Chapter 4: Cuing a Cut
4.1 Experiment 1: Introduction
126.96.36.199 Independent Variables
The editing conditions were controlled by three independent variables each with three levels: Exit Percent,Entry Percent, and Time Gap.
Figure 4-8: Diagram of Entry and Exit positions. The circles with dotted outlines indicate positions where the focal-object is partially or fully occluded by the screen edge.
There were four Exit groups: 0%, 50%, 100%, and Random. The percentage value referred to the amount of the focal-object off-screen in the frame immediately prior to the edit. An Exit value of 0% meant that all of the focal-object was on-screen but that it was touching the right screen edge. 50% meant only half of the focal-object could be seen whilst half was hidden behind the screen edge. 100% meant the focal- 62 Psychology Software Tools Inc. (http://www.pstnet.com/products/E-Prime/default.htm)
131 object had just left the screen edge prior to the edit (see Figure 4-8). In the Random condition the Exit percentage for each edit was randomly chosen from these three options at the start of each shot. This made it unpredictable. It was decided that the degree of Exit would be represented as the percentage of the focal-object off-screen as this is how the matched-exit/entrance cut has been specified by editors (see 188.8.131.52).
Exit Percent was designed to test the Occlusion Expectation and Advantage hypotheses.
Entry percent had six levels, -100%, -50%, 0%, 50%, 100%, and Random, referring to the amount of the focal-object on-screen in the frame immediately after the cut (see left circles in Figure 4-8). An Entry value of 0% meant that the focal-object was completely off-screen immediately after the cut but touching the left screen edge. An Entry value of 50% or 100% meant that it was half or fully on, respectively. An Entry of -50% or -100% meant that the focal-object was fully occluded and positioned with a gap equivalent to half (-50%) or all (100%) of its width between its right edge and the screen edge. The Random condition meant that the degree of entry was unpredictable and changed between shots.
Entry Percent was designed to investigate two different hypotheses. The positive Entry and Random conditions were chosen to investigate the pursuit initiation hypothesis (i.e. full occlusion after the edit is optimum; see 184.108.40.206). The negative Entry conditions and 0% Entry were used in combination with Time Gap to investigate the clearance hypothesis (see 220.127.116.11).
18.104.22.168.3 Time Gap
The Time Gap variable had four levels: 0 frames, 3 frames, 6 frames, or Random. It refers to the amount of time between the cut and the point at which the focal-object re-enters the screen (i.e. reaches 0% Entry). Time is measured in frames as this is the
132 smallest unit of time in an animation. The animation was presented at 24 frames per second so each frame was equal to 41.67ms.
To create these time gaps the focal object was positioned at negative Entry immediately after the cut. The object then moved towards the screen edge (out of view), taking the allotted Time Gap to reach 0% Entry and begin re-entering the screen. The animation was designed so that the focal-object moved a distance equivalent to a sixth of its own width every frame. Therefore, to create a Time Gap of 3 frames the focal-object was positioned at -50% Entry immediately after the cut (see Figure 4-8). The focal-object then required 3 frames to move to 0% Entry. For a Time Gap of 6 frames the focal-object must initially be positioned at -100% Entry63.
22.214.171.124 Experimental Conditions
The Entry and Time Gap variables were varied within subjects and the Exit variable between subjects. A complete crossing of these three variables within subjects was not possible due to the large number of experimental conditions that would have been required. By setting Exit as a between-subjects variable it could be used to investigate whether subjects showed better adaptation to 100% Exit compared to the other Exit values (occlusion advantage hypothesis).
At the start of each experiment, subjects were allocated to one of the four Exit groups (0%, 50%, 100%, and Random). For each subject, the Exit value would then always be the same for every cut during the entire experiment. The only exception was the Random Exit group for whom the Exit would always be Random i.e. unpredictable.
63 The decision to have the focal-object move during the Time Gap rather than just pause the animation was made because 1) it was thought to be more authentic, when Clearing the frame the film is not paused, and 2) for technical reasons. If the animation had stopped there was an increased chance that the display duration of each frame would change from that achieved whilst moving the object.
133 The remaining variables, Entry and Time Gap, were mixed to create nine experimental conditions (see Table 4-1 page 133). Four of these conditions (2, 4, 6, and 8 in Table 4-1) varied Entry whilst controlling Time Gap as 0 frames. Conditions 2, 3, 5, and 7 varied Time Gap whilst Entry changed to accommodate (see 126.96.36.199.3). The focal-object in these Time Gap conditions was always at 0% Entry once the Time Gap reached 0 frames.
Condition Number Entry (% of object) Time Gap (frames) 1 Random Random 2 0 0 3 0 Random 4 50 0 5 -100 6 6 100 0 7 -50 3 8 Random 0 9 Random Random
Table 4-1: Experimental Conditions
Two extra conditions were included: 1 and 9. Both Entry and Time Gap were randomized in these conditions. These conditions were used to test the occlusion advantage hypothesis. By comparing performance in condition 9 to condition 1 any adaptation to the Exit condition, which is consistent for each subject, should have been detected. Entry and Time Gap were randomized to ensure that any increase in performance was due to Exit not either of the other factors.
The full list of experimental conditions can be seen in Table 4-1. Conditions 1 and 9 were always presented first and last, respectively. The presentation order of the other seven conditions was rotated within Exit subject groups to protect against learning, task improvement, and fatigue. Seven different permutations were needed to ensure that each condition was presented at each position during the experiment. Balancing
134 this across the four subject groups (for the four Exit levels) required a minimum of 28 subjects.
The subject’s task during this study was to follow the focal-object throughout the animation, over a series of cuts, and react as quickly and accurately as possible to a reaction time (RT) cue. The RT cue was a black digit from 2 to 9 superimposed on top of the focal-object. It was presented for 1 frame (41.67ms at 24fps); long enough for the digit to be identified if it is being fixated but not if it is only seen peripherally. The presentation time of 41.67ms is also too short for a saccade to be performed to fixate it. The subject’s task was to identify if the digit was odd or even and respond by pressing the appropriate button: ‘E’ for even and ‘O’ for odd (on a QWERTY keyboard). The hands used to press each key were counter-balanced across subjects to protect against handedness bias. Subjects were told that accuracy and speed were important so they should respond as soon as they think they can correctly identify the digit. If they were unsure they were asked not to guess. This was done to limit the effect on guessing on the correct response rates.
The correct response rate and average reaction time for each cue position were recorded as dependent variables.