Deposition-Based 3D Generative Modeling: A Framework for Three-Dimensional Object Generation Through Gas-Phase Particle Deposition, Local Adsorption, and Branching Growth
Abstract
Three-dimensional generative modeling has advanced through approaches such as 3D diffusion models, neural radiance fields (NeRF), 3D Gaussian Splatting (3DGS), point-cloud generation, and voxel-based generation. Despite their differences in representation and optimization, these approaches generally construct objects by predicting, refining, or reconstructing spatial representations. This paper proposes Deposition-Based 3D Generative Modeling (D3GM), a conceptual framework that formulates three-dimensional object generation as a dynamic process of gas-phase particle transport, local surface adsorption, solid deposition, and branching growth. Inspired by physical deposition and frost formation, D3GM represents a target object as the outcome of successive deposition events rather than solely as a directly predicted geometric structure. A three-dimensional environment contains mobile candidate particles, spatially varying deposition conditions, local attraction fields, and evolving solid structures. Particles move through the environment, interact with existing surfaces, and become incorporated into the developing object according to learned or prescribed deposition rules. Local growth conditions determine whether deposition thickens existing structures, initiates new branches, reinforces selected regions, or terminates growth. A learned controller can adapt particle transport, adsorption probability, branching behavior, and deposition intensity to satisfy target geometry and appearance constraints. The resulting structure can be represented as a point cloud, voxel field, mesh, implicit surface, or a collection of three-dimensional Gaussian primitives. This paper defines the conceptual architecture, mathematical formulation, generation algorithm, training strategy, evaluation protocol, and principal research challenges of D3GM. The framework is proposed as a research direction, not as an experimentally validated method. Its potential contribution is the explicit use of deposition events and local growth dynamics as the primary generative mechanism for three-dimensional structures.
Related
- Layerwise Deposition-Field Generative Modeling: A Process-Conditioned Framework for Three-Dimensional Generation Through Thin-Film Deposition and Surface Growth
- Fluvial Erosion–Deposition Generative Model (FED-GM): A Process-Driven Framework for Three-Dimensional Terrain Generation Through Sediment Transport and Geomorphic Feedback
- Granular Flow and Deposition Generative Modeling: A Physics-Driven Framework for Three-Dimensional Shape Generation Through Discrete Particle Dynamic
- Reaction-Driven Three-Dimensional Generative Computing: A Framework for Constructing 3D Objects Through Local Chemical Transformation, Deposition, and Structural Growth
- Collective Biological Construction: A Multi-Unit Generative Framework for Three-Dimensional Shape Formation Through Local Deposition and Collective Structural Evolution
BibTeX
@misc{zhang2026depositionbased,
title = {{Deposition-Based 3D Generative Modeling: A Framework for Three-Dimensional Object Generation Through Gas-Phase Particle Deposition, Local Adsorption, and Branching Growth}},
author = {Jincheng Zhang},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23635850},
url = {https://zenodo.org/records/23635850},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Deposition-Based 3D Generative Modeling: A Framework for Three-Dimensional Object Generation Through Gas-Phase Particle Deposition, Local Adsorption, and Branching Growth
AU - Jincheng Zhang
PY - 2026
DA - 2026/10/25
PB - Zenodo
DO - 10.5281/zenodo.23635850
UR - https://zenodo.org/records/23635850
AB - Three-dimensional generative modeling has advanced through approaches such as 3D diffusion models, neural radiance fields (NeRF), 3D Gaussian Splatting (3DGS), point-cloud generation, and voxel-based generation. Despite their differences in representation and optimization, these approaches generally construct objects by predicting, refining, or reconstructing spatial representations. This paper proposes Deposition-Based 3D Generative Modeling (D3GM), a conceptual framework that formulates three-dimensional object generation as a dynamic process of gas-phase particle transport, local surface adsorption, solid deposition, and branching growth. Inspired by physical deposition and frost formation, D3GM represents a target object as the outcome of successive deposition events rather than solely as a directly predicted geometric structure. A three-dimensional environment contains mobile candidate particles, spatially varying deposition conditions, local attraction fields, and evolving solid structures. Particles move through the environment, interact with existing surfaces, and become incorporated into the developing object according to learned or prescribed deposition rules. Local growth conditions determine whether deposition thickens existing structures, initiates new branches, reinforces selected regions, or terminates growth. A learned controller can adapt particle transport, adsorption probability, branching behavior, and deposition intensity to satisfy target geometry and appearance constraints. The resulting structure can be represented as a point cloud, voxel field, mesh, implicit surface, or a collection of three-dimensional Gaussian primitives. This paper defines the conceptual architecture, mathematical formulation, generation algorithm, training strategy, evaluation protocol, and principal research challenges of D3GM. The framework is proposed as a research direction, not as an experimentally validated method. Its potential contribution is the explicit use of deposition events and local growth dynamics as the primary generative mechanism for three-dimensional structures.
ER -