CS-184: Computer Graphics. Administrative

Transcription

1 CS-184: Computer Graphics Lecture #17: Introduction to Animation Prof. James O Brien University of California, Berkeley V Administrative You should have turned in project proposal Assignment #4 due April 14th 2

2 Today Tessellation Introduction to Animation 3 Adaptive Tessellation Recall early discussion of curve drawing Uniform Sampling Original Adaptive Sampling 4

3 Adaptive Tessellation Midpoint test subdivision 5 Adaptive Tessellation Midpoint test subdivision Possible problem 6

4 Adaptive Tessellation Midpoint test subdivision Possible problem Simple solution if curve basis has convex hull property If curve inside convex hull and the convex hull is nearly flat: curve is nearly flat and can be drawn as straight line Better: draw convex hull Works for Bézier because the ends are interpolated 7 Bézier Subdivision Form control polygon for half of curve by evaluating at u=0.5 Repeated subdivision makes smaller/flatter segments Also works for surfaces... 8

5 Adaptive Tessellation Given surface patch If close to flat: draw it Else subdivide 4 ways 9 Adaptive Tessellation Avoid cracking Passes flatness test Fails flatness test 10

6 Adaptive Tessellation Avoid cracking Cracks may be okay in some contexts... Crack in the surface 11 Adaptive Tessellation Avoid cracking 12

7 Adaptive Tessellation Avoid cracking Test interior and boundary of patch Split boundary based on boundary test Table of polygon patterns May wish to avoid slivers 13 Introduction to Animation Generate perception of motion with sequence of image shown in rapid succession Real-time generation (e.g. video game) Off-line generation (e.g. movie or television) 14

8 Introduction to Animation Key technical problem is how to generate and manipulate motion Human motion Inanimate objects Amorphous objects Control 15 Introduction to Animation Technical issues often dominated by aesthetic ones Violation of realism desirable in some contexts Animation is a communication tool Should support desired communication There should be something to communicate 16

9 Introduction to Animation Story Shading Layout Modeling Animation Lighting Camera From Parent, p Introduction to Animation Story Shading Layout Modeling Voice Animation Foley Lighting Camera Postprocessing For more detailed diagram, see Kerlow p.54 18

10 Introduction to Animation Key-frame animation Specification by hand Motion capture Recording motion Procedural / simulation Automatically generated Combinations e.g. mocap + simulation 19 Key-framing (manual) Requires a highly skilled user Poorly suited for interactive applications High quality / high expense Limited applicability From Learning Maya

11 Motion Capture (recorded) Markers/sensors placed on subject Time-consuming clean-up Reasonable quality / reasonable price Manipulation algorithms an active research area MotionAnalysis / Performance Capture Studio 3 Okan Arikan 21 Figure 1.1: Optical motion capture systems work by tracking retroreflective markers placed on the body. Since these markers appear as bright dots to the cameras filming them, they are easy to detect. Multiple cameras are then used to triangulate the position of these markers. as bright dots on the infrared cameras looking at them (see figure 1.1). Optical systems use a number of such infrared cameras to detect the two dimensional positions of every marker. Because the cameras can be calibrated to a known common coordinate system, a marker s three dimensional position can be triangulated from its two dimensional projections on the cameras that see it. The maximum magnetic field that can be generated by the source and the minimum amount of field that Motion Editing can be measured by the markers mean that the markers can not get too far away from the source. Magnetic systems are also prone to magnetic interference generated by metallic objects, which are almost impossible to avoid. Due to these reasons, optical motion capture systems are more common than magnetic ones. Typical optical motion capture systems are quite expensive. To obtain high accuracy, many cameras need to be employed. This is especially true if markers are to move in a large area. This is due to the perspective effect which means projections of father away points move less as they move in space. In order for a marker to be triangulated, its projection on several cameras need to be found. Since all a camera sees is a cloud of bright dots, the markers corresponding to each other in all camera views must be identified. Some motion capture systems accomplish this correspondence using active markers that emit an infrared pulse which uniquely identifies every marker. Since these Arikan, Forsyth, O Brien, SIGGRAPH

12 Motion Editing Arikan, Forsyth, O Brien, SIGGRAPH Model Construction Kirk, O Brien, Forsyth, CVPR

13 Simulation Generate motion of objects using numerical simulation methods g v x t+!t = x t +!t v t + 1 2!t2 a t 25 Simulation Perceptual accuracy required Stability, easy of use, speed, robustness all important Predictive accuracy less so Control desirable 26

14 Simulation Feldman, Arikan, O Brien, CVPR What to do with animations? Video tape Digital video Print it on yellow sticky notes 28

15 Video Tape Analog tape formats VHS/SVHS Beta SP 3/4 U-matic Digital tape formats Digi Beta DV Tape DVD (yes, I know DVDs are not tapes) 29 NTSC Standard Used by DVD, DV, and VHS 720x486 resolution (sort of) 1.33 aspect ratio Limited color range 30 frames per second (sort of 29.97) Interlaced video Overscan regions 30

16 Digital Video Wide range of file formats QuickTime MS Audio/Visual Interleaved (AVI) DV Stream Bunch o images Some formats accommodate different CODECs Quicktime: Cinepak, DV, Sorenson, DivX, etc. AVI: Cinepak, Indeo, DV, MPEG4, etc. Some formats imply a given CODEC MPEG DV Streams 31 Digital Video Nearly all CODECs are lossy Parameter setting important Different type of video work with different CODECs Compressors not all equally smart Compression artifacts are cumulative in a very bad way Playback issues Bandwidth and CPU limitations Hardware acceleration Missing CODECs (avoid MS CODECs and formats) 32

17 Path totape Not much of an issue any longer Cheap ( < $100 ) devices can give good amateur quality output Pro quality also cheap ( < $5000 ) Beware many cheap solutions over use compression Good analog tape decks still expensive 33 Editing Old way: Multiple expensive tape decks Slow Difficult Error prone New way: Non-linear editing software Premiere, Final Cut Pro, others... Beware compressed solutions May take a long time for final encoding 34

18 Motion Blur Fast moving things look blurry Human eye Finite exposure time in cameras Without blur: strobing and aliasing Blur over part of frame interval Measured in degrees (0..360) 30 tends to often look good 35 Motion Blur Easy to do in a sampling framework Interpolation is an issue 36

19 motion Blur Velocity based blur often works poorly 37