Haptic feedback is becoming more and more common in the visual simulation of various critical tasks. To this end, the physical properties of the underlying material need to be measured or evaluated, and used to simulate the forces applied to the virtual tool that manipulates the surface or volume of the three-dimensional objects with which the user interacts.

Very often, the demand for visual quality of the three-dimensional objects in terms of surface smoothness is very high, resulting in highly tessellated meshes for visualization. Unfortunately, trying to perform the physical simulation of the surface or volume using a spring model or the finite element method on the highly tessellated version of the object can result in noninteractive rates. Alternatively, the mesh used for physical simulation can be of a lower resolution than the visualized one, and the selected level of detail may be determined adaptively in a global manner (for the whole mesh), or via local subdivision of the triangulated surface.

The authors present a usability evaluation of the various alternatives to full-resolution simulation, and measure their impact on the perceived quality of the haptic feedback. A simple test case is used in the paper, and various tests are performed using a group of different individuals, measuring their ability to detect a difference between the full-resolution underlying mesh and one of the alternatives, namely, fixed or adaptive global low-resolution mesh substitution and adaptive local subdivision.

The results of the presented case study are interesting because they provide a detailed credibility study of the various methods. The case study confirms that adaptive local subdivision provides a good alternative to full subdivision, and matches full-resolution physical simulation in terms of feedback quality. Unfortunately, due to the very limited set of test cases (one very simple surface), the reliability of the results may be questioned. Furthermore, although the authors discuss the limits of tactile detail discrimination, no tests are provided for the problematic case of rough or grainy but elastic or viscous surfaces, where very high local tessellation may be required.