PoreDiT: A Scalable Generative Model for Large-Scale Digital Rock Reconstruction

TL;DR

PoreDiT is a scalable 3D generative model using Swin Transformers for large-scale digital rock reconstruction, enabling efficient, high-fidelity pore structure generation at gigavoxel scales on standard hardware.

Contribution

The paper introduces PoreDiT, a novel 3D Swin Transformer-based generative model that significantly improves scalability and efficiency in digital rock reconstruction.

Findings

Able to generate $1024^3$ voxel digital rocks on consumer hardware.

Preserves key topological features like porosity and permeability.

Achieves physical fidelity comparable to state-of-the-art methods.

Abstract

This manuscript presents PoreDiT, a novel generative model designed for high-efficiency digital rock reconstruction at gigavoxel scales. Addressing the significant challenges in digital rock physics (DRP), particularly the trade-off between resolution and field-of-view (FOV), and the computational bottlenecks associated with traditional deep learning architectures, PoreDiT leverages a three-dimensional (3D) Swin Transformer to break through these limitations. By directly predicting the binary probability field of pore spaces instead of grayscale intensities, the model preserves key topological features critical for pore-scale fluid flow and transport simulations. This approach enhances computational efficiency, enabling the generation of ultra-large-scale ($1024^3$ voxels) digital rock samples on consumer-grade hardware. Furthermore, PoreDiT achieves physical fidelity comparable to previous state-of-the-art methods, including accurate porosity, pore-scale permeability, and Euler characteristics. The model's ability to scale efficiently opens new avenues for large-domain hydrodynamic simulations and provides practical solutions for researchers in pore-scale fluid mechanics, reservoir characterization, and carbon sequestration.