Perceptually Driven Simplification for Interactive Rendering

Abstract

We present a framework for accelerating interactive rendering, grounded in psychophysical models of visual perception. This framework is applicable to multiresolution rendering techniques that use a hierarchy of local simplification operations. Our method drives those local operations directly by perceptual metrics; the effect of each simplification on the final image is considered in terms of the contrast the operation will induce in the image and the spatial frequency of the resulting change. A simple and conservative perceptual model determines under what conditions the simplification operation will be perceptible, enabling imperceptible simplification in which operations are performed only when judged imperceptible. Alternatively, simplifications may be ordered according to their perceptibility, providing a principled approach to best-effort rendering. We demonstrate this framework applied to view-dependent polygonal simplification. Our approach addresses many interesting topics in the acceleration of interactive rendering, including imperceptible simplification, silhouette preservation, and gaze-directed rendering.

Gaze-Contingent Multiresolutional Displays: An Integrative Review

“…In fact, several gaze-contingent multi-resolutional VR display systems have been developed (e.g., Levoy & Whitaker, 1990;Luebke, Hallen, Newfield, & Watson, 2000;Murphy & Duchowski, 2001, September). Each used different methods of producing and rendering gaze-contingent multi-resolutional 3D models, but all have resulted in a savings, with estimates of rendering time savings of roughly 80% over a standard constant-resolution alternative (Levoy & Whitaker, 1990;Murphy & Duchowski, 2001, September).…”

“…In most cases, the methods of multi-resolutional image production in Table 3 have been based on neurophysiological or psychophysical studies of peripheral vision, under the assumption that these research results will scale up to the more complex and natural viewing conditions of GCMRDs. This assumption has been explicitly tested in only a few studies that investigated the human factors characteristics of multiresolutional displays (Duchowski & McCormick, 1998;Geri & Zeevi, 1995;Kortum & Geisler, 1996b;Loschky, 2002;Luebke et al, 2000;Peli & Geri, 2001;Sere, Marendaz, & Herault, 2000;Yang, Coia, & Miller, 2001) Figure 2, panel A, by creating multi-resolutional images based on it and on functions with steeper and shallower drop-offs (as in Figure 2, panel D). Consistent with predictions, a resolution drop-off shallower than that in Figure 2, panel A was imperceptibly blurred, but steeper drop-offs were all perceptibly degraded compared to a constant high-resolution control condition.…”

Visual saliency guided textured model simplification

“…Therefore, researchers turned their focuses on generating similar rendered results for simplified objects against their original ones rather than focusing only on geometric difference minimisation. Luebke and Hallen [26] adopted the contrast sensitivity function (CSF), developing an empirical perceptual model to measure the human perceived quality of the rendered output from a simplified object. Qu and Meyer [34] developed a visual masking technique based on the Sarnoff visual discrimination metric and the visual masking tool in JPEG 2000, supporting mesh reduction according to surface texturing, light variation, surface reflectance, etc.…”

An efficient bit allocation for compressing normal meshes with an error-driven quantization

“…Several distortion measures have been exploited by single-rate mesh coders [16][17][18][19]. In this paper, we choose as reconstruction error the symmetric root mean square error between two surfaces [20] also called the S2S distance.…”

“…Several distortion measures have been exploited for compression of irregular meshes [16][17][18][19]. For instance, Karni and Gotsman (2000) introduce a metric which captures the visual difference between the original mesh and its approximation [17].…”