Computer Graphics I
Computer Graphics I CS 4620
Popular in Course
Popular in ComputerScienence
This 12 page Class Notes was uploaded by Lacey Collier on Saturday September 26, 2015. The Class Notes belongs to CS 4620 at Cornell University taught by Staff in Fall. Since its upload, it has received 18 views. For similar materials see /class/214338/cs-4620-cornell-university in ComputerScienence at Cornell University.
Reviews for Computer Graphics I
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 09/26/15
Animation CS 4620 Lecture 20 Cornell CS4620 Fall 2008 Lemme 20 2008 Steve Marsmner l Animation Industry production process leading up to animation What animation is How animation works very generally Artistic process of animation Further topics in how it works Cornell CS4620 Fall 2008 Lemme 20 2008 Steve Marsmner 2 What is animation Modeling specifying shape Animation specifying shape as a function of time Just modeling done once per frame Need smooth concerted movement Controlling shape the technical problem Using shape controls the artistic problem Cornell CS4620 Fall 2008 Lemme 20 2008 Steve Marsmner 3 Approaches to animation Straight ahead Drawanimate one frame at a time Can lead to spontaneity but is hard to get exactly what you want Posetopose Topdown process Plan shots using storyboards Plan key poses rst Finally ll in the inbetween frames Cornell CS4620 Fall 2008 Lemme 20 2008 Steve Marsmner 4 Posetopose animation planning 33 94 i r U J1ll Hi i r Wquot i 1 ill 5 it 2 E7 i w n W I m bum Reserch rst work out poses that are E 39yquotquot39t6 ti 39 i quotr y Next ll in animation in between Cornell CS4620 Fall 2008 I Lecture 20 2008 Steve Marschner 5 Keyframe animation I Keyframing is the technique used for posetopose animation Head animator draws key poses just enough to indicate what the motion is supposed to be Assistants do inbetweening and draws the rest of the frames In computer animation substitute user and animation software Interpolation is the principal operation Cornell CS4620 Fall 2008 I Lecture 20 2008 Steve Marschner 6 Keyframe animation Bryce Tutorial httpwwwcadtutornetddbryceaninVanimhtml Cornell CS4620 Fall 2008 I Lecture 20 2008 Steve Marschner 7 Walk cycle Christopher Luiz httpwwwanimationsnippescom Cornell CS4620 Fall 2008 I Lecture 20 2008 Steve Marschner 8 Controlling geometry conveniently Could animate by moving every control point at every keyframe This would be labor intensive It would also be hard to get smooth consistent motion Better way animate using smaller set of meaningful degrees of freedom DOFs Modeling DOFs are inappropriate for animation Eg move one square inch of left forearm Animation DOFs need to be higher level Eg bend the elbow Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 9 Character with DOFs GreenbergPellacini CIS 565 4 w maI dmcnprimi nf mponiblz mavenmms ir the gui 39El Cornell CS4620 Fall 2008 Lecture 20 gm I j 47 MnJU SE 59er 2008 Steve Marschner l0 Rigged character Surface is deformed by a set of bones Bones are in turn controlled by a smaller set of controls The controls are useful intuitive DOFs for an animator to use CIS 565 sta Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner l l The artistic process of animation What are animators trying to do Important to understand in thinking about what tools they need Basic principles are universal across media 2D handdrawn animation 2D computer animation 3D computer animation The following slides follow the examples from Michael Comet s very nice discussion on the page httpwwoomet cartoonscomtoonsdeocscharanim Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner l2 Animation principles timing 39 Speed of an action is crucial to the impression it makes examples with same keyframes different times Michael s Comet 60 fr looking around 30 fr no 5 fr just been hit Cornell csmo muoos ism 10 1008 SM Manchner i3 Animation principles ease lnout 39 Real objects do not start and stop suddenly animation parameters shouldn t either Michael 3 Carmen straight linear interpi a little goes a long way just a few frames acceleration or ease inout deceleration for snappy motions Cornell cwm muons Lmurelo e was 5m Marxchner i4 Animation principles moving in arcs 39 Real objects also don t move in straight lines generally curves are more graceful and realistic Michael s Comet Cornell csmo muoos ism 10 1008 SM Manchner i5 Animation principles anticipation 39 Most actions are preceded by some kind of windupquot Michael s Comet mini 3 Carmen Cornell cwm muons Lmurelo e was 5m Marxchner i s Anlmatlon principles exaggeration Animation principles squash amp stretch Animation is not about exactly modeling reality Objects do not remain perfectly rigid as they move Exaggeration is very often used for emphasis Adding stretch with motion and squash with impact models deformation of soft objects indicates motion by simulating exaggerated motion blur E u e v E u E 5 Cornell C9520 Fell1008 Lecture 10 1008 Steve Marxchner l7 Cornell C9510 PAHIOOE Lecture 10 1008 Steve Marxchner l8 Anlmatlon principles follow through Anim principles overlapping actlon We ve seen that objects don t start suddenly Usually many actions are happening at once 39 They also don t stop on a dime a a E E u u E i u Cornell C9520 Fell1008 Lecture 10 1008 Steve Marxchner l9 Cornell C9510 PAHIOOE Lecture 10 1008 Steve Marxchner 10 Animation principles staging I I Want to produce clear goodlooking 2D images need good camera angles set design and character positions Michael B Cornet Cornell CS4620 Fall 200E 39 Lecture 20 200E Steve Marschner 39 2 Principles at work weight Michael B Cornet Cornell CS4620 Fall 200E 39 Lecture 20 200E Steve Marschner 39 22 Extended example Luxo jr Cornell CS4620 Fall 200E 39 Lecture 20 200E Steve Marschner 39 23 Computer generated motion Interesting aside many principles of character animation follow indirectly from physics Anticipation followthrough and many other effects can be produced by simply minimizing physical energy Seminal paper Spacetime Constraints byWitkin and Kass in SIGGRAPH I988 Cornell CS4620 Fall 200E 39 Lecture 20 200E Steve Marschner 39 24 Controlling shape for animation Start with modeling DOFs control points Deformations control those DOFs at a higher level Example move rst joint of second nger on left hand Animation