2D vs 3D, Implications on Spatial Memory

Monica Tavanti, Mats Lind
[ACM Link]
Depth is a holistic perception
  • depth cues add linearly to the perception of depth
  • some cues are weaker or stronger than others

Binocular (Stereo) Display

  • different images at each eye
  • over a short distance
  • hard to do with computers (HMD, glasses, etc)
  • very powerful cue

Motion

  • motion paralax - as you move, objects closer move faster
  • use of head tracking to add motion paralax can have effects as good as stereo display
  • strong cue

Shadows

  • In our world, sun is above, and shadows are below
  • attached shadow: the shadow is on the object itself (shading)
  • cast shadow: an object casts a shadow on another object or background
  • can affect perceived size, elevation, and releative depth
  • sharp shadows are better: soft shadows can be detrimental to determining object shape in the absence of other cues.
  • Kinetic Depth Effect: Moving shadows allow for perception of the casting object's shape

Occlusion

  • Nearer objects cover up objects father away
  • partial occlusion (Zhai et al 96) w/ semitransparent surfaces helps see shapes through other's as further away.

Contrast

  • Objects with lower contrast (IE pastels) appear further away
  • as if looking through fog or haze

Study aims at judging the effects of shadows, light sources, stereo, and background on resizing and positioning tasks.

  • Shadows had no effect on accuracy.
  • Introducing a second light source reduced the accuracy
  • Stereo improved task performance measurably.
  • shadows decreased the performance when added to stereo presentation

(quotes Wickens a lot!)

The relative contributions of stereo, lighting, and background scenes in promoting 3D depth visualization

Geoffrey S. Hubona et al
[ACM Link]
This paper investigates whether spatial memory differs in 2d and 3d interfaces. A study on the Data Mountain application used to organize bookmarks as thumbnails on a rising plane. They tested users with similar 2D and 3D interfaces for the same tasks. Performance times were very similar, though 2D showed faster mean times. Subjects showed a slight preference towards the 3D interface.

Depth Cues

[website]

A great site for a review of depth cues in relation to art

Here are some cues not mentioned in the previous paper:

  • Relative Size: smaller objects are further away
  • Relative height: closer to the horizon, the further the object
  • Linear Perspective: parallel lines converge in the distance
  • Texture Gradient: textures get smaller at further distances
  • Aerial Perspective: objects at a distance have softer edges due to atmospheric effects
  • Hue: red appears above blue

Vision and the graphical simulation of spatial structure

W. A. van de Grind
[ACM Link]

3D perception


3D or not 3D? Evaluating the Effect of the Third Dimension in a Document Management

Andy Cockburn and Bruce Mckenzie
[google.scholar]
This paper investigates whether spatial memory differs in 2d and 3d interfaces. A study on the Data Mountain application used to organize bookmarks as thumbnails on a rising plane. They tested users with similar 2D and 3D interfaces for the same tasks. Performance times were very similar, though 2D showed faster mean times. Subjects showed a slight preference towards the 3D interface.


Revisiting 2D vs 3D implications on spatial memory

Andy Cockburn
[ACM Link]
This paper investigates whether spatial memory differs in 2d and 3d interfaces.

"Ehret (2002) provides interesting insights into how users learn the location of items in a user interface. His experiment shows that users learn locations more effectively when targets poorly represent their function. In other words, the higher the ’Äòevaluation cost’Äô (the degree of effort the user must put into finding the function of an item), the better the location is learned. Ehret presents a theory that predicts how well users learn the location of interface items as evaluation cost varies. "

"Jones and Dumais (1986) provide some cautions on overreliance on spatial organization. Their evaluation indicates that semantic labels provide stronger retrieval cues than spatial organisation alone, but indicate that combinations of semantic and spatial organization enhance performance. "

Mentions a paper by Wickens (97) which might be of use.

The experiment once again compared a position based recall task in 3D and 2D rendering. The added perspective DID NOT aid in the recall task. Participant's estimation of their own performance was similar across 2d and 3d conditions.

"The results support our prior work showing that perspective effects in monocular static computer displays do not significantly influence the effectiveness of spatial memory."


Empirical Study of a 3D Visualization for Information Retrieval Tasks

Gregory B. Newby
[ACM Link]
Abstract: ... An experimental evaluation of a prototype interface for visualizing IR results is described. Results indicate that the 3D navigation interface for IR search results was usable, but that subjects had difficulty with some aspects. ...