| The image generation is taking a great importance nowadays in multiple areas. Increasingly, the level of realism of these virtual images is critical for their success, being the techniques based on radiosity one of the basic pillars to render realistic images.; This thesis is focused on the study and development of radiosity algorithms, contributing with developments and techniques that will make possible to run these algorithms on low-cost platforms (personal computers). The developed algorithms present high level capabilities, not only in the image quality but also in the efficiency. This is not a trivial problem because the main disadvantage of radiosity is the high hardware requirements that is needed to carry out its calculations. Moreover, these requirements are not only hardware (data processing power, memory) but also temporary, since the simulation of the behaviour of the light in which is based the radiosity is a costly operation.; Between the new features incorporated in this thesis, a novel technique based on intensities has been developed to achieve the radiosity calculations. Its main advantage is the 20% temporary cost reduction from the classic approach.; Likewise, the developments have been integrated to be used under PC-clusters with an almost lineal temporary reduction. This integration allows a new possibility of remote scene rendering, with the subsequent minimum hardware requirements for the client computer that request the render.; This thesis has been applied in virtual image generation of different environments---as rooms and scenes in general. The results achieved show the utility of the system in these cases. |
