easy-to-use interface. These are not toy programs, but industrial strength crunchers manipulating tens of megabytes of data. This happens because our lab director is committed to teaching despite the indifference of federal and industrial grant agencies. Interactive graphics software is difficult to write. Better toolkits are needed to hide unnecessary details of the computer for the average scientist. People at some workstation companies and the supercomputer centers are developing some interesting things, but there is a ways to go.
eugene@eos.UUCP (Eugene Miya) (12/02/89)
Hit 'n' if you are not interested in the topic. In <7153@portia.Stanford.EDU> rick@hanauma.UUCP (Richard Ottolini) writes: >Jim Helman at Stanford divides SciVis into NerdVis and EdVis (and ShowVis). >Mainly a differences in quality. Exchanging some mail with Jim, I better understand what he is looking for: Jim is seeking a greater degree of interaction in the Ed and NerdVis than say the ShowVis. NerdVis has to be easily usable by a user. It should be quick and dirty if the case is needed. If Ron Levine arranges of the January meeting location and date, I assume I can make it, then I will try to fill out meeting content. I am already overloaded. We can talk about it there. I would hate to see this Bay Area TIG die. rsc@altair.csustan.edu (R. S. Cunningham) wrote: >This is an excellent statement of one of the overlooked opportunities of >computer graphics. Lots of people have written lots of "his (or her) own >thing for his (or her) own class with little cooperation or sharing" and >it shows -- we really aren't making much progress in the area. There are several reasons for this: the original report was good in this respect: there are the dilects of different programming languages, operating systems, hardware; of course NO ONE IS WILLING TO CHANGE. What's being done isn't science in some cases. Call it pre-science. Science is a somewhat fleeting thing. It's also at time some what competitive. Cooperation is kind of a new thing as well, e.g., tenure is a one person evaluation. The problem is science does not get a lot of support, either in money or resources or even pats on the back. Another things that people see is that when you write a code, there is that potential for money. They think about making more money. In the past, scientific information had a somewhat greater degree os sharing (not always). It didn't have an immediate economic value. I think in computer science the computer graphics and artificial intelligence people have recognized the value of their codes more than anyone. This has hurt the non-computer sciences as well as CS to some degree. You don't see Richard Stallman writing graphics code (might be nice if we had a Stallman for graphics code). I digress, but one more If you did, you might get some interesting consequences. Consider a non-graphics "Back to the Future"-type case: we send "Biff" back to the late 1940s to prevent AT&T from licensing the transistor to Sony. AT&T has clearly benefited from the basic research (science) of the transistor but it took Sony to develop it. What about all the other science which depended on it being cheap, etc. We will never know. Science cannot afford to have considerations like this, its information must flow freely lest we hinder it be it for money, morals, etc. The character of science is changing. It's more international. The exchange of information is greater than ever. You (we) now want other disciplines to see what we do. Now the question is how to capture some of this in a meeting? The following is really off the wall. Hit 'n' if you think the above is too off the wall. What we are doing is a science of studying how sciences do their science. We need a representative sample of sciences (the "four" basics). It occured to get a casual sample of my off-hours friends. Why not? Don Knuth (March 1985, Amer. Math. Monthly) grabbed a sample of books off his shelf and looked at Page 100 of each one to understand math and CS. So I thought I would ask Bill Burke, skiing and climbing partner, cosmologist, author (2 books), Godfather of the Santa Cruz Chaos Cabral, to come over. Bill computes on a PC. Bill's kind of a neat fellow, he's not a mainstream physicist, I have steered him to write YACC and LEX code. He doesn't way to fly around black holes, he wants to understand what they are doing. He once traced rays thru the bottom of a beer glass. An interesting contrast, a more main line physicist is a friend at LLNL. Solving the diffeqs. He won't be able to talk about his work directly, but he is an interesting fellow and completely on the opposite spectrum from Bill. It has been said physicists are the polo players of science (prima donnas) [Feigenbaum]. A third and fourth set of friends could talk about the biology of wine. These two have made interesting statements to me about mitosis going on within the biology community between mainline biology and the "systematics" people, the gene splicers. Bob Langridge would be a good case as well. So forth with a chemist, others, etc. Funny thing with one exception I do lots of outdoor activities with these guys. Thru the years I have met them on a yearly Caltech ski trip. I think they all have a common set of threads. Oh yes, I think I would like to say a few words about computer performance measurement (CS is a science) and remote sensing (past life). Some of these people do not like computers. Now how do we do it? We each talk about what and how we do our sciences. Math is part of the unifying theme, but we also have to get away from it. That is something I will have to put together. It's not "How computers will help you do science," but "how do you do science right now?" More specific questions (I have several), please send to me. Another gross generalization from --eugene miya, NASA Ames Research Center, eugene@aurora.arc.nasa.gov resident cynic at the Rock of Ages Home for Retired Hackers: "You trust the `reply' command with all those different mailers out there?" "If my mail does not reach you, please accept my apology." {ncar,decwrl,hplabs,uunet}!ames!eugene Support the Free Software Foundation (FSF)
jim@baroque.Stanford.EDU (James Helman) (12/05/89)
Concerning my distinction between visualization for scientists
(NerdVis), students (EdVis) and promotion (ShowVis),
rick@hanauma.stanford.edu (Richard Ottolini) writes:
I think this distinction is mainly one of quality. In our
laboratory our better research graphics software migrates into
courseware. Better is defined as generalized to many datasets
and an easy-to-use interface.
Certainly given software which is totally complete and perfectly easy
to use, running on hardware which can perform arbitarily fast
computations and instantaneous graphics rendering, the distinctions
vanish. When that day comes, my workstation (or workbook) will only
need to run one metaprogram.
As we approach that day, large simulations can be fully integrated
with interactive visualization. Slick photorealistic rendering can be
integrated with extensible measurement tools. But this is still quite
a ways off given limited resources and different priorities. EdVis
needs to have low cost, and interaction with the simulation; problem
complexity can be sacrificed. NerdVis requires the handling of large
data sets and capabilites for accuarate measurement; glitzy graphics
aren't as important. ShowVis needs to have a professional look with
higher quality graphics, good narration and sequencing; getting
precise quantitative information is not vital. Here are the relative
importance of some capablities as I see them (opinions will vary):
Priorities
EdVis NerdVis ShowVis
Interactivity High Med Low
Measurement Med High Med
Slickness Low Low High
As long as hardware and software resources have to be allocated from
limited pools of $$$, talent and technology, I think the best systems
will be those designed for one particular purpose.
Jim Helman
Department of Applied Physics P.O. Box 10494
Stanford University Stanford, CA 94309
(jim@thrush.stanford.edu) (415) 723-4940