It’s certainly gratifying as an author to hear about a disputed fact instead than a plain old error. Some books have whole web pages for errata, but thanks in part to the professionalism of Peachpit (a.k.a. Adobe Press), the publisher, as well as my technical editor Alexandre Czetwertynski (Parisian filmmaker now in L.A. who got the gig due to thorough notes emailed about the previous book) AE 7 Studio Techniques seems to be more or less error free (and wow, it’s hot up here near the sun!).
Presenting at Siggraph this past year in Boston, I had the pleasure to see a bunch of artists and technicians, some of whom I’d corresponded with but never met; I also made new acquaintances, including Peter Lu, a Harvard scientist, who has been discovering new research possibilities using tools you probably know well: After Effects and Photoshop. These apparently haven’t been put even close to full use in scientific research – ordinary technologies such as motion tracking turn out to have breakthrough scientific applications (a potential topic for another post).
Peter read the book (and even contributed the current most popular Amazon review) and had one factual dispute, quoted in full:
One small note, and I’m afraid the physicist in me couldn’t quite keep quiet, but I don’t think your physics explanation of the need for feathered alpha channels on page 109 is correct. If I understand correctly, you’re saying that mass is bending the light around objects, thereby making their edges fuzzy. This is Einstein’s theory of general relativity. However, while this effect happens, I understand it to be way too small to observe in ordinary light. That is, the light-bending effect of the masses of everyday objects (say up to multiple tons) are just too small to observe. In fact, the first experimental confirmation of general relativity, by Eddington around 1919, involved using a telescope to look at the slight offset in the position of some far-away stars that appeared near the surface of the sun (because of the sun’s position at that time) during a total solar eclipse. So even with the mass of the sun, you need a telescope to see how the light is bent from faraway stars. Another piece of corroborating evidence is that, when you look at the sky at night, and the moon passes overhead, say during a new moon phase so you only see a small crescent and the rest is dark, you should see the the light from the surrounding stars bent by a huge amount if general relativity were powerful to affect ordinary objects of a few kilograms, as the moon weighs 7 * 10^22 kg. Yet you don’t see any distortion, and the mass difference between the moon and a baseball is roughly the same as the difference between a cup of water and a single water molecule. So I don’t think general relativity is playing any role at all.
So what causes the blurry edge, and the need for the feathered alpha channel? I think classical optical behavior is working here, and the blurriness potentially may be due to slight imperfections in camera optics causing a slight defocusing due to the fact that there is a slightdifference in how different colors are focused. Good lenses are apochromatic, of course, and should take care of most of this, but no piece of physical glass is perfectly corrected. It might also be due to illumination not originating from a point source, but rather having some spread, like the width of a filament inside a light bulb, that causes light approaching an edge or a corner not to be exactly parallel, but spread slightly over a distribution of angles. That would also blur an edge slightly. I’m not exactly sure about these explanations, but I’m almost positive that bending of light due to mass is not playing a significant role.
I was reminded of this again last week reviewing the debut Digital Color Theory 201 class at fxPhd.com – Lorne Miess pointed out that light “bends” (or one might more accurately say, diffuses) everywhere in nature: shadows of hard-edged objects appear soft.
Thus the phenomenon has more to do to with optics than relativity. The lens in your eye bends light to focus it on the retina, as does that of the camera on the back. Natural light is always somewhat scattered, leading not only shadows but “hard” edges to be ever so slightly soft, at the very least.