USING AGENTS TO ASSIST SPECTATORS TO ORIENT IN INTERACTIVE DRAMATURGY

Vladan Zdravkovic’

University of Gävle, Sweden

ABSTRACT

Non-linear and interactive dramaturgy and story telling can present problems to inexperienced spectators. Younger generations, familiar with computer games, multimedia and Internet are more or less successful in navigation through interactive and non-linear story telling. But people not used to new technology could have problems in orientation as well in reception of content. Even perception could be difficult. We propose a new method for help, as a necessary tool for interactive film and TV applications. The tool is based on agents. In this version an agent is dedicated to specific DVD content, but in the future it could be developed for wide use, as a global agent for non-linear contents. We suggest several solutions, depending on the age, experience and preferences of the users. Users with different levels of skill and experience should have different types of agent.

Introduction

Present discussions show that DVD authors don’t use the potentials of DVD interactivity [3]. Research into Web communication concerning interactive aspects of the Web shows up problems in users’ navigation and usage of interactive features [13], [1]. Research in human computer interactions [10] shows that users could have problems using animated agents. Problems could be anxiety and bad task performance. Research in media equation shows that human – media communication has depended on way of media performance and that people equate media as a humans [9]. The DVD video market is faced with several obstacles. The first is that users are familiar with ways of looking at VHS tapes that offer linear dramaturgy and implement two-way linear navigation. DVD video has a strong potential, especially in interactivity, but most users don’t know how to use this opportunity. The easiest way to enable them to use it would be to use an agent. But what should the agent look like? Could we build an agent upon different users' preferences? Could our agent cope with different users' age, gender and geographic location? Finally should the agent be animated or static? The literature shows that the use of agents has its advantages and disadvantages [1], [10]. Another big problem is the way of using agents in DVD applications. Most of DVD videos have interactive aspects only in the menu level, not in the content level. There are still very rear interactive films or videos so usage of that kind of interactivity is not even in the development. During the last year we are developing two DVD video projects based on interactive and non-linear dramaturgy structure. The first is Babylon – an interactive film, another is Atlantis - a computer animation – feature - documentary interactive film.

Atlantis

Atlantis is an interactive film directed by X, which contains 3D-computer animation, feature and documentary material. It has 3 different beginnings, more then 25 choices during the film, and 5 different ends. The film is currently in production. 42 minutes of animation have been done by studio Wonderland, and during the summer of 2002 the documentary part of the film will be produced. During the autumn 2002, filming with actors will be carried out. Finally, in the winter complete postproduction for interactive and linear versions will be ready. From the interactive film will be taken 4 linear derivatives for cinema and video distribution. In the DVD version 6 different animated characters will help to spectator in navigation through menus as well as in watching the movie. During watching, the area in the bottom left-hand part of the screen will be devoted to the signal for Interactivity. Choosing of path of interactivity could be done with or without one of 6 interactive agents. From the start of the DVD, Agent number 1 – the Boy will turn on in case the spectator asks for his help. At this moment, the spectator could choose to continue with the Boy or change to Neva, Adam, the Dolphin, the Fish or the Monster - five other agents. Spectators will also be able to choose to continue without agent help. During the film, the agent appears if the spectator chosen the agent’s appearance from the start. Otherwise, the agent does not appear, but if the spectator twice in succession presses any button that is illogical or does not make sense, the agent chosen from the start of the DVD (or the Boy if no one agent is chosen) will appear.

Babylon

Babylon is the first feature film based on an interactive manifest [14]. The spectator will be able to see 128 different versions (different in some important aspects) and more than 500 other versions different from the main 128 versions in smaller details (mostly sound). The target viewing media is DVD or some other similar media. But it could be seen in cinema or TV on classical - linear way. That is possible thanks to the fact that film has its linear derivatives. There are 9 linear derivatives, two of them are finished, and we are in process of finalizing the next 3. The others need additional filming that needs to be completed. Every of those 9 linear versions have 3 subversions (with smaller differences like different endings, that makes a total of 27 linear versions.) Any spectator viewing the film could choose from several types of music, which make the film more or less dramatic, increase or diminish the suspense, make the film sad or happy. That kind of interactivity is reflected in “sub-versions”. In future, the spectator could choose his own music and implement it into the movie. That is not possible now, but in next year, that will be one of the spectator possibilities. Spectators could choose all the aspects: type of film, even genre, atmosphere, main character, way of developing the story, the beginning and the ending of the film. The target viewing media for such film is interactive film theatre, but it could also be viewed in ordinary cinemas (just linear versions). Two derivatives Babylon - North Cup and Babylon - Keep Walking have been made in accordance with Dogme 95 film rules. The DVD release will have no animated agents, but it will have a voice agent. Regarding the voice agent, the spectator will be able to choose between 3 different voices and 3 languages.

