Selected work

Representative Papers

Shadowless Projection Mapping for Tabletop Workspaces with Synthetic Aperture Projector

Mar 12, 2026

This work addresses the challenge of shadow occlusion in conventional desktop projection mapping, which disrupts augmented reality experiences and multi-user interaction. The authors propose a synthetic aperture projection architecture based on a dense multi-projector array, incorporating offline blur compensation to mitigate resolution loss caused by subpixel misalignment—enabling high-quality, shadow-free projection with low latency without requiring real-time computation. For the first time, they establish a design framework centered on minimizing the “Sense of Projection” (SoP), integrating user perception studies to optimize system performance. This approach significantly reduces users’ awareness of the projected nature of the content, thereby enhancing both immersion and practical usability.

0 citationsRead paper

MAME: Multidimensional Adaptive Metamer Exploration with Human Perceptual Feedback

Mar 17, 2025

This work addresses the challenge of directly exploring the high-dimensional human perceptual metamer space—a space where stimuli are indistinguishable to human observers. We propose MAME, the first framework enabling online, adaptive, closed-loop search over >50-dimensional metamer spaces. MAME integrates hierarchical CNN feature modulation, generative image perturbation, online Bayesian parameter updating, and psychophysical two-alternative forced-choice discrimination tasks, with real-time human behavioral feedback guiding iterative image generation. Key contributions: (1) First direct empirical characterization of high-dimensional human metamer boundaries—bypassing reliance on model-based metamers and their indirect inference; (2) Discovery that human discrimination sensitivity to low-level feature metamers is significantly lower than to high-level ones—a finding unexplained by classical contrast-sensitivity models and posing a fundamental challenge to prevailing theories of perceptual invariance.

0 citationsRead paper

Recent publications

Latest Papers

Shadowless Projection Mapping for Tabletop Workspaces with Synthetic Aperture Projector

Mar 12, 2026

This work addresses the challenge of shadow occlusion in conventional desktop projection mapping, which disrupts augmented reality experiences and multi-user interaction. The authors propose a synthetic aperture projection architecture based on a dense multi-projector array, incorporating offline blur compensation to mitigate resolution loss caused by subpixel misalignment—enabling high-quality, shadow-free projection with low latency without requiring real-time computation. For the first time, they establish a design framework centered on minimizing the “Sense of Projection” (SoP), integrating user perception studies to optimize system performance. This approach significantly reduces users’ awareness of the projected nature of the content, thereby enhancing both immersion and practical usability.

0 citationsRead paper

MAME: Multidimensional Adaptive Metamer Exploration with Human Perceptual Feedback

Mar 17, 2025

This work addresses the challenge of directly exploring the high-dimensional human perceptual metamer space—a space where stimuli are indistinguishable to human observers. We propose MAME, the first framework enabling online, adaptive, closed-loop search over >50-dimensional metamer spaces. MAME integrates hierarchical CNN feature modulation, generative image perturbation, online Bayesian parameter updating, and psychophysical two-alternative forced-choice discrimination tasks, with real-time human behavioral feedback guiding iterative image generation. Key contributions: (1) First direct empirical characterization of high-dimensional human metamer boundaries—bypassing reliance on model-based metamers and their indirect inference; (2) Discovery that human discrimination sensitivity to low-level feature metamers is significantly lower than to high-level ones—a finding unexplained by classical contrast-sensitivity models and posing a fundamental challenge to prevailing theories of perceptual invariance.

0 citationsRead paper