For years, in-situ monitoring has been positioned as the answer to real-time quality control in additive manufacturing (AM). The promise? Immediate insights into the build process, fewer post-inspection costs, and greater confidence in part quality. But the reality has been far more complicated.
Many early in-situ monitoring systems failed to deliver meaningful, usable data. They relied on visual monitoring or subjective thermal cameras – methods that lacked repeatability and objective measurement, leading manufacturers to mistrust the technology altogether. The misconception that any in-situ monitoring is beneficial has allowed inadequate solutions to persist, leaving users frustrated when the data proved unreliable.
The Pitfalls of AI-Driven Monitoring
Artificial intelligence is a powerful tool, but when applied using subjective data sources, can lead to misleading and dangerous conclusions. Many companies have turned to AI-driven in-situ monitoring, training models on visual data to predict part quality. The problem? These models are only as good as the data they are trained on, and in many cases, that data is subjective and not trained on the machine you are using.
Unlike manufacturing processes that rely on unit-based, quantifiable inspection methods, AI-driven monitoring using visual images introduces a layer of uncertainty. A model trained on past builds might detect anomalies, but it cannot explain why they occur or ensure repeatability across different machines, materials, or production environments.
Figure 1- Fringe Inspection measures every layer of the build, outputting a heightmap (right) compared to a subjective visual image of the build (left), courtesy of Phase3D
This approach leaves manufacturers trusting a black box—one that lacks the objectivity, precision, and traceability required for quality-critical industries like aerospace, medical devices, and defence. Without a clear understanding of the root cause of failures, manufacturers are left reacting to problems instead of preventing them.
Phase3D: A Different Approach to In-Situ Inspection
At Phase3D, we believe the industry deserves more than black-box AI solutions. Fringe Inspection™ provides objective, quantifiable, and repeatable data, measuring every layer of the build to detect anomalies which lead to build failures or part failure during inspection.
Unlike traditional monitoring systems that rely on visual images, heat signatures, or indirect indicators, Fringe Inspection™ delivers unit-based measurements, creating a digital twin of the 3D profile of every layer to detect variations in powder distribution like recoater streaks, part protrusions, and other build-failing anomalies.
Figure 2 – Fringe Operator is a visualization and analysis tool for customers to identify anomalies that cause build failures in real time, courtesy of Phase3D
Phase3D’s impact on the industry means manufacturers no longer have to guess whether a part is good or bad—they can measure exactly what is happening in the build. By identifying the root cause of defects, Fringe Inspection™ enables users to make informed decisions about their builds rather than relying on black box AI-generated predictions.
The Value of Early Detection
The ability to detect and quantify anomalies at the layer level presents a significant opportunity for manufacturers. Instead of relying on expensive post-inspection techniques—such as CT scanning, X-ray, or destructive testing—to validate parts, Fringe Inspection™ enables manufacturers to catch issues early and reduce the number of defective builds.
This does not eliminate the need for post-inspection—but it makes it far more economical. Instead of spending resources inspecting parts that were already defective during the build, manufacturers can now focus post-inspection on certifying high-quality components.
The impact is twofold:
- Improve delivery timelines—Manufacturers can prevent defective parts from being completed, increasing printing utility.
- Decreased costs—By catching bad parts early, companies can allocate post-inspection resources more efficiently.
By using Fringe Inspection™, manufacturers gain a proactive approach to quality control—one that improves efficiency, reduces costs, and increases confidence in AM as a production-ready technology.
Figure 3 – Fringe Qualification tracks build performance across a fleet of printers to track waste for every machine and material, courtesy of Phase3D
The Future of AM Inspection
For AM to scale into a mainstream manufacturing process, in-situ inspection must move beyond passive monitoring and into actionable, data-driven process control. The industry cannot afford to cut corners by relying on unvalidated AI predictions. Instead, manufacturers need reliable, transparent measurement technologies that provide repeatable, traceable, and trusted data.
Fringe Inspection™ is already being adopted by leaders in aerospace, defense, and research organizations who require high-confidence inspection data. By providing layer-by-layer analysis, manufacturers gain real-time insights that drive better decision-making, reduce risk, and accelerate AM’s adoption as a scalable, production-ready process.
We will be presenting more on Fringe Inspection™ and more at Additive Manufacturing Strategies (AMS) 2025, where industry leaders come together to explore the future of AM technologies.
Learn more about Fringe Inspection™ and Phase3D at www.phase-3d.com.