|
|
Ivan Bretan and Per Kroon
Telia Research AB
S-136 80 Haninge
SWEDEN
{ivan.bretan,per.kroon}@haninge.trab.se
The design of a user environment for a video-on-demand service
through an interdisciplinary style of collaboration called îconcurrent
engineering" is described. The process encompasses pre-prototype
behavioural studies, traditional user studies, graphical design
of interface objects, industrial design of input devices and interaction
design of interface dialogue.
Interdisciplinary design, interactive TV, video-on-demand.
A US video rental store may stock in excess of 15000 titles, and, as has often been remarked, video-on-demand (VOD) is video rental pursued through a different distribution channel. Together with the fact that computerised tools make it easier to cope with a large selection of films, this leads us to conclude that it is reasonable to expect video-on-demand interfaces to have to support the navigation among and selection from a very large volume of films.
The need for interdisciplinary design of user environments has
been accentuated the last years (see e.g. [1]) as additional degrees
of freedom have been introduced into interface construction by
multimedia technology. The most striking need for additional competence
lies in the area of graphical design, since multimedia often means
visually intense interfaces. As the discipline of human-computer
interaction moves into the realm of consumer electronics, it is
obvious that the knowledge of industrial designers concerning
ergonomic design of physical equipment becomes increasingly interesting.
People with this type of background could therefore be a natural
part of user interface development teams, along with interaction
designers, usability experts and many other types of specialists.
Such an interdisciplinary style of working is well-known in the
car manufacturing industry, where designers, human factors specialists
and engineers work together on new car designs from the outset.
Concurrent engineering means parallel interdisciplinary work,
communicating over wide channels and a broad agenda, as opposed
to sequential, segmented team-work with well-defined narrow interfaces.
At Telia Research, a multidisciplinary student project has been
carried out with the goal of assessing the benefits of concurrent
engineering by applying this method to the experimental development
of a VOD user interface prototype.
What is it people do when they rent films using a VOD service?
That is hard to answer, since such services are in their infancy.
In order to provide a rationale for the design of VOD interfaces
from a user-centred point of view, it makes sense to study the
more general search and decision processes that are part of ordinary
video rental. We have conducted detailed interviews with 18 persons
renting videos from a video store in central Stockholm, from which
we derived the following behavioural scheme: 1) Check the chart
of films most frequently rented and/or the shelf containing new
films. If this proves unsatisfactory, move on to one or several
specific categories (action, drama, etc.). 2) Pick up a film recognised
due to marketing (magazine advertisements, trailers, etc.), recommended
by a friend or a reviewer, or just because the cover looks interesting.
3) Check to see whether any well-known actor stars in the film
(and/or whether a well-known director made the film). Read the
information on the reverse of the cassette. 4) Apply remaining
selection criteria: production year, store recommendations (from
staff or other customers), playing time, price etc. Categories
are often used as filters: there is no need to browse among the
films belonging to a category known to be uninteresting. Although
the study did not produce sufficient data to confirm this, we
do believe that a select and rank strategy is relatively common:
gathering, physically or mentally, several films and then selecting
one or a few from this set (particularly useful when more than
one viewer is involved in the decision process).
The behavioural studies revealed that the picture - cassette cover or poster - is by far the most important piece of information used by the video renter when deciding which film to view. Therefore, dominating the main screen in our VOD interface design is a sequential presentation of five movie posters, constituting a magnified portion of a virtual reel of films, where the middle poster is îin focusî (figure 1). Since the posters are designed by the film companies to be graphically attractive and eye-catching, there is no need - and at any rate no space - for any colourful interface widgets but the most important ones. Due to the limited resolution of a TV-screen, more than five posters would hardly be useful. The icons to the left and right on the screen are actually superimposed over the movie pictures, transparent but clearly visible since the leftmost and rightmost parts of the pictures are shaded to black. The film poster sequence is aligned with an Alphaslider [2] used to specify which portion of the film reel is displayed. The size of the Alphaslider îthumb" can be varied to illustrate the ratio of the five currently shown films to the total reel. The general feeling of the visual design is relatively two-dimensional, giving associations to modern magazine layout. Experiments were also carried out using a version of the graphical design which added a sense of depth, but the evidence was inconclusive as to whether this actually increased the aesthetical or functional value of the interface.
