EDITING ATTENTION: the perceptual foundations of continuity editing

1
EDITING ATTENTION the perceptual foundations of
continuity editing

• Tim J. Smith
• tim.smith_at_ed.ac.uk
• http//homepages.inf.ed.ac.uk/s9732397
• School of Informatics, University of Edinburgh
• Supervisors John Lee, Helen Pain, Graeme Ritchie
• CCSMI Conference, July 23rd, 2004

2
Continuity Editing

• What?
• A system of heuristics, guidelines, and patterns
  specifying how different camera shots can be
  edited together.
• Why?
• To make the visual transition between shots
  invisible to the viewer.
• Avoid a jarring transition Defensive Response
• To ensure viewer comprehension of space, action,
  and narrative across an edit.
• Example
• the 180 Rule. v
  X

3
Continuity Editing

• Why does it work?
• Analogue for real-world shifts of point-of-view
  (Griffith Lindgren).
• Eye blink as a mental cut (Huston Murch).
• Analogue of mental experience (Munsterberg
  Reisz Messaris).
• Inertia of narrative comprehension (Fairservice).
• Compatibility with rules of ecological
  perception (Gibson Anderson).
• Continuity of eye location and movement across
  the screen (Block).
• Goal of this study

HIGH
Level of perceptual processing
LOW
To empirically investigate viewers awareness of
an edit at the level of graphical perception.
4
A simple edit Lateral Stitch
5
Cutting for Continuity

• What is the best way to edit the Lateral Stitch?
• The cut should be made whilst the subject is
  still partially within the frame (Katz, 1991).
• 50 Exit
• The subject should be partially within the frame
  immediately after the cut (Katz, 1991).
• 50 Entry
• Why?
• The effect is to make the cut smoother and to
  speed up the flow of action (Katz, 1991).

6
Questions

• Do viewers construct visual expectations based on
  the principal objects motion relative to the
  screen edge?
• Are the suggestions of continuity editing
  compatible with these expectations?
• Do these expectations result in periods of low
  attention during which the visual discontinuity
  of the edit could be perceptually hidden?

7
Visual Occlusion

• Real-world objects do not suddenly cease to
  exist, change identity, or relocate without
  associated transitions (Michotte, 1950).
• Effects of occlusion
• Continued object perception during occlusion
  (Baker et al., 2001).
• Maintain spatiotemporal expectations.
• Easier initiation of occular smooth pursuit.
• We utilise the same perceptual assumptions when
  viewing film as in real-world perception
  (Anderson, 1996).
• Hypothesis
• The viewer expects the principal object to be
  fully occluded on exit (100) and on entry (0)
  to create the maximum impression of continuity.

8
Attention and the Saccade

• Goal is to relocate fovea, most sensitive part of
  the retina
• Visual attention is attenuated during saccadic
  eye movements due to saccadic suppression
• Duration 200ms
• 50-100ms preparation and recovery
• 30ms actual movement
• Therefore
• A visual discontinuity can be hidden in a saccade
  if the change occurs during the period of
  saccadic suppression.
• Hypothesis
• Predictable edits (100 Exit 0 Entry) will
  show a fading out of attention before the edit
  and a brief fading in after the edit.

9
Experiment

• 9 different animations.
• Identical except for Entry and Exit conditions.
• Exit 0, 50, 100, Random
• Entry 0, 50, 100, Random
• Attention tested at 13 points across the screen.
• Accuracy measured
• Reaction times and smoothness ratings also
  recorded (not reported here).
• 28 subjects

10
Results

• 100 Exit shows evidence of predictive saccades.
• 0 Entry shows slowest recovery of attention
• 100 Entry creates maximum attention, but
• 50 Entry shows predictive saccading.

Ideal distribution of attention across a
predictive saccade
Percentage of Correct Response ()
Cue position relative to edit (frames)
11
Results An Ideal Edit?

• 100 Exit
• 50 Entry
• optimum perception?

• 100 Exit
• 100 Entry
• optimum attention

Continuity Rules specified the ideal edit as
50 Exit 50 Entry Our results support this
whilst improving upon it. But, without a method
of testing continuity perception we cannot show
the benefit of occlusion on entry (50 Entry).
12
Conclusion

• In the Lateral Stitch viewers expect the
  principal object to be fully occluded by the
  screen edge before an edit occurs.
• This expectation manifests itself as saccade
  preparation prior to the edit.
• Initiating a saccade prior to the edit masks the
  visual discontinuity with saccadic suppression.
  This removes the viewers ability to directly
  perceive the edit and be affected by it (e.g.
  defensive response).
• Our evidence agrees with the continuity editing
  rules whilst improving their precision and
  validating them with empirical evidence.
• Future research will continue to unlock the
  connections between continuity editing,
  attention, and perception.

