Pixels
and Lead - Part 2
Doron
Meir, a classic animator retrained for CG, examines
the relations and differences between the two types
of animation, and offers some insights, tips, methods
and conclusions.
The first part of this article discusses the methods
of working in classic 2D animation. The second part
compares that to CG.
By
Doron A. Meir
06.12.03 |
|
In
the first part of this article, we've explained
the methods of working in classic 2D animation.
So what is the difference between this process and
the method of animation in CG?
Well, the answer is: none. I work in quite the same
way in both. The difference lies elsewhere: classic
animators use papers.
Animating
on paper
• Constantly flipping the drawings - all
the time, even while drawing - with the previous drawing
(checking proportions and motion), the two previous
drawings (checking flow and arcs), or the three or
four previous ones (not as often - a deeper checking
of flow). With inbetween drawings you flip between
the previous and the next drawing. All it takes to
get back to the current drawing is a small movement
of the fingers.
• Occasionally you change the order of the papers.
It's very easy. The animator sets up a sequence of
drawings he would like to examine, and flips them
or turns on the light table to see them overlap. The
order of the papers in the work process is entirely
unrelated, of course, to their final order in the
film (contrary to CG animation, in which the work
is done directly on the timeline).
• You can easily get the onion skin effect (several
transparent, overlapping drawings) - just turn on
the light table.
Here
the differences between CG and classic animation are
quite evident. Some of it, like working independently
of the timeline, does not exist in any of the currently
available 3D software. Other features (like onion
skin effect) can be found in some of the programs,
but never as simply as pushing a button.
Why
is this relevant to CG animation? Because the classic
way of work, especially the flipping, is the very
thing that develops the famous "sense of motion"
- that feeling that guides the animator and tells
him whether a certain motion "works". A
classic animator would usually flip to previous drawings
before he even starts drawing the next keyframe, in
order to feel and imagine roughly how it's supposed
to look.
These
issues are solvable. It can be programmed. I hope
someday, someone will actually program it. But these
are differences that, important as they are in my
view, are still superficial. The deeper differences
are still ahead.
Fundamental
differences between classic and CG animation
Indirect animation
Some of the basic differences, I've noticed very
early in my CG experience. I felt, and I still do,
that in CG animation I wasn't actually animating:
I was explaining the software, through its user interface,
how I want it to animate. In other words,
the work is indirect - contrary to classic
animation (as well as puppet animation, for instance).
Every animator who ever yelled at his computer "but
I asked you to do it like that!"
will understand what I mean. Not to mention missing
or too complicated features.
Intuition
In classic animation one can create quickly, with
a few lines, a rough pose. This allows a simple, flowing,
harmonic animation process, concentrating on acting
and motion without being distracted by details. The
drawings can then be carefully redrawn, with all the
necessary details, keeping the original flow of the
free, intuitive animation. This is impossible with
3D animation: each pose includes tedious picking,
moving and rotating of different objects. In classic
animation you think of the pose, then
analyze it to its parts; in CG animation you have
a multitude of parts from which you assemble a harmonic
pose. A hand is perhaps the ultimate example: a few
seconds of pencil time, a lot more digital time.
|
|
|
|
A
hand in a few seconds. Posing a hand in 3D requires
using up to 20 objects (3 bones for each finger,
4 in the palm itself and one for the wrist).
|
A
whole pose in less than half a minute, including
character design. Drawing is much more intuitive
than 3D posing.
|
Rigging
The rigging problem accentuates the above two issues
even further. Each rig has a sort of user interface,
with its own features, missing features and bugs.
Thus another thick layer is added to separate the
animator from his animation. And on top of that, the
rigging issue is very much responsible for the intuition
problem.
A
classic animator thinks less about bones and more
about the forces working within his character. The
character is trying to reach something high? The animator
visualizes a power that supposedly pulls the character
up, stretching it through its fingers and influencing
the whole body. The character reaches to scratch its
head? The animator doesn't move the arm: he moves
the hand. The character hits the floor after a jump?
In the mind of the classic animator there is a squashed
mass, along with a sense of weight and the strength
of impact. There is nothing similar in the current
rig technology.
Illusion of motion
This is, I think, the most inert and fundamental
difference between classic and CG animation, and it's
so simple that it goes completely overlooked. I noticed
it for the first time while working on this article.
Classic
animation is an illusion of motion.
CG animation really moves. In other
words: in 10 frames of classic animation there are
10 static characters, but they are drawn and arranged
in such a way that it seems to be one character who
moves. In 10 frames of CG animation there is only
one character.
This
simple but unobvious difference has significant consequences.