Since the interface is designed with an interactive television
setting in mind, the natural choice for an input device is some
kind of remote control. The user must be able to carry out all
actions available in a whole range of interactive television services
using the same device, including controlling a video (pause, rewind
etc.), entering a personal code and, of course, moving a pointer.
Figure 2 shows the type of remote control that the industrial
designers of the project eventually arrived at after user tests
and input from the other team members. It has a finger joystick
controlling the pointer's position on the screen, operated using
minimal thumb movements, while the index finger rests on the select
button. This design is well suited to operating the Alphaslider
and creates a restful grip for the hand and relieves the entire
lower arm. Note that the remote is symmetrical and therefore also
easy to use for a left-handed person. The slim shape of the remote
has been made possible by leaving out an alphanumerical keyboard,
relying on Alphasliders and virtual keyboards. An alternative
version of the remote having an extendible alphanumerical keyboard
has also been developed, but it is ergonomically somewhat inferior.
The user navigates among the videos using the Alphaslider located
below the five-poster sequence, encompassing the entire search
space of the film reel, or by dragging the poster sequence to
the left or right. The Alphaslider has several indisputable advantages,
among which some of the most important ones are: the fact that
no typing is needed (reducing the need for a alphanumerical keyboard
on the remote); that the user only needs to recognise rather than
recall a name; and that no misspellings are possible. By clicking
on one of the icons on the left-hand side of the display, the
user specifies the movie category (action, comedy, drama, horror
or children), i.e., which reel of films is shown. The icons in
the group above changes the sorting order of the films on the
reel (by recency, by popularity (rental frequency) or alphabetically
by title). The same method is used regardless of whether the user
is searching for a particular film or just browsing. When an interesting
film has been found the user can obtain more information by clicking
on the picture or using a colour-coded key on the remote control.
The information provided with each video is: a plot summary, the
cast and other data from the recording, a trailer (and perhaps
an interactive commercial, when available). An additional click
with the remote orders the film.
Advanced searches, where the user for instance wants to select
a film from the set of movies recorded in Sweden in the seventies,
are carried out by clicking on the search icon. The user then
selects the search criteria, choosing between actor, director,
country, composer, specialised genre (westerns, science fiction
etc.) and many more - different criteria can be combined. Selecting
actor, for instance, is done by means of an Alphaslider, which
at this point will be a technique familiar to the user. Having
specified the actor, the reel displays films from the relevant
set, which can be sorted in the same way as earlier. Additional
conditions can then be added incrementally in the same way. There
are occasions when an interesting video will be found but where
the viewing of the film has to be postponed, for instance if the
film found requires more time than is available. The user can
then put the film in a îbasket" (which in effect is
a user-specified category, or reel of films), saving it for a
later occasion. Clicking on the basket icon shows the films currently
collected there. This function is also useful when a group of
people is trying to agree on a movie to watch. Each person in
the group can then put films that they are interested in into
the basket and eventually compare the candidates, settling on
a choice that satisfies everyone (or annoys no-one). There is
also an option of side-tracking. If the user turns this function
on at a time when it is indicated as available, a link to another
relevant film will be presented. The reasons for linking videos
together vary: they are based on the same story; they are filmed
at the same location; or perhaps they have a theme in common,
like aristocratic decadence or urban gang wars.
We would like to thank the other DUVA project members: Fredrik
Ampler, Walter Frost, Lukas Hansson, Thomas Nyström, Markku
Virtanen and Jon Ölmeskog. Thanks also to Catriona MacDermid
for valuable comments.