Perception of Objects in TechnicalIllustrations - A Challenge in Technical Communication Debopriyo Roy Associate Professor University of Aizu December 12, 2010 Presentation at IEEE PCS-J, Nagano City, Japan
Depth Perception Depth perception is the visual ability to perceive the world in three dimensions (3D)and the distance of an object. Monocular cues provide depth information when viewing a scene with one eye. We perceive depth with 2 main sources of information: A. Binocular disparity B. Monocular cues
Depth Perceptions & Technical Illustrations How should technical illustrators demonstrate physical orientation to understand procedures What should be the characteristics of the display plane for visualizing procedures?
Speciﬁc Issues for Technical Writers How do we design body/object positions in user manuals most effectively for second language speakers such that a minimalist text approach is viable. Would readers in EFL context understand the subtle difference within body and object positions /orientations, based on difference in tasks when explained with minimal English text?
Affordance - 1 (Image Plane)Anatomical Image Plane -Axial - Image slices that are perpendicular to the long axis of the bodyCoronal-Image slices that bisects the front and back sides of the bodySagittal- Image slices that bisects the left and right side of the body
Affordance -2 (Optimal Perspective) Canonical View - Front, side and top views of an object.
Affordance - 3 (Task-based)Orientation of Illustrations Characteristics of Display Planes Distance between legsWeight of the ball Exact location of the ballPressure on the knees Angle of shoulder bendPressure on the shoulder Angle between hips and legsGrip to lift the ball Head PositionLeg fold Extent of Knee foldsAnkle pressure Direction of ball liftTransfer of weight Displacement during liftBend of vertebral column Overall Camera angle
Visual Systems People use at least two visual systems when performing tasks in the three dimensional world. One system involves primary processing of visual information by the back of the brain with subsequent processing by its sides. Side areas seem to specialize in object identiﬁcation (e.g., cube vs. cylinder). The other system performs secondary visual processing at the top- rear of the brain and helps with body orientation in space.
Object Manipulation vs. Physical Orientation Research ﬁndings suggest that people use different parts of their brains when doing tasks requiring object Ball Grip manipulation. However, the extent of such use diminishes when doing tasks requiring physical orientation of their bodies in three-dimensional space. Ball Release
Two Dimensional Views of ThreeDimensional Plane For on-screen or print materials, people only have monocular cues to help them interpret what they see. This reduction of depth cues makes interpretation more difﬁcult and an illustration’s point of view more critical.
Object vs. Body-Centered View Research has shown that the effects of mental rotation depend on point of view. Spectator’s point of view Performer’s point of view
Perception of Distances People are better at judging distances directly across the display plane. In Figure 1, the height of the box relative to the man’s waist is clearly visible. This is because the distance is across the display plane. However, the distance between the box and the man’s chest is not visible because the movement is into the display plane.
Perception of More Complex Physical Postures In this ﬁgure, the distance between any two object is recognized mostly as a function of other body postures and parts. To understand this object, different angular veriﬁcation along each of X, Y and Z axis is required.
Illustrations and Instructions Heiser and Tversky found (for a furniture assembly task) that illustrations showing objects partially rotated is easier to understand than objects showed head on. Szlichcinski found illustrations showing slightly rotated objects to have more task-based affordance. He also found that hand positions in illustrations are duplicated in actual tasks exactly as shown.
Image Perceptions Force up / down Push / Pull Angular Push / Pull Pressure up/down Showing direction Showing shape Grip Throw Turn Squeeze Twist Entire Body Posture
Variations of Viewpoints - Body PosturesHeight /Angle of Camera Const. -AlongWaistlineNo manipulation Z & Y axis manipulation Z, Y, X axis manipulationX & Z axis manipulation Z, X & Y axis manipulation X, Z and Y axis manipulation
Variations of Viewpoints - Camera Rotationalong waist line (X & Z axis manipulation)2/3rd Front 1/3rd Side Side 1/3rd BackBack 1/3rd Back 2/3rd Front
Variations of Viewpoints - Z axismanipulation Movement along Z axis Movement along Y axis
Object / Posture / Visual Angle ComplexityBend angle Bend angle, ankle movement Hand pressure, grip Exact box position, leg pressure, grip, leg rotation, twist, finger folds to hold the box
Hypotheses Major Research Question - Previous studies do not resolve conﬂict between orientation of illustrations versus characteristics of display plane. Illustrations that show a performer’s point of view will be easier to use. Illustrations that place important distances across the display plane will be easier to use.
Methods - 1 Test subjects were asked to evaluate body images via a matching task. 20 subjects were tested and each subject rated 20 sets of images. Poser Figure Artist software used to design body positions and angles. It sustains accurate three dimensional relationship among body parts. Variations of viewpoints and body positions were tested.
Methods - 2 (Operation Matrix)Task # 1: Man with the box
Summary of Findings Subjects were most conﬁdent for images shown directly from front or back. Somewhat less conﬁdent of the images rotated one-third from the front or back. (contrary to literature) Least conﬁdent of images shown from side. Two views did poorly - images rotated 1/3rd from front and back. (contrary to literature) Average number of correct choices were around 45%. Average correct matches for direct head-on view, 1/3rd back and back views cluster around 40-50%. 1/3rd front and side views vary tremendously.
Ongoing Analysis on Performance: Accuracy Scores Correlation between accuracy rates for different heights of the box along the Y axis. Correlation between accuracy rates for the displacement types. (Extensive further analysis required for Z-axis manipulations into the display plane). 5 different coordinates were chosen for trial: (2, 6, 0); (3 ,6, 2); (4, 6, 4); (4, 6, 6); (3, 6, 5) Correlation between camera angle rotations along the waist / chest.
Future Analysis Variation in displacement types on the basis of action:1. Carrying 2. Lifting 3. Pulling 4. Pushing Differentiate between postures on the basis of props.1. Chair 2. Ball 3. Box 4. Handle 5. Sphere/Triangle. - Purpose is to determine whether object shape is a function of performance and accuracy with physical tasks. OS = f[A f1(D, CA)] Variation in displacement types based on movement: E.g., left, center and right oriented displacement variations. E.g., variation along the Y axis. E.g., variation into display plane (z axis).
Free Viewpoint - Looking into Future Free viewpoint television (FTV) is a system for viewing natural video, allowing the user to interactively control the viewpoint and generate new views of a dynamic scene from any 3D position. With FTV, the focus of attention can be controlled by the viewers rather than a director, meaning that each viewer may be observing a unique viewpoint. This technology might revolutionize the way we look at interactive procedural user manuals. Readers will be able to switch views as and when they need to, based on task. However, this might not necessarily improve procedural accuracy, because of uncontrolled access to viewpoints, enabling spatial misconception.
Japan’s Promise for WC Football 2022 Japanese organizers say each game will be ﬁlmed by 200 high deﬁnition cameras, which will use "freeviewpoint" technology to allow fans to see the action unfold from a players eye view -- the kind of images until now only seen in video games. (CNN.com - 1st Dec, 2010)