GeoDict Quantify brings AI-powered 3D image analysis to additive manufacturing

Math2Market GmbH launched GeoDict Quantify on September 29, 2026, a software package for processing and quantitatively analyzing data from computed microtomography, FIB-SEM, and other high-resolution 3D imaging methods. The package was unveiled during the plenary session of the GeoDict Innovation Conference 2026 in Frankfurt am Main.

Quantify combines data preparation, AI-assisted segmentation, quantitative characterization, reporting, and automation. The package reaches customers through selected CT and FIB-SEM system manufacturers, typically bundled with a new imaging system.

The manufacturer itself stresses that the tool was not developed exclusively for additive manufacturing, while acknowledging the importance of tomography to the sector. In metal 3D printing, the usability of a part is often determined long after printing is complete, on the CT scanner, where an image of its internal structure still has to be converted into numbers that can withstand scrutiny from customers. More interesting than the feature list itself, however, is what the package deliberately leaves out.

Built on GeoDict

GeoDict has been under development since 2001, while Math2Market itself was founded in 2011 as a spin-off from Fraunhofer ITWM in Kaiserslautern. Quantify is therefore a focused subset of a mature platform, something the company openly acknowledges by describing it as a way of making established capabilities available to a broader range of users.

From Raw Data to Segmentation

Data preparation includes removing ring artifacts from CT images and denoising them using the non-local means method. For FIB-SEM, or focused ion beam scanning electron microscopy, the workflow also includes alignment of successive slices and correction of the curtaining effect.

Segmentation can be performed using thresholding or AI-based methods, including the training of full 3D U-Net models, neural networks commonly used for image segmentation and image-quality enhancement.

Additive Manufacturing Use Cases

The applications most closely related to additive manufacturing are presented through two examples. The first involves an additively manufactured sample produced in the laboratory of Dr. Mehrshad Mehrpouy at the University of Twente.

An analysis of wall thickness shows a dominant value of approximately 0.9 mm, along with secondary peaks at 1.2 and 1.4 mm corresponding to internal and external wall connections.

The second example is a 10 mm metal cube, also produced by additive manufacturing. The software identified pores measuring up to 0.277 mm in diameter and characterized them through distributions of size, elongation, and sphericity.

Reproducibility Over AI Hype

Artificial intelligence has made its way into the headline of the announcement, as it seemingly has into every software launch these days. In additive manufacturing, however, greater significance may lie in a feature described much more quietly.

Every processing and analysis step is automatically saved as a recipe. The workflow can then be rerun on subsequent samples, by different users and across different projects. This is complemented by macros, Python scripting, and standardized reports.

Anton du Plessis, head of image processing and analysis at Math2Market, points out that a high-resolution image is only the starting point. Its value emerges when the result can be reproduced, reported, and compared. In an industry where part qualification depends on repeatability, that statement carries more weight than another mention of U-Net.

A Deliberately Limited Package

The most revealing part of the announcement concerns a limitation. Quantify is a fixed package of ten GeoDict modules intended exclusively for image analysis, and it cannot be directly expanded with simulation modules.

The company maintains this boundary carefully. Its documentation even notes that geodesic tortuosity is a geometric descriptor and should not be confused with numerically calculated transport properties.

Users seeking digital material modeling or multiphysics simulations must move to a standard GeoDict configuration. Math2Market CEO Andreas Wiegmann describes this as a clear development path and makes no secret of the company’s long-term ambition to make GeoDict a standard choice for 3D image analysis and material characterization.

A Focused Route Into the GeoDict Ecosystem

This is a classic sales funnel, described without pretending to be anything else, which happens less often in press releases than it probably should.

The distribution strategy is equally deliberate. Among the OEM partners for the GeoDict platform, Math2Market lists Carl Zeiss, Rigaku, ProCon X-Ray, and Bruker. The software is intended to reach laboratories together with new imaging systems, at the point when a major purchasing decision is already being made.

Price and Availability

The obvious gap is the price. The company describes the package as cost-effective and competitively priced, but does not disclose an actual figure.

Organizations that already own a micro-CT or FIB-SEM system cannot purchase Quantify as a standalone package in its current form. Instead, Math2Market will offer them a demonstration using representative data and the same functionality as part of a standard GeoDict configuration.