Drexler on How Nanotech Animations Should Be Slower Tuesday, Feb 10 2009
nanotechnology 7:07 am
As you might have heard, Eric Drexler got a blog a while ago (last October), and has been producing some nice content. One interesting post from December was about how Drexler considers diamond mechanosynthesis (atom-by-atom diamond fabrication) a bad short-term objective, instead arguing we use protein or pyrite, which are easier to work with. He dispels the notion that protein would be a ridiculous material to use for mechanosynthesis (was there one?), encouraging us to think of protein as horns, made of hard keratin, rather than meat, which is over a million times weaker. Horns and meat — who said nanotechnology wasn’t exciting?
Today’s post has to do with how the colorful molecular machinery gifs floating around on the web mislead scientists into thinking that the very idea of MNT, and by extension him, are nutty, because the gears in those animations are running at speeds comparable to their (apparent) thermal motion, which would cause them to overheat and break down almost immediately. Examples of such animations can be found on Drexler’s blog, at CRN, and a past post of mine here. The one he uses on his blog looks even more speeded up than usual. Drexler provides a video from J. Storrs Hall showing realistic gear motion — very slow relative to random thermal motion. The bearing in the last video is rotating at 1 GHz.
This most recent post is related to a post of his from December that points out some really cool biomolecular videos and remarks, “the videos lie because they must”. He writes, “They lie about how biomolecular machines move. Where they show smooth, purposeful-looking mechanical movement, the reality is instead a frenetic dance of Brownian motion.” There’s a further page on Nanorex about the “stroboscopic illusion” — the limited frame rate of animations captures atoms dancing around at only 24 frames per second, even though they are actually dancing around much faster than that. To display the gear motion in a time frame that wouldn’t cause the viewer to get bored, you have to “fast-forward” the video, producing an illusory effect of slower thermal vibration relative to the purposeful motion.




What’s interesting is that these videos can be made physically correct with minimal effort. All the tools needed for this (multisample motion blur / tempral antialiasing and the SDK for motion controller plugins) have been built into 3DS Max since Precambrian.
As for the video speed problem, there’s a simple solution — divide the video into 2 parts. The first one with perceptible rotation and blurred thermal motion, and the second one with perceptible thermal motion and ‘frozen’ rotation. Both parts would come with explanatory texts.
(I posted a couple of more detailed comments at Eric’s blog, but the first one has stuck in the moderation, probably due to length and links).