RE: stillness capture

From: Richard Widgery (motion@widgery.demon.co.uk)
Date: 11/16/01


Scott,

I have uploaded something onto my website that should help to describe the
motion within stillness in terms of motion capture for the games and feature
film industries....

http://www.kinetic-impulse.com/text/download_movies_free_motion_capture_data
/experimental_art/human_motion_capture_stillness.html

Sorry about the rather long URL.



As for developing software, I could do that; the movie above is from my
proprietary motion capture & control software:
http://www.kinetic-impulse.com/text/services/custom_tools_development.html


The use and implications of this kind of still motion should be clear from
the example above, but I'll try to explain in words for anyone who may have
trouble viewing this clip that requires Microsoft IE & Media Player.

On the left (of the first web-page) you will see a still frame of a motion
captured actor.  On the right is a very short movie clip of the same actor
doing practically nothing at all.

In terms of believability I want to break the subject into three scenarios,
for video games, feature films/TV/movies, and robotics.


1) Games
--------
The image on the left is "life-less" and leads the player to start to ask
questions as to whether there is something wrong with the game / computer,
"Has it frozen or crashed?" because of one simple point .. there is nothing
moving at all.  (Robbie Shaw is correct to say it is just a "pose").

OK, what if there is something else moving in the scene such as a background
object?  Then it is probable that the audience will start to ask "Why is my
character not moving AT ALL if everything else is?  Is there something wrong
with the game?"

These are both usually answered by user interaction, they would press some
buttons to get something to happen, but it does not detract from the fact
that the user has already asked the question.  This would lead to a
disjointed experience in terms of "absorption" into the fictitious
role-playing scene that most games try to create.  No living thing is
totally static in real life and a game tries to envelope a players senses as
much as possible, so to have started down a road of believability only to
suddenly question why this character has become totally life-less is
detrimental to the player's "fluid" enjoyment of the game.


2) Feature Films / TV
---------------------
Similar to the game explanation above, the audience can start to ask
questions such as "Is this a new movie effect technique, and if so why is it
being used?"  This is more of a narrative question than a believability one.

We are so used to 25 frames per second (30 USA / 24 film) that to suddenly
have what looks like a still picture, the audience want to know "why?".
Even a little movement is better than none at all.  This can be seen very
clearly from cellulose grain.  I personally dislike digitally slowed down
footage that has been shot from film because film grain is visible noise
(and sometimes very aesthetically pleasing noise), and by slowing the
footage down you slow the frequency of the noise down, and the technique
becomes immediately apparent.

Now take this point into consideration for a "life-less" motion captured
character being show at 24fps including the film grain noise.  The audience
would ask (even though subconsciously) "Why do I still have 24fps noise but
0fps motion?"

Noise in motion is a natural part of our visual diet, and to remove either
the grain or the movement leaves the audience with an unsure state of
security about the intention whether it be technical difficulties or
artistic direction.


3) Robotics
-----------
Scott writes:
> "Given that there is so much use of dynamic motion synthesis in
robotics --
> are there dynamic stillness states that a robot slips into between tasks?"

Generally .. no.

There are a number of problems concerning the control of robots with dynamic
motion and stillness including inertia, friction, stiction, electrical
interference, and noisy gears.  These collectively provide difficulties in
obtaining the exact motion required.  Fine Tuning a robot has to be done
within a specified range of parameters including acceleration, braking, and
velocity ranges.  It is easy to tune for one, but the problem exponentially
expands if trying to tune for all eventualities.

Let me give an example:  If you construct a robot armature such as a replica
of a human right arm and you want that arm to be able to move at 2m/s
velocity with an acceleration of 10m/s2 (i.e. a very fast reacting arm that
can move with high speed) you will have difficulties trying to get that arm
to do a very slow and subtle movement.  You have to construct the armature
and electronics in such a way that can respond to very fast control signals.
If you now try to get that armature to represent stillness you usually end
up with a jerky and erratic (noisy) result unless you implement large
dampening constraints - but by applying dampening you reduce its
responsiveness.

Confused yet?  Basically its a catch 22 situation, the more dynamic your
robot the less able it will be cope with subtle and delicate motion, and
vica versa.

There is a solution that allows you to implement dynamic tuning that can be
taught what kind of motion to expect and re-tune on the fly, and many of
these are providing excellent results - but in general for feature film
animatronics you generally know what kind of motion you require from your
robot before you build it.

Dynamic stillness is applied to some of the advance animatronics we have
done such as is the feature film of Mary Reilly - Jekyl and Hyde where we
motion captured John Malkovich's body and applied the data to an animatonic
robot and computer graphics model.  In the transitional scene where he
change from one character to the other, the shot comprised of both still and
rapidly moving torso motion (representing the mental cognitive time to work
out that something was happening to his body, and a painful and "thrashing"
motion of the transformation).

