Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: $Revision: 1.6.2.16 $; site datacube.UUCP Path: utzoo!linus!philabs!prls!amdimage!amdcad!amd!vecpyr!lll-crg!seismo!harvard!think!datacube!shep From: shep@datacube.UUCP Newsgroups: net.graphics Subject: Re: SIGGRAPH '85 review(chatty) Message-ID: <6700024@datacube.UUCP> Date: Thu, 1-Aug-85 12:21:00 EDT Article-I.D.: datacube.6700024 Posted: Thu Aug 1 12:21:00 1985 Date-Received: Sat, 10-Aug-85 20:29:07 EDT References: <239@weitek.UUCP> Lines: 28 Nf-ID: #R:weitek:-23900:datacube:6700024:000:1500 Nf-From: datacube!shep Aug 1 12:21:00 1985 SIGGRAPH was interesting this year. My "fave rave" was Steve Gabriel's paper on rotationaly invariant splines. His "physical proof" of 720 degrees of rotation should have been videotaped. (was it?) Here is a question; more of a homework problem: The PIXAR demo showed a "real-time" 2-D FFT on a 128 * 128 image. It looked more like 10 transforms per second, so let's say it took 100mS to do the transform. The CHAP (channel processor) board on the wall had 4 29116's, each with their own 16-bit multiplier. Assuming a 100nS cycle for the 29116, and free to guess about the PIXAR's internals, how many CHAP boards were in the system? ps: I don't know the answer. I really just want to hear people's ideas on architectures for novel FFT techniques. The choice of decimating in time or frequency is some help; but if we assume that the PIXAR has some high speed -block- access to the image-store, that could be the clue. One last SIGGRAPH note. Did everyone attending see the ABEKAS A62 digital disk recorder? It would seem that you "shaded cone heads" would need something like that to deposit your pictures into. Is this going to replace the SONY 1" still-frame widget. why/why-not? Shep Siegel ihnp4!datacube!shep Datacube Inc. ima!inmet!mirror!datacube!shep 617-535-6644 decvax!cca!mirror!datacube!shep 4 Dearborn Rd. decvax!genrad!wjh12!mirror!datacube!shep Peabody, Ma. 01960 {mit-eddie,cyb0vax}!mirror!datacube!shep