P3DMM@ICME2026

Summary

Reliable 3D modelling is a foundational capability for many multimedia applications, yet achieving metrically accurate and physically meaningful 3D representations in real-world environments remains challenging. Variations in illumination, material properties, motion, sensor configurations, and environmental conditions often undermine the robustness and interpretability of purely data-driven approaches. This workshop focuses on advancing physically informed and physically interpretable 3D modelling, learning, and perception methods that explicitly incorporate physical principles to improve reliability, consistency, and trustworthiness across multimedia scenarios. This workshop aims to provide a unified forum for discussing how such physical principles can be systematically embedded into modern learning frameworks, including neural fields, radiance models, and multimodal foundation models.

In addition to physics-guided methods, the workshop welcomes contributions that integrate physical priors with diverse multimedia and sensing modalities, such as RGB-D, multi-view and video data, IMU and robotic kinematics, force and tactile sensing, acoustic measurements, and spectral or hyperspectral imaging. Particular emphasis is placed on methods that enhance physical consistency, interpretability, and measurement fidelity, enabling reliable 3D modelling for applications such as digital twins, intelligent manufacturing, robotics, biomedical imaging, and computational multimedia systems.

Call for Papers

We invite original submissions that address challenges and advances across the full spectrum of Physical Principles 3D Modelling. Topics of interest include, but are not limited to:

Learning-based 3D modelling with physical principles
Physics-coherent neural fields and radiance models
Shape, lighting and material decomposition with physical consistency
Modelling contact, collision and rigid/deformable body behaviour
Data-driven methods enriched by physical cues
Reliable 3D modelling in complex multimedia environments
Physical cues for digital twins and manufacturing
Multimedia applications requiring physically interpretable 3D models
Datasets, metrics and evaluations for physics-informed 3D modelling

Submission Guidance: Submit via CMT

Download CFP (PDF): Click here to download

Important Registration Note: All accepted papers need to be covered by a full registration.

Keynote Speakers

Prof. Junhui Hou

(CityU, HKSAR)

Prof. Huiyu Zhou

(University of Leicester, UK)

Schedule

Note: The schedule is for reference only and is subject to change.

Time	Event
9:30 - 9:40	Welcome & Session Introduction
9:40 - 10:10	Keynote 1: Prof. Junhui Hou (30 min)
10:10 - 10:40	Keynote 2: Prof. Huiyu Zhou (30 min)
10:40 - 10:50	Coffee Break
10:50 - 11:40	Oral Session (5 contributed talks, 8+2 min each)
10:40 - 12:00	Poster Session (10 posters)
12:00 - 12:10	Awards & Closing (Best Paper, etc.)