Best Friend shot 02, current progress

I last commented on this shot in August. 

Version 95, from around mid-summer. I overhauled the shot, placing the male character more center stage now. Formerly he was screen R; but it seemed a bit of a viewing strain to be passing the conversation back and forth over such a screen distance, and he had too far to travel for the grab.

Here's some overdraw I did to resolve the grabbing motion. Ref video I have for this is far too uninteresting. However, this still doesn't "land" it; the shot's fairly realistic, and this motion still needs to be toned down further for the guy. Also, his motion needs to be sudden and unexpected, so that actually calls for very little anticipation. This anticipation is way too big.

Version 130. The female character's body is now well polished, and I'm re-blocking the male character with more carefully studied details and subtleties. He's going to be fidgeting a fair bit, I've been needing to better convey a sense of nervousness with him and that will help. I filmed completely new ref video for his frames 1-245. The male character's anticipation is now more toned down.

This shot is proving to be quite complex - which isn't unexpected, of course: this is intended, as I want this to be my graduating demo's opening shot, and want to be able to demonstrate a broad scope. So there's a whole lot going on on different levels. I still haven't gotten into the facial emotive content, but it's coming along bit by bit.

Spacing analysis: Don Knotts, The Shakiest Gun in the West

Another arc-tracking study. This example of shaky motion comes from "The Shakiest Gun in the West", with actor Don Knotts.
(A longer segment from this scene can be seen on youtube here)
I tracked the tip of the gun. As an instrument, it magnifies a motion that is complex primarily in rotation; the translation would be fairly smooth. Incidentally, if this were animated it would make sense to animate this with a constraint, and it would be important to locate the COG somewhere around the middle of the gun rather than his wrist.
I think a key think to notice from this is that the shaky motion is fairly random. It does hold for about 2 frames and jump in about 1 frame, so there's a slight clustering of points. But it's more "white noise" than "pure tone" vibration.

Stewie Tennis Camp, finaled

Here's the final version of my Stewie Tennis Camp shot, which I completed in AM Class 6.
I made a frames-to-seconds chart and used Audacity to line up the sound effects after the timing was locked down. Incidentally, I later applied a 2 frame delay to get the audio to sound right. It is always best to animate to audio without a time offset, but when everything's done and over with, putting in that slight 1-2 frame delay does sometimes sound better, especially when it's a given that the camera is a bit of a distance from the subjects. Sound travels 340.29 meters per second, or just over 14 meters per frame / about 42 feet per frame. A tennis court is 39 feet from net to rear, 78 feet in total length - the assumed distance-to-camera in this shot. So, strange as it may seem, a 2 frame sound delay for this particular scene is pretty much spot-on.
Introducing audio also led to introducing a third unseen character: Pro's opponent. in keeping with Pro's girl-power, it only seemed logical that she would be besting a man, to boot. (Who knows where Kid's balls are coming from! Whoever or whatever, the source remains silent and unnoticed.)
Frames 100 - 200 were scrapped and re-done three times over before I finally got that part right. Ironic that the very segment where "nothing in particular" was happening posed a particular challenge. But then again, thinking in such a manner was perhaps the fundamental problem with that segment to begin with. Generally speaking, no moment should be dramatically vacant. "Something in particular" does now happen there, a series of subtle but clear emotional beats. 
Getting the hand-held camera to feel right took a few passes. It was initially over-active and jittery. While this may have been truer to life, it was distracting; hand-held action is best kept low-key.

Pro's hair was fun to animate. It is simply composed of AM's Tailor rig constrained to her head, a hair solution I've seen other AMmers use before. I resolved it intuitively (via "mental simulation," I like to say) by overdrawing existing playblast frames (via Digicel Flipbook), then exported and rotoscoped directly in-camera within Maya. (I used a duplicate non-animated "steadycam", because the rendercam was handheld-animated by then.) When rotoscoping like this, it helps to move the rotoscope's plane close to the camera and make it semi-transparent. Posing to the drawover is then relatively straightforward. I find it interesting that the observing mind does not appear to be very particular about the details of motion with things it does not intuitively understand, so long as the general sense of it seems to hold true. Any sort of complex dynamics (fabric, water, hair) would fall into this category, as does complex physics such as the pendulum dynamics of two or more bones in a chain. What this amounts to is a great opportunity for artistic or creative treatment in these contexts. For example, the sinuous motion of this hair in the air is not physically accurate, but it looks great this way. No physics simulation is going to obtain these kinds of results. Nor will it "know" where motion needs to be subdued and where it needs to be more active. All of which supports the point that dynamic non-living objects are also best given the art-of-animation treatment whenever possible.