In classic animation I can do whatever I like. If,
all of a sudden, just for a single frame, I want my
character to stretch, or completely distort, or turn
into a messy scribble or a tomato - anything - and
I think it fits the action, I do it like that. In
CG animation, by contrast, I am limited to the existing
character and to the "laws" made by the
rigger - that is, unless I model something entirely
new, just for that single frame, and also the inbetweens
(which is of course not practical, at least for regular
commercial projects).
This
is the reason for many advantages and disadvantages
of the classic vs. CG animation, but mainly it draws
us to the conclusion that compared to classic animation,
CG animation is inescapably a restrained medium, simply
because it's harder to distort and exaggerate when
you are trapped in a web of laws such as exists in
CG animation, and completely absent in classic animation.
More
important differences between CG and classic animation:
• Classic animation is done from a single point
of view - that of the camera. 3D animation is done
simultaneously from every point of view.
• The classic inbetweener is human, and has human
intelligence (usually). The computer understands nothing,
and has no common sense at all.
• The result in CG animation can be viewed instantly
and altered at any point, while in classic animation
it's impractical to test every minor adjustment, and
when the inbetween is done the animation is usually
not changed. Therefore classic animation demands more
imagination and planning.
• In CG animation each object has its own keys,
while in classic animation the keys are generally
set for the character as a whole.
Conclusions
General
• The spontaneous nature of drawing is irreplaceable.
CG animators, too, will gain from a solid quick-sketching
and drawing ability, which will enable them to create
appealing, interesting, flowing thumbnails that can
be used as reference for the actual animation.
• Animation is tough enough even without technical
complications, but the 3D animator is working with
double gloves - the interface and the limitations
of the software, and those of the rig. Precisely because
his work is less comfortable and intuitive, the 3D
animator has to better understand what he's doing,
whereas a classic animator can use his intuition.
• Since in CG animation you can immediately view
the results, the animator has a better chance of getting
exactly what he was after. Classic animators must
be extremely experienced to achieve this goal, especially
when it comes to timing. CG animation can shorten
the learning process, and on the other hand may cause
the animator to neglect the planning and visualizing
stage - which is dangerous in my opinion.
• Working in 3D space -animating from all angles
simultaneously - can contribute to better understanding
of pose, weight and anatomy. Nevertheless, since a
holographic medium is yet to be invented, the final
result is two dimensional, seen from a single angle.
Therefore the camera has to be set prior to the animation
process, and the animator should take specific care
of the pose from that point of view.
Software
• The rig technology is still not fully formed.
I think the animator should be given the option to
use not just his rational but also his intuition.
The objective I would suggest is for the rig to enable
work on forces (weight, squash, stretch, line of action),
leaving the work on bones for fine tuning. Also, the
entire IK scheme is problematic and should significantly
improve in order to be really intuitive.
• Several features can be added to let the CG
animator work similarly to the classic way of working
with papers. For example: an option to set a few poses
that can easily be flipped with the current frame.
There should be a hotkey that toggles ghosting (onion
skin effect) for these poses. Also it should be easy
to rearrange the selected poses without influencing
the timeline arrangement.
• There should be a simple option
of setting the nature of inbetweens (slow in, slow
out, smooth) for all the character's objects at once.
Most of the major programs now include such a feature,
but it's not simple enough, and when the character
is very complicated (includes many objects) it may
be time consuming and cause technical problems.
Work
methods
• It is well to keep the keys grouped for the
whole character as much as possible. It simplifies
the animation process, gives the animator greater
control over the final result, and prevents some technical
complications. Once the animation generally works,
you can fine tune by changing and altering the keys
for specific elements.
• Sometimes it is more practical and convenient
to animate more freely, i.e. to work on each object
separately. When the animation is essentially done,
you can group the keys by creating grouped keys (i.e.
on all objects at once) in the important frames and
deleting the rest of the keys. When correctly done,
the animation will stay almost the same and you can
refine timing and poses without getting all tangled
up.
Sample: in Test
A the arm as a whole has 18 keys, spread
amongst the many objects, while Test
B only includes 7 grouped keys.
• Occasionally I would begin the work on a scene
by setting the important poses regardless of the timeline.
I set pose 1 at frame 1, pose 2 at frame 2 and so
on. This inhibits the temptation to make premature
previews, and forces me to plan and imagine my timing
in advance.
• The stepped test of classic animation, which
takes a little practice to use, helps to better understand
the role of keys and extremes in creating the animation.
In my opinion, CG animators can also use it to their
advantage. Keith
Lango's excellent article presents a similar
method.