PROBLEMS

Spectators are not familiar with interactive films and video so the main obstacle regarding such projects is to find the way that spectators will use the advantages of interactivity in DVD video. Even when the spectator is an experienced computer user, aspects in interactive story telling are not familiar. The spectator should be aware of the potentials of interactive dramaturgy and how to exploit non-linearity maximally. Several questions arise. How will spectators be informed that places of interactivity occur, without disturbing the flow of story? How will the warning of interactive potentials (hot spot) be presented to them? How long they will be able to react when they are aware that they can? The solution is the introduction of an agent that will help the spectator in navigation, not just through these places, but also through navigation in the whole film.

Human To Media Interaction Aspects

decided to implement animated agents as a solution for improving communicational aspects. In Babylon, we don't implement an animated agent because of the lack of funds necessary for building a model and its animation. We defined different voices and visual appearances so as to have adequate agent for different kinds of users. We divided users into 3 groups on the basis of their experience in working with interactive applications and familiarity with interactive dramaturgy. We defined them in subgroups of masculine - feminine (although there are 4 male and 2 female) users. Besides physical appearance and voice, we chose the agent's gestures and motions – they were animated in different ways in order to distinguish them. We started from the agent no. 1 - the Boy. Searching for the right voice, gestures and motion we find that a minimum of six different agents is necessary for successful implementation. We chose to distinguish two user characteristics, although literature suggests more factors. The main reason is the fact that funds and time didn't allow more characters, so our framework should fulfill the number six. Although we planned from the start to distinguish other characteristic regarding different psychological types of learning, we decided to keep experience as the main factor. In that sense, we divided potential users into 3 groups: beginners, users with some knowledge, and experienced users. For beginners, we targeted Adam and Eva, agents with humanoid faces, smooth motions and human gestures. Their voices are gentle and explanations are broad. For users with some knowledge, we targeted Fish and Dolphin. They are faster in motion and gestures. Their voices are less gentle and tempo of speaking and explaining are faster. For experienced users, we targeted Monster and Boy. We didn't opt for a female character (although the Boy has motions and somewhat feminine gestures) because of the results from Reeves and Nass [9] pp 164. They state that male-voiced computers (we are applying that on DVD) are seen as more knowledgeable about technical subjects then female-voiced, and evaluation from a male -voice is taken more seriously. The Monster was chosen because we wish to add an additional humorous aspect in communication with the experienced user. The reason for that is problems that especially experienced users could have regarding the use of agents, as stated in the introduction. During internal testing, only the agent “Monster” wasn’t evaluated negatively by the experienced users. The Monster has a strong voice, with auto parody elements and violent gestures and motion. The Boy has also elements of self-parody in his gestures and his characteristics could be described as those of a comical TV commentator.

Media Aspects

The complexity of interactive dramaturgy makes sense if spectators have benefits. The idea with interactivity in Atlantis and Babylon is to provide a film closer to spectators' tastes and moods. Also, using an agent should be amusing, bringing a sense of adventure and play. An agent shouldn't disturb all the other aspects of the film as a linear form. And that is impossible. The introduction of a toolbar or even a signal that interactive moments are enabled, a so called “hot spot”, brings awareness to the spectator. Also, the time of making the decision reduces spectators’ concentration on structures projected at that time. An even worst solution would be to stop the structure and leave an unlimited amount of time, where the film would be arrested. The solution that we implemented is the spectators' ability to change interactive paths all the time, until the new interactive choice comes. So, even when the time for the interactive place has passed, the spectator can make a hyper jump and go to another interactive path. If the spectator wishes, she or he could turn off many interactive, linear-looking versions offered from the many linear versions, and choose another aspect of the film before its projection.

