Perception of Human Motion with Different Geometric Models

Abstract: Human figures have been animated using a variety of geometric models including stick figures, polygonal models, and NURBS-based models with muscles, flexible skin, or clothing. This paper reports on experimental results indicating that a viewer's perception of motion characteristics is affected by the geometric model used for rendering. Subjects were shown a series of paired motion sequences and asked if the two motions in each pair were the same or different. The motion sequences in each pair were rendered using the same geometric model. For the three types of motion variation tested, sensitivity scores indicate that subjects were better able to observe changes with the polygonal model than they were with the stick figure model.

Perception of Human Motion with Different Geometric Models

This paper provides an incisive examination of how varying geometric models impact the perception of human motion. The authors, Jessica K. Hodgins, James F. O'Brien, and Jack Tumblint, explore the significant differences in motion sensitivity when using polygonal models compared to stick figures. The research utilizes experimental data wherein subjects observed motion rendered through different geometric models.

Key Findings

The paper posits three hypotheses concerning viewer perception: that simpler representations could potentially allow more precise distinctions, that complex and accurate models might enhance detail recognition, and that model complexity may be largely irrelevant to human judgment. The study relies on controlled laboratory experiments where subjects were presented with pairs of motion sequences using different models—polygonal and stick figure. Results demonstrate that model type indeed affects the perception of motion changes. Specifically, subjects showed higher sensitivity to motion changes with polygonal models, evidenced by significant H-F sensitivity scores across tests (p-values ranging from <0.001 to <0.012 depending on the specific motion variation tested).

Implications

These findings have profound implications for the field of computer animation, where perceptual accuracy is paramount. Given that subtle nuances are more readily observed in polygonal models, animators and designers can harness this knowledge to craft more lifelike and emotionally resonant animations by opting for more complex models. Furthermore, these results suggest a potential avenue for standardization in the research and development of animated models, fostering consistency in evaluations of motion synthesis methodologies across studies.

Future Prospects

Future work could expand on these findings by investigating whether more abstract or complex models without human likeness serve particular purposes in other domains. Research might also focus on optimizing rendering techniques by emphasizing factors that influence motion perception, perhaps affecting both conscious and unconscious processing of visual information. Exploring psychophysical models to refine how animations are rendered could unlock more authentic representations of human motion, useful across various applications from entertainment to robotics.

The study's meticulous approach in understanding perceptual differences lays a solid groundwork for advancing techniques that capture and express human-like movements effectively, encouraging further investigations into how geometric complexities can affect judgments in dynamic scenarios.