CSE 165: 3D User Interaction. Lecture #5: Selection

Transcription

1 CSE 165: 3D User Interaction Lecture #5: Selection

2 2 Announcements Project 1 due this Friday at 2pm Grading in VR lab B :30pm Two groups: even hours start at 2pm odd hours at 3pm

3 3 Selection and Manipulation

4 4 Why are Selection and Manipulation Important? Major methods of interaction with physical environments virtual environments Affect the quality of entire 3D interface Design of 3D manipulation techniques is difficult

5 5 Selection vs. Manipulation Selection: specifying one or more objects from a set Manipulation: modifying object properties (position, orientation, scale, shape, color, texture, behavior, etc.)

6 6 Goals of Selection Indicate action on object Query object Make object active Travel to object location Set up manipulation

7 7 Selection Performance Variables affecting user performance Object distance from user Object size Density of objects in area Presence of occluding objects

8 Canonical Parameters Selection distance and direction to target target size density of objects around the target number of targets to be selected target occlusion Manipulation Positioning distance/direction to initial position distance/direction to target position translation distance required precision of positioning Rotation distance to target initial orientation final orientation amount of rotation 8

9 9 Input Device Parameters Number of control dimensions Control integration: how many DOF are controlled simultaneously Force vs. position control (relative vs. absolute location) Form factor: impact on accuracy Sensor attached to hand Sensor rolled with fingers

10 10 Technique Classification by Metaphor Manipulation techniques Egocentric metaphor Virtual pointer metaphor Ray-casting Two-handed pointing Flashlight Image plane Direct manipulation Classical virtual hand Go-Go Exocentric metaphor World-in-miniature Scaled-world grab Hybrid techniques Voodoo Dolls

11 11 Isomorphic vs. Non-Isomorphic View Isomorphic Geometrical on-to-one correspondence between hand motions in physical and virtual worlds Natural interactions Non-Isomorphic Magic virtual tools (laser beams, rubber arms, etc.)

12 12 Ray-Casting User points at objects with virtual ray Ray defines and visualizes pointing direction First intersected object is selected p( α) = h + α p h = 3D position of virtual hand p = ray attached to h 0 < α < determined by first object intersection

13 13 Two-Handed Pointing Ray casting with 2 hands More control Distance between hands controls length Allows pointing at things behind other things p( α) = hl + α ( hr hl ) 0 < α < is fixed hl = 3D position of left hand h = 3D position of right hand r

14 14 Flashlight Soft selection technique Does not need precision Conic selection volume Tip of cone in wand Cone direction determined by wand direction Fixed cone size If multiple objects in cone Object closer to center line of cone is selected If multiple objects are equally close to center line: select object closer to device

15 15 Virtual Hand Select and manipulate directly with hand Hand represented as 3D cursor Intersection between cursor and object indicates selection p v = α pr, R v = Rr pr, Rr = position and orientation of real hand p v, R v = position and orientation of hand in VE α = fixed scaling factor

16 16 Go-Go By Poupyrev, 1996 Arm-extension technique Touch objects to select, like simple virtual hand Non-linear mapping between physical and virtual hand position Requires torso position v Local and distant regions r v where rr = length of R r = length of R D, α are constants r r if if r r D D = rv F= ( F r )( r= ) = 2 2 r + r α+ ( α( r r D ) Dotherwise ) where r = length of Rr rv = length of R v D, α are constants v r

17 17 World-in-Miniature (WIM) By Stoakley, 1995 Dollhouse world held in user s hand Miniature objects can be manipulated directly Moving miniature objects affects full-scale objects Can also be used for navigation

18 18 Image Plane Techniques Require only 2 DOF Selection based on 2D projections Use virtual image plane in front of user Dependent on head/eye position Framing Lifting Palms Head- Crusher Sticky Finger

19 19 Scaled-World Grab By Mine et al., 1997 Often used with occlusion At selection, scale world down so that virtual hand touches selected object User initially does not notice a change in the image, until head or hand is moved

20 20 Voodoo Dolls Pierce et al Two-handed technique Builds upon image plane and WIM techniques Developed for pinch gloves Requires finger pose tracking Creates copies of objects (dolls) for manipulation Non-dominant hand: stationary frame of reference Dominant hand: defines position and orientation

21 21 Selection by Dwell Time User points at object with any technique Virtual pointer Eye gaze Action is triggered after dwell time threshold is exceeded Works without physical buttons Frequently used in controller-less VR: Google Cardboard, Samsung Gear VR

22 22 3D UI With the Leap Selection Hover w/timeout Trigger with non-dominant hand gesture Two finger near-pinch Manipulation Hand orientation 3-finger orientation 2-finger orientation (2 DOF)

23 23 Menus with the Leap Hover over buttons Leap API-Supported gestures: Rotate Swipe

24 24 General Tips for the Leap Finger pinches hard to detect More than 3 fingers hard to distinguish Fingers hard to distinguish when hand not close to horizontal Hand detection (left/right): need to bring hands into FOV from back edge Options for camera motion: rotate around circle, set with non-dominant hand, map orientation of non-dominant hand

25 25 Related: Forced Perspective Museum of Simulation Technology

26 Quaternions

27 Rotation Calculations Intuitive approach: Euler Angles: Simplest way to calculate rotations Defines rotation by 3 sequential rotations about coordinate axes Example Z-Y-X:

28 Problems With Euler Angles Problems with Euler angles: No standard for order of rotations Gimbal Lock, occurs in certain object orientations Video (0:20-1:12) Better: rotation about arbitrary axis (no Gimbal lock) Can be done with 4x4 matrix But: smoothly interpolating between two orientations is difficult Quaternions

29 Quaternion Definition Given angle and axis of rotation: a: rotation angle {nx,ny,nz}: normalized rotation axis Calculation of quaternion coefficients w, x, y, z: w = cos(a/2) x = sin(a/2) * nx y = sin(a/2) * ny z = sin(a/2) * nz

30 Useful Quaternions w x y z Description Identity quaternion, no rotation turn around X axis turn around Y axis turn around Z axis sqrt(0.5) sqrt(0.5) rotation around X axis sqrt(0.5) 0 sqrt(0.5) 0 90 rotation around Y axis sqrt(0.5) 0 0 sqrt(0.5) 90 rotation around Z axis sqrt(0.5) -sqrt(0.5) rotation around X axis sqrt(0.5) 0 -sqrt(0.5) 0-90 rotation around Y axis sqrt(0.5) 0 0 -sqrt(0.5) -90 rotation around Z axis

31 Quaternions in Bullet Physics Quaternions can be specified by rotation angle and axis: btquaternion(const btvector3 &_axis, const btscalar &_angle) Or by an x, y, z, w tuple: btquaternion(const btscalar &_x, const btscalar &_y, const btscalar &_z, const btscalar &_w) Bullet defines mathematical and other operators: +, -, *, /, inverse, getangle, getaxis, slerp, etc.

32 Quaternions: Further Reading Gamasutra: Rotating Objects Using Quaternions 86/rotating_objects_using_quaternions.php Quaternions in Ogre3D: +Rotation+Primer Quaternions in OSG: wiki/support/maths/quaternionmaths