Uncertainty is an intrinsic property of almost all data, regardless of the data being measured, simulated, or generated. It can significantly influence the results and reliability of subsequent analysis steps. Clearly communicating uncertainties is crucial for informed decision-making and understanding, especially in biomolecular data, where uncertainty is often difficult to infer. Uncertainty visualization (UV) is a powerful tool for this purpose. However, previously proposed uncertainty visualization (UV) methods lack sufficient empirical evaluation. We collected and categorized visualization methods for portraying positional uncertainty in biomolecular structures. We then organized the methods into metaphorical groups and extracted nine representatives: color, clouds, ensemble, hulls, sausages, contours, texture, waves, and noise. Furthermore, we assessed their strengths and weaknesses in a twofold approach: expert assessments with six domain experts and three perceptual evaluations involving 1,756 participants. Through the expert assessments, we aimed to highlight the advantages and limitations of the individual methods for the application domain and discussed areas for necessary improvements. Through the perceptual evaluation, we investigated whether the visualizations are intuitively associated with uncertainty and whether the directionality of the mapping is perceived as intended. We also assessed the accuracy of inferring uncertainty values from the visualizations. Based on our results, we judged the appropriateness of the metaphors for encoding uncertainty and suggested further areas for improvement.

Surface noise provides a geometric alternative to color for encoding scalar information on surfaces, yet its perceptual characteristics remain insufficiently understood. We present a systematic investigation of how variations in noise amplitude and frequency are perceived when applied to 3D surfaces. Across three online perceptual studies with 142 participants, we gathered similarity judgments for surface noise stimuli and modeled the resulting perceptual space using multi‐dimensional scaling (MDS). Our analysis shows that the amplitude–frequency perceptual space requires at least three dimensions for accurate reconstruction; however, the stimuli lie near a 2D manifold. Building on these findings, we derive locally perceptually uniform reparameterizations for both amplitude and frequency, improving the suitability of surface noise as a mapping for scalar data. These results provide perceptual guidance for the design of geometric encodings based on surface perturbation.