As for using dynamic stillness for "task transitions", I would say that
depends on the duration between the tasks.  If a robotic character / avatar
was to run in to a scene, stop for a few seconds before doing another action
then I would definitely implement dynamic stillness to maintain the
believability of the characterisation.  If the time between the two moves is
under 1 second, I would guess that there would be so much inertia of
overcome anyway, it is probably not worth the effort.



Question:
---------
Scott writes:
>>Richard, to what extent can 'stillness' be synthesized? Could this be
quite easy?

Stillness can be synthesized from learning about the body motion - I don't
see a problem is developing tools to learn anything the user wants from any
performance.  Every-"body" is different and as such any synthesis would be
just one interpretation.  Part of the interest of working with different
motion capture actors is their individuality that they bring to a
performance and would like to ask what context/application you would apply
synthetic stillness.



Footnote:
---------
I believe that motion in stillness is an essential component of creating a
performance but on it's own is it is really interesting.  Like the narrator
that uses pauses in speech to good advantage to accentuate dramatics, the
speech is nothing without the actual spoken words.


There have been a lot of very interesting points brought up by other people
in the newsgroup and will try to comment on some of them in future replies,
but I hope that this one has helped to explain some of the uses of stillness
in motion capture and look forward to your comments.

Regards,

Richard Widgery
richardw@kinetic-impulse.com


KINETIC IMPULSE - Digital Performance Art & Science
--------------------------------------------------------
http://www.kinetic-impulse.com
--------------------------------------------------------






-----Original Message-----
From: owner-dance-tech@lists.acs.ohio-state.edu
[mailto:owner-dance-tech@lists.acs.ohio-state.edu]On Behalf Of Scott
deLahunta
Sent: 15 November 2001 10:24
To: dance-tech@lists.acs.ohio-state.edu
Subject: RE: stillness capture


Hi richard --

Thanks very much for your contribution... during my relatively few
opportunities to work with professionals in the animation industry I have
never seen anyone capture stillnesses in that way... so I appreciate your
informing us about this based on your more extensive experience. How
capturing people in a state of not moving can be useful to maintaining the
narrative and believability of the game is quite interesting. Richard, to
what extent can 'stillness' be synthesized? Could this be quite easy? Given
that there is so much use of dynamic motion synthesis in robotics -- are
there dynamic stillness states that a robot slips into between tasks?
Perhaps these are silly questions, perhaps they might be explored in an
interesting way... I'm not sure.

Besides maybe using 'stillness' as a way of asking a question about robots,
I was also proposing to perhaps extract stillness (like Rogalsky) from
movement -- to see what it might tell us about the event associated with
those motions. So, could someone write a program that would say extract and
assemble all the moments of stillness (within parameters) from a given
motion captured sequence? Defining these parameters would be quite a
challenge...

A dance maker/ performer might define a typology of stillnesses/ different
forms -- the stillness just after stopping, the stillness just before
starting, nearing stillness/ deceleration, moving away from stillness/
acceleration, the stillness of trying not to be seen, a symbolic stillness
of resistance, different stillnesses in different body parts, the stillness
of waiting, the stillness in meditation, the stillness that sets in when
foregrounding another sense (i.e. listening)...

Dance makers interested in perception (of the watcher) may use stillness in
the same way John Cage did with 4:33 -- because your ear doesn't focus on
the sound of the piano you attend to all other noises. Cage was explicitly
political with his silence to enable an emancipation from traditional forms
of music. Pina Bausch or Vera Mantero choose stillness at certain points to
invite a particular way of seeing. Within the realm of experience, Steve
Paxton's "small dance" (something one tends to do instead of watch) is
sensing the movement in stillness -- exactly the sort of no-motion you
refer to (Danny Lepkoff has written an article that looks at stillnesses in
a recent issue of Contact Quarterly).

So, anyway -- enough on stillness. I have also enjoyed the banter and did
my own back channel response to Bruno's e-quip about getting money for
'doing nothing' -- so just to prove that I do have a sense of humour.

**Dear Bruno: and I was also thinking that perhaps this could all be
synchronized with the stillnesses that happen in the silences in order to
optimize the
moments of emptiness and generally contribute to the accumulation of
considerable amounts of money.**

------- but I have less funny things to say in response to Christy's
comment that perhaps there is some connection with a "rarely addressed"
lack of content issue. Just as with Richard's proposal that there is "no
such thing as no-motion" -- there is no such thing as no content. Please
give me an example of where you might perceive there to be a 'lack' of
content and what are the means for identifying this lack? what evidence do
you look for? I would be happy to read something of interest on this topic.

Best,

Scott



This archive was generated by hypermail 2b30 : 01/24/02