Perception of variety in forests in relation to reducing polygon count : Exploring human perception in computer graphics and computer graphics education

Abstract: By taking human perception into account in computer graphics, time and processing power may be utilised more efficiently when modelling and rendering scenes. Exploring possibilities for reducing the complexity of forest scenes is particularly interesting because of the geometric complexity of trees, the effects of which are compounded by dense, large-scale forests. Most studies of rendering simplifications for forests concern level-of-detail methods, simplifying scenes in areas where simplifications are less perceptible to the viewer. The present study instead approaches the topic by considering that humans, to certain extents, are insensitive to the exact composition of objects in aggregates. In forest scenes that use a variety of geometrically different trees, this can be utilised to favour the geometrically simpler tree. A perceptual study was conducted to explore how perceptive viewers are to the ratio between trees, in forest scenes that feature two geometrically different tree models. Specifically, the tree models varied in leaf and branch density. The impacts of forest density, layout and camera angle were studied. Overall, it was found that density and layout had significant effects on thresholds measured for participants’ perceptiveness of ratio, while camera angle had little impact. The results indicate that the ratio between two different tree models can be skewed considerably in favour of the geometrically simpler model, without participants detecting changes. These findings provide some guidelines for reducing polygon count in forest scenes while preserving appearance. Motivated by these findings, the educational part of this study considers the role of perceptual topics in computer graphics courses. Specifically, the study focuses on the inclusion of perceptual topics in introductory computer graphics courses at Swedish universities, as well as the attitudes of computer graphics educators toward teaching perception. 22 course syllabuses at 12 Swedish universities were collected and analysed. Interviews were conducted with three computer graphics educators, while 13 educators responded to an online survey. The results present a diverse set of approaches and attitudes toward perception among educators, signifying that the role of perception in computer graphics education is uncertain. Perceptual topics were included in about half of the surveyed courses, with teaching approaches ranging from marginally including perceptual topics, to fully dedicated lectures, though perception was rarely mentioned in the intended learning outcomes of courses. Overall, approaches seemed influenced mainly by the background of the student group, educators’ own biases and experiences, as well as challenges posed by needing to address foundational skills such as programming. By investigating perceptual thresholds through a psychophysical experiment and exploring pedagogical practices in relation to perceptual topics in computer graphics courses, this study links empirical research with pedagogy, highlighting the role of perception in computer graphics from different perspectives.