CompuServe Thread

OK. I'll work up a few test-cases, benchmark them, and run 'em down at Corky's place this week. Flame and Vapor sound like good candidates. -Alan

Tyler, Our company has done a bit of FE modeling in the past. In fact our company president did an FE model of the human brain many years ago. But as you know, that kind of modeling can complicated and tedious. We do not generally use it for litigation for several reasons. Mainly because of the higher burden for admissibility. Such an analysis is really a simulation, and not an animation. Animations are used mostly as demonstrative evidence to illustrate an expert's testimony. Simulations, on the other hand, can prove facts indepently. For that reason they are much more difficult to get admitted into court. The reliability of the computer code and the data are called into question. The computer code itself sometimes must be submitted far in advance of trial for the other side's expert to review. This can be a bit of a problem if the code is proprietary. It must also be shown that the model can be validated. Then there is the accuracy of the input data to consider. These issues by theselves can lead to pretty extensive litigation. Basically your opening an expensive can of worms. For that reason a lot of attorneys shy away from it. I've personally heard of a case where an attorney got burned to the tune of $100,000 and had his simulation kicked out of court. There are a lot of Recon people who use the EDC programs such as EDSMAC and EDCRASH. We use them too. These are 2D accident reconstruction programs. They have proven themselves over time in many jurisdictions and are now widely accepted, so admissibility is not too difficult. However, even with these programs, your input data is still open to scrutiny. We do do quite a few simulations, but we use a widely known (in the Auto Industry) and used mathematical modeling software package called MADYMO. It is not a finite element model–rather you work with lump masses defined as either ellipses or planes. You can assign surface elasticity values, joint stiffnesses, object mass, center of gravity, velocity…etc. You can even integrate FE models. The real beauty, though, is that every automobile manufacturer uses it in designing their cars! Therefore, they cannot object to its reliability and, thus, admissibility. Recently we even got it admitted, without objection, in a federal criminal case. We have have been looking at ways to link MADYMO and 3D Studio through IPAS–thus making 3DS a sort of post processor. MADYMO looks a little rough since it just shows you a bunch of animated colored ellipses. 3DS could dress it up nicely for court use. But, on the other hand, I had one attorney that preferred the "technical, computer look" of MADYMO because he thought the jury might belive it more. So who knows. Anyway, I don't know what to tell you about getting into FEA, other than it could be a real battle. You might start simple, just testifying as an engineer. Do some 3DS animations to illustrate your testimony. Then write some custom IPAS routines which add the effects of gravity, friction, mass and collision detection. Then you could sell your routines to the rest of us 3DS users! Hope this all helps -Parris

<Have you got any interesting forensic work for the Siggraph reel? (besides police brutality cases, I hope)> I had a great one – It was a police butality case though, but this was against the police. I was all excited about putting it on my demo reel. Unfortunately, the case settled for millions of dollars and the condition for the settlement was that I could not show the animation (especially to the Press) or discuss the facts of the case to ANYONE for five years–so save room on the Siggraph '99 reel! I have another which resulted in huge settlement from GM, and another which is kind of humorous. I'll see if I can't dub 'em to beta and send 'em in. As to that infamous case, this week it finally ended for me. We've been working for the city in the civil case, which as you no-doubt have heard, is finally over. Thank goodness. -Parris

Mei-Tien, In the "Advanced CMOS Setup" configuration I set the parameter for DISABLE SHADOW MEMORY SIZE to 64kb and the line below that –DISABLE SHADOW MEMORY BASE to D0000h. Hope that helps. Parris

<<I have been getting 250% to 275% speed increase using a pentium 60!! What gives???>> I really haven't done a scientific study on this. I have a couple of 486-66 machines with 40mb of ram, a few 486-50s, and now a Pentium-90 with 40mb of ram. What took me around 3 minutes a frame on the 486-66s now takes about a minute on the Pentium. I bought the Pentium 90, not just for the speed, but for the improved (cooler) chip design. I really wanted jump at those 60s and 66s but figured I was better off waiting for the 90. I bought a Gateway since both Dell and Gateway (and a few others) use the same Intel motherboard–no manufacturer had a clearly superior design. Compaq is too expensive and I've been burned by ALR on compatability in the past. So far I have been happy with the Pentium. It does have the PCI chip set COM port bug but I can work around that. Maybe if time permits I will do some serious benchmarks, and if I do I'll get back with you and pehaps compare with it your 60. -Parris