eugene@wilbur.nas.nasa.gov (Eugene N. Miya) (06/15/90)
A meeting was just held here at Ames on the topic of scientific visualization, where it's going, etc. There was a lot of discussion about the hype aspects, is it just a solution looking for a problem? how to get real work, etc. The visualization mailing list was brought, up, and so forth. I have a retorical question, may not be a good one, but science asks questions, re-poses problems and creates new models. See if and how you answer it. You can mail (I will gather and repost), or post yourself. Many examples (analogies) were given at the meeting (I was only able to attend part of the last day) and I draw upon them. The question concerns one of those Back-to-the Future what-If questions where I am fortunate to have a bit of additional information. The question concerns one of the most important scientific discoveries and one of the best documented discoveries in recent years. The question concerns to discovery of the structure of DNA. There are several well known books on this, a film, and many of the major participcants are still alive. If one would have sent a modern graphics workstation (with documentation and full color), back in time to 1952, would there have been signficantly different change (new discovery, etc.) in the way Watson and Crick would have uncovered the structure of DNA? Would they have gained further insight? No modern biosoftware, of course. Note: I didn't cross post this to sci.biology. That would be cheating. The question is directed to graphics people. The bio people should sit back and watch. Note: imagery played an important role, the work was done using X-ray crystallographic pictures. [Part of the idea came to me because of a comment in one meeting about radiology]. Remember this part of the relucance to use color in this field. Too many unimportant details. Note 2: W&C spent a lot of time manipulating ball and stick models. Would they have gained the insight from a 2-D screen as they did with their 3-D models? Note: I know that this type of pattern matching is over an order of magnitude harder than string pattern matching, I've a reference from the human genome project. Is the interaction from mice/dials sufficient or were W&C using solid balls and sticks important to their final discovery? I bring this up because I have solid models sitting on my workstation and no 3-D rendering will replace them: not stereo, nothing. Will visualization get get as far because it doesn't produce hardcopy output like Alpha1 hardcopy? Also note: would one of those "Reniassance" teams as proposed with programmers, artists, etc. be necessary to have had the two of them make their discovery? [I'm skpetical, but don't let me stop you from posting.] Okay you can send the team back in time with your box if you must, just no memory of the bio-future. 8) Since the BA/ACM/SIGGRAPH TIGSV meeting was brought up, I should note that for those who were there, Peter Ray was at the Cavandish when W&C did their work. Calling Peter is also off limits. Would Watson and Crick have learned more, faster using a workstation than their simple ball and stick models? Why? How? --e. nobuo miya, NASA Ames Research Center, eugene@orville.nas.nasa.gov {uunet,mailrus,other gateways}!ames!eugene Should I offer prize money for the best answer?
rick@hanauma.stanford.edu (Richard Ottolini) (06/15/90)
In article <6719@amelia.nas.nasa.gov> eugene@wilbur.nas.nasa.gov (Eugene N. Miya) writes: >A meeting was just held here at Ames on the topic of scientific >visualization, where it's going, etc. There was a lot of discussion Scientific visualization has been absolute essentially in the petroleum exploration, both in the analysis of 3-D data and devising algorithms to improve the appearance of the data. Before 1975 seismic prospecting was essentially irrelevant in the discovery of oil. Seismic pictures did not well match what was inside the earth. Since then geophysics has gotten so much more accurate that we are discovering oil "too fast" resulting in a worldwide glut. One of the more important tools have been the surface-slicing volume visualization tools used for past dozen years or so. We could neither understand the complicated 3-D interrelationship of geologic objects nor aspects of seismic wave propagation without these tools. I think the remote sensing / weather satellite people and medical people could make similar claims. Note that these three cases involve OBSERVATIONAL DATA ANALYSIS. Many visualization demos use PSEUDO DATA generated by modeling / simulation programss. Many of this program generate weak or unconvincing results because the science is imperfectly understood or the computers aren't powerful enough yet. Garbage into a visualization program will result in a scientifically unconvincing picture. The goal of scientific visualization is insight. Jazzy display features or poor inputs don't fool scientist for long.
leech@homer.cs.unc.edu (Jonathan Leech) (06/15/90)
In article <1515@med.Stanford.EDU> rick@hanauma.stanford.edu (Richard Ottolini) writes: >Note that these three cases involve OBSERVATIONAL DATA ANALYSIS. >Many visualization demos use PSEUDO DATA generated by modeling / simulation >programss. Many of this program generate weak or unconvincing results >because the science is imperfectly understood or the computers aren't >powerful enough yet. Garbage into a visualization program will result in >a scientifically unconvincing picture. It can go the other way, too (pseudo data generating very convincing results). A few years back, I borrowed a video of a simulated Mars-sized body impacting with the proto-Earth from David Stevenson (Caltech Geology) to show to other people in the graphics group there. While talking to him, he mentioned that many planetary scientists seemed more prone to accept the impact theory (of lunar formation) after seeing the simulation. Apparently their critical skills could be suppressed by the pretty pictures. If the trained people in a field are prone to this, how much more impressionable is the naive (in that field) person to fancy graphics? Well, the impact theory seemed very reasonable to me after watching the film. -- Jon Leech (leech@cs.unc.edu) __@/ ``Even by the 22nd century, no way had yet been discovered of keeping elderly and conservative scientists from occupying crucial administrative positions. Indeed, it was doubted if the problem ever would be solved.'' - Arthur Clarke, _Rendezvous with Rama_
MELTSNE@gecrdvm1.crd.ge.com (06/15/90)
A convincing simulation will tend to do an end-run around a scientist's analytic capabilities. I did some little movies in Videoworks that made a real difference in getting people to accept some simulation results of mine.. And as for 3-D pictures vs. models, the pictures do work if animated slightly. The depth cues from motion (flip images of 2 degrees left, center, 2 degrees right, and back) work well for me -- I have lousy depth perception. On the other hand, I like models made with balls and sticks, especially if the balls are a decent brand of gumdrop. You don't believe in tetrahedral interstices until you've eaten a few. Ken
montnaro@spyder.crd.ge.com (Skip Montanaro) (06/15/90)
In article <6719@amelia.nas.nasa.gov> eugene@wilbur.nas.nasa.gov (Eugene N. Miya) writes: If one would have sent a modern graphics workstation (with documentation and full color), back in time to 1952, would there have been signficantly different change (new discovery, etc.) in the way Watson and Crick would have uncovered the structure of DNA? Would they have gained further insight? No modern biosoftware, of course. .... Note 2: W&C spent a lot of time manipulating ball and stick models. Would they have gained the insight from a 2-D screen as they did with their 3-D models? .... Would Watson and Crick have learned more, faster using a workstation than their simple ball and stick models? Why? How? If you consider the structure of DNA (using hindsight, of course), its fundamental building blocks (adenine, cytosine, guanine, and thymine? my biochemistry is real rusty) are not all that complex, there are a limited number of stable ways to put them together, and they are few in number. Watson & Crick (and Pauling, for that matter - he was perhaps blinded by his earlier work on protein structure) knew what they were; they had to figure out which way matched the x-ray crystallography results. I doubt a modern graphics workstation sans Biograph would have been very helpful. On the other hand, consider the flow of gases through a jet engine, the ultrasonic energy profile over the surface of a part being machined, or the temperature profile over an airfoil. I submit that the "building blocks" of those problems (however you care to define them) are somehow much more complicated than those of DNA. In cases like that there's no substitute for the computer`s ability to quickly create a picture that has a strong tie to physical reality. Put another way, there are no simple ball-and-stick models for many of the problems being tackled by so-called scientific visualizers. -- Skip (montanaro@crdgw1.ge.com)