controls control those DOFs at a higher level Example openclose left hand Both cases can be handled by the same kinds of deformers Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 25 Rigid motion the simplest deformation Move a set of points by applying an af ne transformation How to animate the transformation over time Interpolate the matrix entries from keyframe to keyframe 39 This is ne for translations but bad for rotations Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 26 Parameterizing rotations Euler angles Rotate around x then y then 2 Problengimbal lock 39 If two axes coincide you lose one DOF Unit quaternions A 4D representation like 3D unit vectors for 2D sphere Good choice for interpolating rotations These are rst examples of motion control Matrix deformation Anglesquaternion animation controls Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 27 Hierarchies and articulated figures Robot assignment as an example Small number of animation controls control many transformations Constraint the joints hold together Robotics as source of math Methods Forward kinematics Inverse kinematics Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 28 Articulation in robotics a a rectangular or cartesian b cylindrical or posttype c spherical or polar a djointarm or articulated e SCARA selective compliance assembly robot arm Cornell CS4620 Fall 2008 Lecture 20 b 2008 Steve Marschner 29 Motion capture A method for creating complex motion quickly measure it from the real world thanks to Zoran Popovic39 for many visuals Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 30 Motion capture in movies Cornell CS4620 Fall 2008 Lecture 20 Final Fanatsy 2008 Steve Marschner 3 Motion capture in movies The Two Towers New Line Productions Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 32 Motion capture in games quot Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 33 Magnetic motion capture 39 Tethered 39 Nearby metal objects cause distortions 39 Low freq 60Hz Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 34 Mechanical motion capture 39 Measures joint angles directly 39 Works in any environment 39 Restricts motion Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 35 Optical motion capture 39 Passive markers on subject Cameras with IR illuminators 39 Markers observed by cameras Positions via triangulation Retroreflective markers Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 36 Optical motion capture 8 or more cameras Restricted volume High frequency 240Hz Occlusions are troublesome Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 37 From marker data to usable motion 39 Motion capture system gives inconvenient raw data Optical is least information case accurate position but 0 Which marker is which 0 Where are the markers are relative to the skeleton X039 Yb 21 X1 Y 2 X2 Y2 22 Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 38 Motion capture data processing Marker identi cationzwhich marker is which Start with standard rest pose Track forward through time but watch for markers dropping out due to occlusion Calibration match skeleton nd offsets to markers Use a short sequence that exercises all DOFs of the subject A nonlinear minimization problem Computing joint angles explain data using skeleton DOFs A inverse kinematics problem per frame Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 39 Motion capture in context 39 Mocap data is very realistic Timing matches performance exactly Dimensions are exact 39 But it is not enough for good character animation Too few DOFs Noise errors from nonrigid marker mounting Contains no exaggeration Only applies to humanshaped characters 39 Therefore mocap data is generally a starting point for skilled animators to create the nal product Cornell CS4620 Fall 2008 Lecture 20 2008 Steve Marschner 40 Basic surface deformation methods Mesh skinning deform a mesh based on an underlying skeleton Blend shapes make a mesh by combining several meshes 0 Both use simple linear algebra Easy to implement rst thing to try Fast to run used in games 0 The simplest tools in the offline animation toolbox Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 4 Mesh skinning 0 A simple way to deform a surface to follow a skeleton 1 ad 34 g f r 391 39 39lt39 l 5 J 1 g J l g v 7 74 7 gt 7 quott g ll 7 D 7 I 4 l C l 3 g W 391 m l r quot U i ELM g s W 4 9 Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 42 Mesh skinning math setup Surface has control points p Triangle vertices spline control points subdiv base vertices Each bone has a transformation matrix M Normally a rigid motion 0 Every point bone pair has a weight wij In practice only nonzero for small of nearby bones The weights are provided by the user Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 43 Mesh skinning math 0 Deformed position of a point is a weighted sum of the positions determined by each bone s transform alone weighted by that vertex s weight for that bone Pl szijpi 7 LLCWID CL GI DLUUKHI39FI LUUUJ Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 44 Mesh skinning Mesh skinning classic problems Simple and fast to compute Surface collapses on the inside of bends and in the Can even compute in the vertex stage of a graphics pipeline Presence Of strong tWiStS Used heavny in games Average of two rotations is not a rotation One piece of the toolbox for offline animation Add more bones to keep adlaocent bones from be39ng too different or change the blending rules Many other deformers also available 8 E E D 9 quot Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 45 Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 46 Blend shapes Blend shapes math 0 Another very simple surface control scheme Simple setup Based on interpolating among several key poses User provides key shapes that is a position for every Aka blend shapes or morph targets control point in every shape pi for point i shape j Per frame user provides a weight W for each key shape Must sum to 0 g1 Computation of deformed shape to E Pi ijpij E j E 0 Works well for relatively small motions a Often used for for facial animation Runs in real time popular for games Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 47 Cornell CS4620 Fall 2008 0 Lecture 20 2008 Steve Marschner 48
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'