Digital storytelling significantly improves the immersion of the users into virtual environments. The perception of the information contained in the digital story is better perceived if the story is told by a real avatar, rather than the animated character. The paper describes how we improved the approach of inserting the real avatar recorded against a green screen using a sequence of images with an alpha channel in an X3D real time virtual environment.

Smell is a key human sense which can significantly effect our perception of an environment. Although, typically not as developed as our other senses, the presence of a pleasant or unpleasant smell can alter the way we view a scene. Such a cross-modal effect can be substantial with parts of a scene literally going unnoticed as the smell dominates our senses. This paper investigates the cross-modal affect on the perception of the real-time animation of a field of grass in the presence of the smell of cut-grass. Rendering the high level of detail of a close-up view of a field of grass is computationally very demanding. In the real world the smell of grass would be present, and especially strong if the grass had just been cut, for example in preparation for a sports event. By exploiting the cross-modal interaction between smell and visuals we are able to render a lower quality version of a field of grass at a reduced computational cost, without the viewer being aware of the quality difference compared to a high quality version.

Visual perception is becoming increasingly important in computer graphics. Research on human visual perception has led to the development of perception driven computer graphics techniques, where knowledge of the human visual system and, in particular, its weaknesses are exploited when rendering and displaying 3D graphics. It is well known that many sensory stimuli, including smell, may influence the amount of cognitive resources available to a viewer to perform a visual task. In this paper we investigate the influence smell effects have on the perception of object quality in a rendered image. We show how we can potentially accelerate the rendering of images by directing the viewer's attention towards the source of a smell and selectively rendering at high quality only the smell emitting objects. Other parts of an image can be rendered at a lower quality without the viewer being aware of this quality difference. By doing this, we can significantly reduce rendering time without any loss in the user's perception of delivered quality.