• Before adding the inbetweens, the stepped test
should already include arcs and timing suggestions,
including ease in/out. If the stepped test works,
the influence of the inbetweens (of which the animator,
of course, has less control) would be minor. Further
more, since the stepped test already describes the
desired motion, the inbetweens can be treated in terms
of slow in/out/linear (instead of tedious work with
curves and beziers). It is simpler, and also creates
a less mathematical and more natural movement.
• Where cartoon animation is needed, you may
want to consider:
• Simple, flowing and graphic character, without
too much detail in model or texture.
• Rig - as flexible as possible, thinking ahead
of possible distortions (wildly open mouth, violent
stretch/squash, etc.).
• Animators with classic background can draw
sketches of the animation, in order to better feel
the flow and the distortions needed.
• Once the animation is done, the model (mesh)
itself can be animated where distortions are needed.
• Don't be afraid to hold the character. The
conventional CG animation way of thinking says that
a hold does not look good in 3D. I think the main
reason for that is that contrary to drawn animation
- which is perceived as an icon or caricature - CG
animation is seen as a simulation of reality, and
in reality holds don't exist. I think cartoon will
be seen as caricature in CG as well, and therefore
the holds will hold (so to speak).
Pointing
finger: example of "classic" way of work
Since we're talking about a very simple
animation, I skipped on thumbnails and went straight
to the extremes. After having planned the desired
timing, I set the two main poses: initial and final.
Then I set the anticipation. I watched the animation
and made sure that the timing I planned roughly works.
In every keyframe there is a key for all objects.
"Extremes"
test
Now
it was time for some breakdowns. To see the arcs I
worked with ghosting (onion skin effect) and trajectory
options. I set two breakdowns to make sure I would
get the arc I wanted (in classic animation one breakdown
would be enough, with a drawn arc that the inbetweener
can follow). The second breakdown pose is quite close
to the anticipation extreme, but it's not as close
in the timeline, which makes a slow in. Between the
anticipation extreme and the final pose I added another
breakdown, defining a very flat arc and adding a "drag"
on the hand and index finger. After the "pointing"
extreme I added a recoil pose that also serves as
a moving hold. I went through the important objects,
making sure the arcs work. The result, still in stepped
keys, can be seen here:
Initial
stepped test
As
you can see the animation already works, even though
there are still no inbetweens - just like in the classic
test in the beginning of the article. Now I turned
to some specific adjustments: stretched the bones
in the pointing extreme to get some more elasticity,
and added a key for one frame in which the hand is
stretched, to simulate a blur (this is a 2D cartoon
trick. A fine example of this can be seen in Eric
Goldberg's animation for the genie in "Aladdin").
Final
stepped test
It's
time to add inbetweens. I selected all the objects
(as you may remember, every keyframe contains keys
for all objects), and changed the curves: accelerate
from the first to the second key, smooth to the third,
then slowing in to the forth (anticipation extreme).
Slow out towards the pointing, no inbetweens between
the stretched key and the pointing itself, and then
slow in towards the recoil, final pose. Made a test,
here and there I made some timing corrections (very
easy, as the keys are grouped), and the animation
is ready:
Final
test
No
tedious curves work has been made, and yet the animation
is smooth, elastic and also somewhat cartoony. More
importantly, this method does not destroy creativity
and may even improve it: instead of investing a lot
of time in curves and motion, most of the time was
spent on the initial extremes, the important ones,
those that tell the story; a little less time was
put into the secondary poses, the ones that make the
motion; and the shortest time was spent on the inbetweens,
whose sole function is to smooth out the motion. Furthermore,
since every keyframe is set for the whole character
(the whole hand, in this case) and not parts of it,
I was able to maintain high level of control over
the final result.
I chose not to go over the animation and clean up
redundant keys, or smooth the arcs in the graph editor,
because I wanted to keep the more natural timing that
you get when the arcs are not completely mathematical
and "correct".
It is important to emphasize that there is no intention
to claim that this is "the right way to work
in CG animation", but merely a possible example
of how classic animation methods may be put to work
in the CG world.
And
finally…
Finally,
animation is animation. This article examined through
a magnifying glass - no, through a microscope - the
differences between classic and CG animation. But,
when all is said and done, it's more of the same thing,
and the goal is always the same: Nice motion and interesting,
convincing acting. And so all animators, in all techniques,
are sitting in the same boat, trying to understand
where the hell we're paddling to.
I'd
like to thank Kfir
Ram from Asifa
Israel, Tal
Flint and Nir
Hermoni, and also the guys at Tapuz
animation forum, for their help with this article.
The author is a 2D/3D animator, animation teacher
and a generally odd man.
For questions, comments, compliments or hate-mail,
contact: doron_mei@bezeqint.net
