I've finished Animation Mentor now. Had a blast at the graduation, it was great finally meeting friends and classmates that I had gotten to know over the past year and a half.
I continue to polish my shots toward a completed demo. I'd like to share stages on "Best Friend." This became my main "final effort" in Class 6, the last class at AM. I'll share versions up to my last submission first, then I'll share polish stages and the eventual completed version - whenever that comes about.
Class 6 was busy. A lot of time was consumed polishing my Tennis Camp sequence. This was completed, I was glad to get that wrapped up at least. Meanwhile, Best Friend became the main contender for my lead shot, and I was starting this one from scratch. I challenged myself with a decent amount of activity blended into the dialogue mix, and with a relatively long clip, 16 seconds, as follows:
Filming ref was a process. After failing dismally at acting out the female role - my body just can't move that way - my wife obliged and did a spot-on bit of acting for the role. I still ended up splicing together a bunch of clips to get what seemed best, and I also ended up with parallel ref versions. This is the main sequence just for Pond:
Oh well. Sean said he can block a scene in a day this way. Either it just takes practice or the typical scene doesn't involve quite so much physics and moving around, and is shorter anyway. Yeah, I'm not sure what to make of it all, time-wise. But there's no denying that the high level of detail in blocking, wherever I did have it, produced more accurate motion and reduced complications farther down the road. It's certainly very easy to deceive oneself into thinking there's enough motion-information in there when there simply isn't yet, when blocking.
Another thing that became clear about this method was that it's best for situations where you have straightforward reference to work from, and where the style of motion is fairly realistic. Splices in reference really call for careful attention to resolving the body mechanics in the scene, and those resolutions are fabricated and take more time to iron out.
Furthermore, it isn't just splices that throw this approach off. It's also re-timing. I ended up doing a fair bit of retiming. For this approach to work well, it's good to have reference frames loaded directly into a Maya viewport for easy frame-specific reference. I did experiment a bit with animating the rotoscope timing to make it fit, but ultimately the two no longer meshed most of the time.
At the end of it all, I finished classes with this version of Best Friend in my final submission: