Stereo-Consistent Apparent Ridges
A Gradient-Based Geodesic Search for View-Dependent Lines in Virtual Reality
T. Tian (TU Delft - Electrical Engineering, Mathematics and Computer Science)
E. Eisemann – Mentor (TU Delft - Computer Graphics and Visualisation)
P. Kellnhofer – Mentor (TU Delft - Computer Graphics and Visualisation)
A. Zaidi – Mentor (TU Delft - Computer Graphics and Visualisation)
C.A. Raman – Graduation committee member (TU Delft - Pattern Recognition and Bioinformatics)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Virtual Reality (VR) demands high-performance stereoscopic rendering to maintain immersion, yet current techniques for Non-Photorealistic Rendering (NPR) often struggle with stereo-consistency. Despite line drawings, specifically apparent ridges, being a powerful tool for conveying shape and depth with minimal visual clutter, their view-dependent nature present a significant challenge in stereoscopic settings. As the viewpoint changes, view-dependent lines can have discrepancies between left and right viewpoints, triggering binocular rivalry and undermining the user's depth perception.
This thesis proposes a framework for synthesizing stereo-consistent lines through a gradient-based geodesic search in world-space. Rather than attempting to match existing view-dependent lines in stereo-viewpoints, we introduce a three view correspondence algorithm. Using the midpoint of the stereo viewpoints as an anchor, we use an iterative pathfinding algorithm that performs geodesic walks along the surface gradient of the curvature field. This process traces the transition of apparent ridges, a view-dependent line, from the center reference point to the stereo viewpoints, identifying consistent surface regions across the mesh.
To enhance depth and shape perception for the user, we extend this algorithm to apply hatching lines on the marked surface regions. By rendering curvature-aligned hatching lines along the geodesic paths, we provide additional surface cues that apparent ridges cannot convey. This method ensures that both eyes are presented with the same lines and information, eliminating most cases of binocular rivalry.
The efficacy of this approach is measured using a user study conducted in VR, alongside a performance analysis to ensure the algorithm meets the 90 FPS requirement of VR. Our results demonstrate that there is a statistically significant improvement in 3D shape and depth perception for users, while maintaining the visual comfort that apparent ridges provide.