Having worked directly with motion-capture technology myself, on a variety of projects (with both 'good' results, and 'bad') .. I was prepared to offer some of my own viewpoints, based on my experiences.
Until I clicked into the thread, and saw that the first poster's mind was already vehemently made-up .. that motion-capture was not only aesthetically 'displeasing', but
EEE-ee-e-e-villl-l-l-l, somehow.
Ben's last post kinda sums up my own reaction to that attitude.
----
EDIT: this was originally all I was going to say. But I figured, what the heck, I'll offer my views, anyway:
------------------
IMO, mo-cap is better-suited to 'certain' types of projects. In particular, video games, live-action visual-effects, or forensics. It is an excellent tool for any project whose primary goal is realistic, physics-accurate motion.
But,
99.9% of the time (
no exaggeration), mo-cap is used with prodigious human 'assistance' and input:
Most systems place a keyframe on
every frame, for each marker .. often with multiple sets of data per marker; this tremendous amount of data produces 'jerky' motion when played back 'raw', at speed. Most of these keyframes must be discarded, so that only the 'key' extreme positions for each marker remain .. resulting in smooth 'arc-ing' motion graphs. Also, of course, markers which become 'obscured' for any reason (ex: passing behind an arm or leg) must be interpolated. While many studios do use code/scripts to wade through a certain amount of these 'clean-up' tasks, a human is ultimately responsible for what data gets kept, and what gets thrown away. Also, of course -- particularly for characters who must interact -- the motion data must be re-timed, or even enhanced with hand-keyed animation, in order to synchronize the performances. Trouble often comes when those tasked with data 'clean-up' are not necessarily the best 'animators' themselves, or even (especially at some 'cheap-o' overseas studios), seem not to have actual 'artistic' skills of any kind.
The quality of performance from the actor is also a major factor. I've found that the best people for this sort of work usually have some form of dance or gymnastic/acrobatic training, which makes them 'hyper-aware' of their body movements; also, of course, pantomime acting ability is extremely important.
None of that makes up for the fact that humans simply
do not 'move' like cartoon characters can & do. What looks aesthetically pleasing when drawn (or hand-keyframed) is always the result of carefully-planned-out arcs and judicious use of squash-n-stretch, ease-in/out, and any number of other tried & true principles which produce appealing (though, again, often
not 'realistic') animation.
This is why, IMO, mo-cap is not very well-suited to cartoon-styled animation. The 'disconnect' between how cartoon characters are
expected to move, and the rather clumsy way that real people
actually do move, is enough to place the result firmly in the 'uncanny valley'.
------------------
The main reason that Hollywood will never "put an end to motion-capture" is that it
can, indeed, save productions time and money.
However, less money and less time almost necessarily means you'll have a 'lesser' end result than hand-keyframing (which costs more money, mostly
only because it takes more time).
And, of course, when more time
is taken to 'massage' mo-cap data .. the time
and money savings is 'nullified', and the end result can
still be less-than-pleasing (for most of the reasons already stated above).