Navigation Aspects

The first problem in interactive films is that spectators are not familiar with interactive films. Although most spectators are familiar with interactivity in using computers, most spectators are not familiar with DVD abilities. An agent seems a natural choice to assist the orientation of spectators in interactive structures. The use of an agent starts with instructions for using a toolbar and choosing interactive paths. It is not yet clear whether it is better to use animated agents, as in Atlantis, or simple voice agents, as in Babylon. This remains to be tested, but in the final instance, the decision could be left to the spectator, who would be able to choose from several forms of appearance. The final implementation of Atlantis will have incorporated agents in the form of bodiless agent for users who prefer not to use agents. An agent also helps in different problems related to non-linear storytelling, as a reminder for parts that have already passed, as helper in choosing a part where the application doesn't specify what the different possibilities are, and so on. An agent could indicate all the dramaturgical aspects of the chosen interactive path and offer reminders regarding the story, characters and all aspects that could be of interest to spectators. Finally, agents could exchange experiences over the Internet and show spectators the experiences of other users.

cONCLUSIONS

The model hasn't been properly tested yet, as the films are not ready, but on the basis of internal testing, it shows advantages, especially in the start of the DVD, during navigation, and in choosing an interactive path, especially for the inexperienced user. Proficient users didn't remark on the advantages in using an agent. The full testing is schedule for December 2001, but some appreciation has been found after internal testing. DVD needs at least three different types of agents, with different appearances and ways of helping spectators. Spectators should be able to configure the agent at the start of the application, ideally at the installation of the DVD device. The DVD standard committee should create a simple standard, where basic information about users’ preferences would be stored in the DVD. When the new DVD video starts, the video application should have information from the DVD player and make a basic initialization autonomously. In that initialization, the application should configure the user’s preferences regarding the agent's appearance and different information already available from the menu, such as language, subtitles, skipping trailers from the start, immediately starting the film, starting with the biography, starting with commentaries.

Future Investigations

In Atlantis, the agents are used as pre-rendered animation. Using of MPEG 7 support object control and with the ability of better definition of virtual characters, the animation will be rendered in real time. It will allow users to define on their own all the aspects regarding the appearance of the agent, such as his shape, size, texture and place on the screen, his motion, and eventually, the camera's motion. Future applications, especially interactive films, could have use of this, as users would be able to create an agent in accordance with their own wish. Additional investigation is necessary in psychological studies regarding the best possible use of agents

LITERATURE

1. Cassell, Justin, Sullivan Joseph, Prevost, Scott & Elizabeth, Churchill, (Ed) (2000). Embodied Conversational Agents, MIT Press, Cambridge, (375-398).

2. CHI 2001 Conference Proceedings. (2001). ACM – CHI, New York.

3. DVD Report, (2001). 6 (18), (1-2).

4. Heim, Michael (1998). Virtual Realism, Oxford University Press, New York.

5. Kender, J.R. & Yeo, B.L. (1998). Video Scene segmentation via Continuous Video Coherence, CVPR 1998, CVPR, Santa Barbara.

6. Luger, George F. & Stubbleifield, William A. (1998). Artificial Intelligence, Addison-Wesley, Harlow.

7. Moser, Mary Anne (Ed.). (1996) Immersed in Technology. Art and Virtual Environments, MIT press, Cambridge.

8. Norman, Donald. (1993). Things that Make Us Smart, Parceus Books, Reading.

9. Reeves, Byron & Nass, Clifford. (1996). The Media Equation, Cambridge University Press, Cambridge.

10. Rickenberg, Raoul & Reeves, Byron, The CHI 2000 Conference Proceedings, The Hague 2000, ACM – CHI, New York, (49 -56).

11. Russel, Stuart & Peter, Norving. (1995). Artifical Intelligence A Modern Approach, Prentice Hall, Upper Saddle River.

12. Vicer, Dan. (2001). Hyperactive and Interactive Literature, Multimedia, Belgrade.

13. Zdravkovic, Vladan (2000). Proceedings IWALT 2000, IEEE, Los Alamitos, (296-298).

14. Zdravkovic, Vladan. (1998). Interactive Art - Manifest, AFPC, Zuid-Scarwoude, (5).