• Do viewers construct visual expectations based on
  the principal objects motion relative to the
  screen edge?
• Do these expectations result in periods of low
  attention during which the visual discontinuity
  of the edit could be perceptually hidden?
• Are the suggestions of continuity editing
  compatible with these expectations?

13
References

• Anderson, J.D. (1996) The reality of illusion An
  ecological approach to cognitive film theory
• Baker et al. (2001), Neuronal Representation of
  disappearing and hidden objects in temporal
  cortex of the macaque. Exp. Brain Research
  140375-381
• Katz, S. D. (1991) Film Directing Shot by Shot.
• McConkie, G. W., Currie, C. B. (1996). Visual
  stability across saccades while viewing complex
  pictures. Journal of Experimental Psychology
  Human Perception Performance, 22(3), 563-581.
• Michotte, A. (1950) On phenomenal permanence
  facts and theories.

Tim J. Smith tim.smith_at_ed.ac.uk http//homepages.i
nf.ed.ac.uk/s9732397
14
Experiment

• Hypotheses
• Full occlusion Trigger will show
• saccade preparation prior to the edit,
• quicker recovery of maximum attention, and
• the highest smoothness ratings.
• Full occlusion Entry will show
• a smooth increase to maximum attention following
  the edit,
• be rated as the smoothest edit.

15
Questionnaire

• Frame-by-frame, how smooth do you think the
  motion of the UFO was
• in the centre of the screen?
• very jerky (1) gt (2) gt (3) gt (4) gt (5) perfectly
  smooth no opinion
• on the right edge of the screen?
• very jerky (1) gt (2) gt (3) gt (4) gt (5) perfectly
  smooth no opinion
• on the left edge of the screen?
• very jerky (1) gt (2) gt (3) gt (4) gt (5) perfectly
  smooth no opinion
• After a cut, did you ever expect the UFO to
  appear on the screen when it didn't? Yes No
• At which time point during the whole animation
  did this happen most?
• early - middle - late - throughout
• How often did this happen?
• Rarely - Occasionally - Half - Most -
  All of the time Can't say
• Did the UFO ever seem to jump unexpectedly from
  one side of the screen to the other? Yes
  No
• At which time point during the whole animation
  did this happen most? early - middle -
  late - throughout
• How often did this happen?
• Rarely - Occasionally - Half - Most -
  All of the time Can't say

16
Results
Exit Reaction Times
Entry Reaction times
17
Results

• Attentional Fade-in duration (AFI)
• Hypothesis Long AFI period (6 frames) will show
  
• A higher mean level of attention, and
• shorter period until full attention recovery.

• Attentional Fade-in duration (AFI)
• Hypothesis Long AFI period (6 frames) will show
  
• A higher mean level of attention, and
• shorter period until full attention recovery.
• But, edits with long AFI durations are rated as
  being less smooth than those with a short AFI
  (3 frames).

18
Chronostasis
19
Continuity Editing (2)

• How? 180 Rule1

1 David Bordwell and Kristin Thompson. (2001).
Film art an introduction, 6th edition
20
Visual Attention

• The focusing of mental resources to the
  extraction and processing of a subset of the
  visual information existing in the scene.
• As Overt Orienting
• Fixations and Eye movements
• E.g. Saccades
• As Covert Orienting
• increased receptivity to sensory information
  outside of the fovea
• As Processing Resources
• distributing cognitive resources between the
  perception, restructuring, and storage of the
  internally represented information.

21
Summary Did we achieve our goal?

• Show how film editing can influence eye movements
  and the distribution of attention across a scene.
  
• Occlusion by screen edge acts as trigger for
  saccade initiation.
• Overlap of action can accommodate saccade
  duration.
• Timing of object re-entry to coincide with
  recovery of full attention can maximise attention
  during the objects entire screen time.
• Show how continuity editing can create the
  perception of continuity by hiding the edit in
  periods of low attention.
• Editing decisions can be operationalised and
  their effects empirically tested.
• Potentially, Edits can be hidden in saccadic
  eye movements.
• Occlusion by screen edge creates potential for
  continuity.
• But, viewers cant introspect on continuity.
• Empirical evidence validates continuity editing
  rules.

• Show how film editing can influence eye movements
  and the distribution of visual attention across a
  scene.
• Show how Continuity Editing creates the
  perception of continuity by hiding an edit in
  periods of low attention.