Thesis

In this work a conceptual model is proposed and implemented for a tool capable of applying image filters, based on the behaviour of ant swarm. This tool transfers some of the creative responsibility from the user to itself, when processing an image. To do this, the tool uses a property of the swarm behaviour, the emergence of complex behaviours from simple actions, to process the images. A brief discussion about creativity is made, examples of how swarm has been used are demonstrated and how to perform image processing is explained. The conceptual model proposed is based on layers containing information of the image/images and the swarm/swarms. Based on the parameters of the user, the ants of each swarm behave accordingly, modifying the image by applying local image filters. From the answers of the survey, the results show that even when the user felt he wasn't in full control of the final result, the images produced had an interesting visual impact. This demonstrates that the tool plays an important part on the creative process of producing those images.

The modelling of believable synthetic characters allows analogies with subjects from areas like Biology and Psychology. So, the modelling of synthetic characters might consider the question of survival in living organisms. Based on survival, appear many homeostatic variables that define the balance of an organism. The lack of balance of an organism requires its regulation and interferes with the way of life. A synthetic character might also have homeostatic properties that define its balance on its way of life reacting to the lack of balance of those properties using planning or recurring to known behaviours, seen as preferential, ie, useful behaviours to achieve goals that can increase or decrease its strength over time, in several circumstances. Regarding the second one it is possible to say that it is the concept of habit. If the character can adapt to the environment where acts and demonstrate a behaviour that allows an external observer to detect habits on its behaviour, is believed that the level of credibility might increase. Therefore, it is proposed a model to the adoption of habits from the synthetic characters that assumes certain premises in order to be implemented on architecture for synthetic characters and considers, for example, the behaviour's success level and the circumstances in that are taken. The model was later implemented on architecture for synthetic characters and evaluated. From the evaluation is it possible to conclude that the model is adequate so that external observers detect habits on the characters and, sometimes, even increase its credibility.

The evolution of communication strategies in society now enables culture and entertainment to reach different forms of diffusion such as digital media. Museums and other heritage institutions start to change their storytelling styles, in order to offer different and engaging experiences. An example is to use Interactive Storytelling as a communication media to guide and provide visitors with information on the presented heritage, while increasing amusement and curiosity about the collections of items and transforming exhibits into narrative spaces. Such changes reshape museums' identity in society, turning them into institutions of culture and education diffusion, competing with other entertainment places like cinemas and theaters. In this thesis we propose the ISGT system, a Storytelling engine that guides visitors in interactive exhibits with an augmented reality character. Using the YDreams' Showroom and technology as a case study implementation, we were able to simulate an exhibit with the virtual guide, which provided users with suggestions and information on a collection of items. While the results suggest an increase of visitors' curiosity and the duration of interactions with the presented heritage, we believe that the system's purpose is to provide a customizable base for creating case-by-case guided tours with a minimal development effort.

With the advent of modern processing hardware, it became possible to bring global illumination effects into real-time applications. Namely, several classic global illumination algorithms were recently adapted to run in real-time on the GPU while ray tracing has also become suitable for real-time rendering on the CPU. Although none of these techniques provides a complete solution for simulating illumination in a realistic way, each one of them provides a different but complementary contribution for achieving this purpose. In this line of thought, this thesis introduces a 3D rendering engine that combines GPU rasterization and CPU ray tracing techniques in order to approximate global illumination in real-time. The engine simulates direct lighting using typical local illumination models while it simulates indirect lighting through an implementation of the light propagation volumes technique and a new sky lighting technique that is presented in this thesis which provides realistic indirect lighting for outdoor environments. These illumination effects are then complemented by sharp reflections and refractions generated by an hybrid real-time ray tracer that combines the processing power of the CPU and the GPU to generate these effects efficiently and to combine them seamlessly with the rest of the lighting.