"Finding Nemo": why are the mouth motions still off at this point?

"Finding Nemo": why are the mouth motions still off at this point?

Post by zvoi » Tue, 08 Jul 2003 07:29:52



(I am not a regular reader/poster of this group, but I thought you
guys would know the answer to my question.)

I saw the movie "Finding Nemo" recently. Although it is clearly a
movie that is meant for kids, it is technically brilliant.
I know that putting together one of these computer generated animated
films is a very grueling process; but how come at this point in time,
(technologically speaking), the actors' vocal tracks are slightly out
of sync from the animated characters' mouth movements. When you see a
character talk, some of the lip/vowel positions appear almost
short-cutted, (as in traditional animation). The reason that this
puzzles me is that they apparently have the data of the original
actors' mouth movements and have a program that can translate these
original movements to the animated characters, (right?) Why not put
the whole motion in?
How come they are skimping on these mouth movements/synchronizations?
Is it that aesthetically they just like the way it looks?
Is it a budget issue?

Thank you in advance for your input.

 
 
 

"Finding Nemo": why are the mouth motions still off at this point?

Post by Tonci Neme » Sun, 13 Jul 2003 01:54:32



Quote:> (I am not a regular reader/poster of this group, but I thought you
> guys would know the answer to my question.)

> I saw the movie "Finding Nemo" recently. Although it is clearly a
> movie that is meant for kids, it is technically brilliant.
> I know that putting together one of these computer generated animated
> films is a very grueling process; but how come at this point in time,
> (technologically speaking), the actors' vocal tracks are slightly out
> of sync from the animated characters' mouth movements. When you see a
> character talk, some of the lip/vowel positions appear almost
> short-cutted, (as in traditional animation). The reason that this
> puzzles me is that they apparently have the data of the original
> actors' mouth movements and have a program that can translate these
> original movements to the animated characters, (right?) Why not put
> the whole motion in?

In fact, that's not exactly how things run at Pixar. Other studios may be
contented by using software that directly translates mouth movements into
keyed digital form (by some form of pattern recognition i.e. motion
capture), however (afaik) people at Pixar essentially believe that it is a
shortcut (although they do have a degree of analog control involved to
introduce randomness that is present with natural motion) which defeats the
certain aspects of toon-making.

I have just seen Finding Nemo today. I am not a laureate when it comes to
cartoons in general, but I liked it very much. As far as animation goes,
Pixar did it once again! Two thumbs up!

Quote:> How come they are skimping on these mouth movements/synchronizations?

I have only seen it once and it didn't bother me. I'll pay more attention
next time around.

Quote:> Is it that aesthetically they just like the way it looks?

Well, since the movie is viewed and re-checked more than a hundred times in
its entirety before it even leaves Pixar Animation Studios, I'd say they
thought the animation was satisfactory.

Quote:> Is it a budget issue?

I don't think it's a budged issue.
Seriously, after working for years on storyboarding, design, modeling,
shading and animation I don't find it likely to run out of juice in the
mouth-keying ;) Besides, you can always check Monsters Inc. and compare. The
method they used is essentially the same.

Regards,

Tonci

 
 
 

"Finding Nemo": why are the mouth motions still off at this point?

Post by rick » Tue, 15 Jul 2003 02:24:49


It didn't seem out of line to me.

They usually don't motion-capture the voice-actor's face.
A common technique is to have a dictionary of face-phonome motions,
first-cut sync them to the speech-phonomes and speech-rate, then
post-edit.  Several vendors sell these dictories for various
animation packages.  Some of these may have been derived from motion capture.

About two weeks ago, the LA Times ran an article on the voice
capture of name-actors.  They aren't paid-much upfront, about $10,000 / day,
but share in revenues.  Brad Pitt said it was relatively easy work
compared to war-action-adventure he is filming now.
The animation companies have promo-reels of acttor recordings,
and they there rarely is motion capture.  And one mentions she does
touchups from overseas studios, should a director need adjustments.

And film movies can have flaws too.  If the sound editor hasnt converted recording
microphone projection to theater distances properly, that can look funny.
A large fraction of outdoor scenes are post-dubbed, due to sound problems.
A documentary on Star Wars Phantom Menance shows this.

 
 
 

1. different *point*,*vector*,*normal* "why"

Hi all

I was wondering why there is "different" typecast concerning some
*build-in* functions in RiSpec.

ex. In "RiSpec3.2" some or all function that have variable *N* is casted
to *vector*

but is this always true or accurate in name of all shader a have code
and have look at
use *normal* for N and not *vector*

There where also people that declare *point* cast to a normal. but not
me I always use *normal*

ex

normal Nf = faceforward(normalize(N),I);

also *Arman* do declare a *vector* to this functions example from this
book

color diffuse(vector N)
color specular(vector N, V; float roughness)
also another type of function do this.

So if I take a look in "Larry Gritz" header *shading models* he use
always *normal* for N

So it's kind of *confused* to think that Arman and RiSpec3.2 are calling
geometric typecast *point-like* data
("point* , *vector* and *normal*) to share a commen name *vector*

Is this the case ???

regards,
Fredrik Gustafsson

2. limited budget graphics card

3. ****"""""BUG IN PROGRAM"""*****

4. TIFF files format info

5. "Morph_Gizmo"ed mouth wanted

6. read jpg files

7. """"LOOKING FOR METRAEYES MODELLER AND TURBO.PXP FOR AN EXCHANGE"""""

8. Arg! Trying to create *2D* terrain

9. "Off screen" rendering in IrisGL

10. "Scientific American Frountiers" rips off SIGGRAPH

11. "HELP ME" - Rip-off warning

12. off topic: anyone heard of "upFront"

13. Bug: "Dumb Terminal" Plots Cut Off