Following our previous announcement—“InfiPoints Standard Package Now Includes BIM/CAD Export To Enhance Scan-to-BIM Workflows – InfiPoints May 2025 Update – 1”—we’re excited to unveil the second major enhancement in the InfiPoints May 2025 Update.
This release introduces two powerful new features designed to dramatically improve how engineers work with massive point cloud datasets: automatic segmentation and textured polygon generation. These capabilities make it faster than ever to extract functional components and generate simulation-ready models directly from your laser-scanned point cloud data.
From Billions of Points to Meaningful Segments—Automatically
Point clouds captured using 3D laser scanners can span entire facilities or construction sites and contain hundreds of millions of points. Each point holds precise XYZ coordinates and color information, enabling the digital replication of the real world with exceptional accuracy.
While a point cloud may appear to be just a mass of dots floating in space, humans can easily recognize real-world objects and structures within it—walls, equipment, pipes, and more. More than that, we naturally segment the scene in our minds: “This part is machinery,” “That section is structural.” We intuitively group points into meaningful areas based on shape and position.
At Elysium, we’ve taken this human ability and turned it into an intelligent algorithm. We are excited to introduce auto-segmentation, a new feature in the upcoming May 2025 Update version of InfiPoints. This technology enables InfiPoints to automatically divide massive point cloud data into logical and functional segments, streamlining the workflows for modeling, simulation, and layout planning.
Unlock the Full Potential of Point Cloud—Smarter Visualization, Faster Conversion
Point cloud segmentation enhances the utilization of 3D-laser-scanned point cloud data.
For example, in a digital twin of a manufacturing factory, isolating points for a specific piece of equipment and moving them to another layer allows you to visualize the space after the equipment removal simply by hiding that layer. Similarly, with a digital twin of a park, by isolating points for a tree, you can instantly preview how the landscape would look without the tree—helpful for planning and design.
Segmentation is also essential when converting point clouds into polygon models for simulation software. InfiPoints includes a feature that converts only the points registered in a specified layer into polygon data. This avoids processing unnecessary parts of the scan and greatly reduces conversion time. Instead of converting the entire point cloud, you can quickly generate only the 3D models needed for simulation—boosting system performance and engineering productivity.
Next-gen Point Cloud Segmentation
Since the product launch, InfiPoints has offered a feature to classify point cloud by layers and isolate points in a certain region to a separate layer by manually specifying the target point using clipping tools. While this gave users precise control over how the data was segmented, and has been a must-have feature for many customers, there has been a demand to reduce the time and effort required for this manual process.
The new auto-segmentation feature, included in the latest version of InfiPoints, directly addresses this feedback. It automatically identifies individual components within the point cloud—such as equipment, columns, and floor surfaces—and assigns them to separate layers without user intervention.
How It Works: Automatic Segmentation in Action
Organize Complex Point Clouds with a Single Click
To illustrate the power of the new automatic segmentation feature, let’s take a look at a real-world example—a large-scale incineration facility filled with piping, pumps, and various types of equipment.
Using the new functionality is remarkably simple.
With a single click of the [Auto-segment] button in the InfiPoints menu, the software automatically analyzes the point cloud, identifies distinct elements, and assigns them to separate layers. These segments are then color-coded and displayed for easy visual distinction.
For example, if you want to simulate what the site would look like after removing specific equipment, all you need to do is uncheck the corresponding layer. In seconds, you can visualize the post-removal environment—making planning and decision-making faster and more efficient.
From Scan to Simulation-Ready: Generate 3D Models with Textures
Next, let’s explore how InfiPoints converts point cloud data into 3D polygonal models using the example of a cylindrical piece of equipment.
Traditionally, creating a 3D model of an installed asset requires referencing photographs, handwritten measurements, or blueprints—starting from scratch. Even when point cloud data is available, it often serves only as a visual guide for manual modeling.
With the new workflow in InfiPoints, if the equipment has already been segmented into its own layer, generating a polygon model becomes as simple as clicking a button. The point cloud within that layer is automatically converted into a 3D polygon model.
Here’s an example of an automatically generated polygon model of the equipment:
Initially, the polygon model is displayed in a solid color. This is sufficient for virtual simulations, but there are situations where more visual realism or intuitive interpretation matters—especially when communicating with non-technical stakeholders.
For these scenarios, InfiPoints now includes a new texture mapping feature that applies color textures to the polygon model, derived from the original point cloud’s color data.
What looks like a high-density point cloud is, in fact, a lightweight polygon model with textures—ready for simulation and easy to handle in downstream applications.
These textured 3D models can also be used directly within InfiPoints’ built-in collision detection feature, supporting tasks like equipment layout planning or carry-in/out route study.
The Future of Elysium’s Proprietary Technology—Building toward the Future of Intelligent Point Cloud Processing
The new automatic segmentation feature in InfiPoints is powered by a proprietary algorithm developed by Elysium. While InfiPoints has long been recognized for its ability to detect planes and cylinders from point cloud data, this latest innovation marks a significant advancement—combining our deep expertise in 3D data processing with new ideas born from years of real-world experience.
The development effort spanned several years, with engineers fully immersed in the challenge—some even joking that “everything around them started to look like point clouds.” The result is a powerful tool that saves time and improves precision, but we recognize that there’s still room for improvement in both accuracy and usability.
Looking ahead, Elysium plans to further evolve these features by integrating AI-based processing. By combining our proven algorithms with artificial intelligence, we aim to unlock even more value from point cloud data—pushing the boundaries of what’s possible in digital engineering.
See Point Cloud Segmentation in Action
Watch the full hands-on demo video to see how engineers can apply auto-segmentation and polygon generation in real-world workflows: