cnm 190 advanced digital animation lec 12 : walk cycles & autonomous motion
DESCRIPTION
CNM 190 Advanced Digital Animation Lec 12 : Walk Cycles & Autonomous Motion. John Cleese “Silly Walk” animated using SPAM software. Dan Garcia , EECS (co-instructor) Greg Niemeyer, Art (co-instructor) Jeremy Huddleston, EECS (TA). Walk Cycles Definition, History References - PowerPoint PPT PresentationTRANSCRIPT
CNM 190CNM 190Advanced Digital AnimationAdvanced Digital Animation
Lec 12 : Walk Cycles & Autonomous MotionLec 12 : Walk Cycles & Autonomous MotionDan GarciaDan Garcia, EECS (co-instructor), EECS (co-instructor)
Greg Niemeyer, Art (co-instructor)Greg Niemeyer, Art (co-instructor)
Jeremy Huddleston, EECS (TA)Jeremy Huddleston, EECS (TA)
John Cleese “Silly Walk” animated using SPAM softwareJohn Cleese “Silly Walk” animated using SPAM software
2/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
OverviewOverview
Walk CyclesWalk Cycles Definition, Definition,
HistoryHistory ReferencesReferences What makes a What makes a
great walk cycle?great walk cycle? ExamplesExamples SPAM demoSPAM demo UCB ResearchUCB Research
Autonomous MotionAutonomous Motion DefinitionDefinition Flow ChartFlow Chart Finite State Finite State
MachinesMachines Brian Mirtich DemoBrian Mirtich Demo
3/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Walk Cycles : Definition & Walk Cycles : Definition & HistoryHistory
DefinitionDefinition A sequence of frames capturing a walk movementA sequence of frames capturing a walk movement
HistoryHistory Since the early days of hand-drawn animation, Since the early days of hand-drawn animation,
these have been used to get characters from A to B these have been used to get characters from A to B and, most importantly, to and, most importantly, to capture & convey capture & convey emotionemotion
www.idleworm.com/how/anm/02w/walk1.shtml
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
4/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
References (1)References (1) The Illusion of LifeThe Illusion of Life : :
Disney AnimationDisney Animation by Frank Thomas & Ollie by Frank Thomas & Ollie
JohnstonJohnston Two of Disney’s Two of Disney’s
“Nine Old Men”“Nine Old Men” “… “… gives many glimpses gives many glimpses
into the workings of the into the workings of the animation masters at animation masters at Disney's during the Disney's during the Golden Age of Golden Age of animation”animation”
On every professional On every professional animator’s bookshelfanimator’s bookshelf
576 Pages576 Pages Disney Editions, 1995Disney Editions, 1995 Out-of-print! ~$40Out-of-print! ~$40
en.wikipedia.org/wiki/The_Illusion_Of_Life
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
5/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
References (2)References (2) The Animator’s Survival The Animator’s Survival
KitKit by Richard Williamsby Richard Williams 352 pages352 pages Faber & Faber, 2002Faber & Faber, 2002 ~$20 new, ~$15 used~$20 new, ~$15 used
Cartoon AnimationCartoon Animation by Preston Blairby Preston Blair 144 pages144 pages Walter Foster, 1994Walter Foster, 1994 ~$15 new, ~$10 used~$15 new, ~$10 used
www.amazon.com
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
6/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
References (3)References (3)
Principles of Principles of Traditional Animation Traditional Animation Applied to 3D Applied to 3D Computer AnimationComputer Animation, , Computer Graphics, Computer Graphics, pp. 35-44, 21:4, July pp. 35-44, 21:4, July 1987 (SIGGRAPH 87).1987 (SIGGRAPH 87). by John Lasseterby John Lasseter He shared the secrets of He shared the secrets of
the kingdom with 3D the kingdom with 3D animatorsanimators
There’s an abridged There’s an abridged version online (URL above)version online (URL above)
www.siggraph.org/education/materials/HyperGraph/animation/character_animation/principles/prin_trad_anim.htm
7/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
What makes a great walk What makes a great walk cycle? (1)cycle? (1)
Four distinct posesFour distinct poses Contact – feet at furthest extension of walk (start Contact – feet at furthest extension of walk (start
here)here) Recoil – where the character impacts the groundRecoil – where the character impacts the ground Passing – an ‘in-between’ of Recoil and High PointPassing – an ‘in-between’ of Recoil and High Point High Point – character’s body stretched to maxHigh Point – character’s body stretched to max
www.idleworm.com/how/anm/02w/walk1.shtml
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
8/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
What makes a great walk What makes a great walk cycle? (2)cycle? (2)
Depending on the Depending on the emotion to convey, emotion to convey, these four basic these four basic poses may changeposes may change No 2-pt contact pose No 2-pt contact pose
for run cyclesfor run cycles E.g., skipping E.g., skipping
requires more than 4 requires more than 4 posesposes
Note several thingsNote several things Path of headPath of head Lean of body, headLean of body, head Arms & hands Arms & hands
matter!matter!
minyos.its.rmit.edu.au/~rpyjp/a_notes/walk_reference_01.html
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
Preston Blair’s Walk Cycles
9/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
What makes a great walk What makes a great walk cycle? (3)cycle? (3)
Four-legged figures Four-legged figures are similar, but are similar, but more complicatedmore complicated Are legs in phase?Are legs in phase? Horses have vastly Horses have vastly
different cycles different cycles depending on depending on speedspeed
This is relevant for This is relevant for both teamsboth teams Elephants, GiraffesElephants, Giraffes Robot DogRobot Dog
minyos.its.rmit.edu.au/~rpyjp/a_notes/walk_reference_02.html
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
Preston Blair’s Walk Cycles
10/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
What makes a great walk What makes a great walk cycle? (4)cycle? (4)
Red line shows off-Red line shows off-balance lean for a balance lean for a typical walk cycletypical walk cycle
Happy charactersHappy characters Bouncy, more time Bouncy, more time
up in the air than up in the air than downdown
Sad charactersSad characters Plod along, drag feetPlod along, drag feet
Step distance also Step distance also signals emotion, signals emotion, ageage Older, shy Older, shy shuffle shuffle
minyos.its.rmit.edu.au/~rpyjp/a_notes/04_walkcycle_project.html
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
11/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
What makes a great walk What makes a great walk cycle? (5)cycle? (5)
The walk cycle The walk cycle should have good, should have good, consistent consistent transverse transverse motions (L-R)motions (L-R) E.g., Note arms; E.g., Note arms;
they don’t just they don’t just rotate in direction rotate in direction of walkof walk
minyos.its.rmit.edu.au/aim/a_notes/walk_reference_03.html
12/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Examples of WalksExamples of Walks
Fortunately, it’s easy Fortunately, it’s easy to act out a walk & to act out a walk & film it for analysisfilm it for analysis
Humans good atHumans good at reading impossible reading impossible
and bad walksand bad walks reading emotion from reading emotion from
walkswalks Make sure to convey Make sure to convey
weight, balanceweight, balance Practice, practicePractice, practice Let’s see some Let’s see some
examplesexamples
minyos.its.rmit.edu.au/~rpyjp/a_notes/04_walkcycle_project.htmlwww.anticz.com/Walks.htm
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Monty Python’s Ministry of Silly Walks
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
13/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
SPAM – FK Walk EditorSPAM – FK Walk Editor I wrote a forward-I wrote a forward-
kinematic walk editor back kinematic walk editor back in 1994in 1994
It uses a cubic BIt uses a cubic Bézier ézier spline to interpolate:spline to interpolate: Q(t) = Q(t) = VV00 (1-t) (1-t)33 + + VV11 3t (1-t) 3t (1-t)22
+ + VV22 3t 3t22 (1-t) + (1-t) + VV33 t t33
I start at forward foot as I start at forward foot as root and work up & downroot and work up & down
digitalassets.lib.berkeley.edu/techreports/ucb/text/CSD-94-860.pdf
14/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Walk Generation –Walk Generation – Use Use MoCap!MoCap!
We saw a MoCap We saw a MoCap demo last week demo last week
With a talented With a talented actor, you can actor, you can create any human create any human motionmotion E.g., LoTR “Gollum”E.g., LoTR “Gollum” You can also use the You can also use the
data as a starting data as a starting point for editingpoint for editing
Might be lots of raw Might be lots of raw data cleaning updata cleaning up
Purists don’t use itPurists don’t use it
en.wikipedia.org/wiki/Motion_capture
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
15/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Walk Cycle Summary…Walk Cycle Summary…PersonalityPersonality
““The objective of the The objective of the [Animation] principles [Animation] principles previously discussed [in his previously discussed [in his SIGGRAPH paper] are to give SIGGRAPH paper] are to give the characters in an animation the characters in an animation a personality that appeals to a personality that appeals to the viewersthe viewers.”.”
““The different principles The different principles should be applied in a fashion should be applied in a fashion to produce a consistent to produce a consistent personality.” personality.”
““This means that This means that the animator the animator must have a good idea of the must have a good idea of the desired personality desired personality beforebefore beginning the animationbeginning the animation.”.”
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
www.siggraph.org/education/materials/HyperGraph/animation/character_animation/principles/personality.htm
16/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
UCB Walk-Related ResearchUCB Walk-Related Research
There has been a lot of There has been a lot of research done here research done here relating to walk cycles relating to walk cycles & MoCap databases& MoCap databases Electronic Arts gave us Electronic Arts gave us
some of their MoCap some of their MoCap datadata
Okan Arikan (now a Prof Okan Arikan (now a Prof at U Texas) led charge, at U Texas) led charge, with Profs O’Brien, with Profs O’Brien, ForsythForsyth
Knowing When to Put Knowing When to Put Your Food DownYour Food Down
Pushing people aroundPushing people around Motion Synthesis from Motion Synthesis from
AnnotationsAnnotations Motion Synthesis & Motion Synthesis &
RetargetingRetargeting
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
3 UCB MooVs
www.cs.berkeley.edu/b-cam/www.cs.utexas.edu/~okan/
17/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Autonomous MotionAutonomous Motion More than simply More than simply
spacetime spacetime constraintsconstraints
This field is VAST, for This field is VAST, for some the Holy Grail!some the Holy Grail! Artificial intelligenceArtificial intelligence Artificial lifeArtificial life RoboticsRobotics Video gamesVideo games Crowd, herd & Crowd, herd &
flocking simulationsflocking simulations Real-time >> BatchReal-time >> Batch
In terms of $, interestIn terms of $, interest
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
18/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Autonomous Animated Autonomous Animated CharactersCharacters
Here’s a general framework…Here’s a general framework…
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
www.kuffner.org/james/anim/framework.php
WhyWhyWhere Where
HowHow
19/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Lots of questions ; this Lots of questions ; this is hard!is hard!
Why / Where, Why / Where, often in real time…often in real time… What behavior is What behavior is
generated / issued?generated / issued? Where do we go?Where do we go? What is the emotion?What is the emotion? What’s our What’s our
motivation?motivation? What is our mood as What is our mood as
the result of the new the result of the new motion / location?motion / location?
When do we have When do we have births / deaths?births / deaths?
How do we, How do we, often in real time…often in real time… select the animations select the animations
from DB? Or…from DB? Or… auto-generate the auto-generate the
animation?animation? blend the animations blend the animations
we have with what is we have with what is requested?requested?
retarget animations retarget animations to capture request?to capture request?
scale pre-set motions scale pre-set motions for new characters?for new characters?
20/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Intelligence : Finite State Intelligence : Finite State Machines (1)Machines (1)
VERY powerful ideaVERY powerful idea Model of behavior Model of behavior
composed of composed of statesstates, , transitionstransitions & & actionsactions
State (1 must be a State (1 must be a start)start) Past informationPast information
TransitionTransition Condition that triggers Condition that triggers
state changestate change ActionAction
Activity performed at Activity performed at a given momenta given moment
en.wikipedia.org/wiki/Finite_state_machine
FSM for gum machine
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
21/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Intelligence : Finite State Intelligence : Finite State Machines (2)Machines (2)
4 Kinds of Actions. For 4 Kinds of Actions. For each, “Execute action each, “Execute action _”_” EntryEntry : …when : …when
ENTERING stateENTERING state ExitExit : …when EXITING : …when EXITING
statestate InputInput : …dependent on : …dependent on
present state and input present state and input conditionsconditions
TransitionTransition :…when :…when performing a transitionperforming a transition
en.wikipedia.org/wiki/Finite_state_machine
QuickTime™ and aTIFF (Uncompressed) decompressor
are needed to see this picture.
22/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Brian Mirtich, 1996 UCB Brian Mirtich, 1996 UCB Ph.D.Ph.D.
Thesis: “Thesis: “ImpulseImpulse - -based Dynamic based Dynamic Simulation of Rigid Simulation of Rigid Body Systems”Body Systems” He focused on He focused on
dynamic constraintsdynamic constraints Very cool work!Very cool work!
““Creature War” Creature War” demodemo His purpose: show His purpose: show
off his simulatoroff his simulator Great example of Great example of
FSM-based AI!FSM-based AI!
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Mirtich’s Creature War
www.kuffner.org/james/software/dynamics/mirtich/
23/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
Boids – Distributed Boids – Distributed Behavior Behavior
Big idea : Give many Big idea : Give many characters their own FSMcharacters their own FSM
Craig Reynolds Craig Reynolds “founded” first Boids in “founded” first Boids in 1986; A-Life1986; A-Life SeparationSeparation – steer to avoid – steer to avoid
crowding local flockmatescrowding local flockmates AlignmentAlignment – steer towards – steer towards
the average heading of the average heading of local flockmateslocal flockmates
CohesionCohesion – steer to move – steer to move toward the average toward the average position of local flockmates position of local flockmates
Emergent behavoir!Emergent behavoir! CrowdsCrowds just another just another
example of same idea…example of same idea…
www.red3d.com/cwr/boids
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
Boids Applet, Movie, It’s a Big Ad
24/24Walk Cycles & Autonomous Motionhttp://inst.eecs.berkeley.edu/~selfpace/CNM190
ConclusionConclusion Autonomous Motion Autonomous Motion
(incl Walking) is the (incl Walking) is the goal of many (esp goal of many (esp games!!)games!!)
Imagine virtual actors Imagine virtual actors interacting in real timeinteracting in real time What do they want to do What do they want to do
and when?and when? How do they generate How do they generate
smooth motion to do it?smooth motion to do it? Can actors die? Be Can actors die? Be
born?born? FSM key to it all!FSM key to it all! Strong vs Weak AIStrong vs Weak AI
Strong: An amazing FSM Strong: An amazing FSM can have “sentience”!can have “sentience”!
Commander Data from Star Trek
en.wikipedia.org/wiki/Strong_AI