Computer graphics field is rapidly growing and is widely used in many industries, especially in cinematics. The visual effects, which is a part of the field, have become more affordable in comparison to special effect. By using more visual effects, safety during the shooting increases and the use of resources decreases. However, the field is still young and there are a lot of flaws to consider when creating such an effect. It is important that CGI in movies are created in such a way that it does not disrupt the viewing experience unless it’s done on purpose. Therefore, CGI in a real-life setting should look real. To create CGI that is closer to reality, this study examines the differences between special effects and visual effects in the aspect of the viewing experience. To investigate this question, we conducted an experiment following a semi-structured interview. 13 students participated in this study where they had to watch video clips representing the special respective visual effect. During the experiment, they did a self- report by filling in a questionnaire. A Valence-Arousal model was used to examine the subject’s emotional reactions. The result suggests that there are no significant differences between the two effects in regard to an emotional reaction. However, it shows that the visual effect had flaws of being too plastic and too perfect which makes it less convincing compared to the special effects. In contrary, the special effects do not have the ability to be exaggerating, which in turn makes it more exciting to watch the visual effects. Although this study does not solve any flaws, it highlights those that need to be fixed for the goals of improving CGI. / Datorgrafik är ett ständigt växande ämne och tekniken kan appliceras så många olika sätt, speciellt inom film. Visuella effekter som är en del av ämnet som har blivit ett mer prisvärt alternativ i jämförelse till specialeffekter. Genom att använda mer visuella effekter, har säkerheten i filmandet ökat och användningen av resurser minskat. Men datagrafiksämnet är fortfarande relativt ny och det finns många brister att tänka på när det gäller att skapa en visuell effekt. Det är viktigt att CGI i filmer är skapad på sådant sätt att den inte stör tittarupplevelsen, såvida det inte är med mening. CGI i en verklighetstrogen miljö bör därför se verklighetstrogen ut. För att skapa CGI som är närmare verkligheten ska denna studien undersöka skillnaden mellan specialeffekter och visuella effekter baserad på tittarupplevelsen. För att undersöka denna fråga har vi utfört ett experiment med följd av en semi-strukturerad intervju. 13 studenter deltog denna studie där de har kollat på videoklipp med specialeffekter respektive visuella effekter. Under experimentet fick de fylla i en enkät där de utvärderar deras reaktion till klippen samt vad de tyckte om respektive klipp. En Valence-Arousal-modell användes för att undersöka deltagarnas emotionella reaktion. Resultatet visar att det inte är någon signifikant skillnad mellan de två effekterna när det gäller den emotionella reaktionen. Däremot indikerar att de visuella effekterna var för konstlad samt för perfekt i jämförelse till specialeffekter. Dessa nackdelar gjorde effekten mindre övertygande. Emellertid kan specialeffekterna inte överdrivas i samma skala som visuella effekter vilket gör den visuella effekten mer spännande. Denna studie löser inte de problem som visuella effekter har idag, men den tar fram problem som behöver fixas för att förbättra tekniken.

Music mood recognition by machine continues to attract attention from both academia and industry. This thesis explores the hypothesis that the music emotion problem is circular, and is a primary step in determining the efficacy of circular regression as a machine learning method for automatic music mood recognition. This hypothesis is tested through experiments conducted using instances of the two commonly accepted models of affect used in machine learning (categorical and two-dimensional), as well as on an original circular model proposed by the author. Polygonal approximations of circular regression are proposed as a practical way to investigate whether the circularity of the annotations can be exploited. An original dataset assembled and annotated for the models is also presented. Next, the architecture and implementation choices of all three models are given, with an emphasis on the new polygonal approximations of circular regression. Experiments with different polygons demonstrate consistent and in some cases significant improvements over the categorical model on a dataset containing ambiguous extracts (ones for which the human annotators did not fully agree upon). Through a comprehensive analysis of the results, errors and inconsistencies observed, evidence is provided that mood recognition can be improved if approached as a circular problem. Finally, a proposed multi-tagging strategy based on the circular predictions is put forward as a pragmatic method to automatically annotate music based on the circular model. / Graduate / 0984 / 0800 / 0413 / zazz101@hotmail.com