Reverse engineering has accelerated product development in a range of industries. In automotive, it allows small shops to recreate worn vintage car parts or design missing parts around the existing components. Forensics, on the other hand, is leveraging reverse engineering to extract more insights from crime scenes. In heavy machinery, it’s used to create replacement parts for items that are no longer on the market.
No matter what area you’re working in, if you’re considering using reverse engineering, you need to understand the time and effort it involves. The key factor determining this are 3D models.
From a 3D scan to a 3D model
The technology underpinning all reverse engineering applications are 3D scanners, whose introduction made it easier to digitally capture the shapes of objects. However, the road from 3D scan data to a 3D model can be long and winding and requires intermediary software. CAD tools are essential for product design, but most CAD software can’t process scan data efficiently.
Geomagic Design X makes this process faster. With a scan processing kernel and a CAD kernel on top, it allows you to operate with scan data as native geometries, and extract the various types of models for manufacturing. Here’s an overview of the most common ones.
Point clouds
A point cloud is the simplest of 3D models used to represent physical objects. It’s an immense collection of data points – millions or billions – in space, with each point position defined by a set of Cartesian coordinates (X, Y, Z). A scanner combines the vertical and horizontal angles created by the laser to calculate the coordinate positioning. As the output of 3D scanning processes, point clouds are the basis for creating CAD models, for metrology and quality inspection.
Mesh
A polygon mesh is a relative of a point cloud. It’s created when software takes the measurements from point clouds together and connects the close ones into triangles, quadrilaterals (quads), or other simple convex polygons. The resulting collection of vertices, edges and faces together defines the shape of an object. A polygon mesh is easy to shade and render on the screen, and all software use it to display models.
CAD
A very distant model type that shares little with polygon mesh and point cloud data is CAD. Made of Non-Uniform Rational B-spline (NURB) surfaces, CAD models are a type of math used to represent 3D shapes within software. They’re easy to edit with formulas and dimensions. Because you can adjust parameters of different types to make a CAD model perfectly cylindrical or perfectly flat, we say they are parametrically driven. You’ll also hear them called “intelligent” meaning they contain a lot more information about the model than a point cloud or mesh.
CAD is the holy grail of 3D models in the engineering and design world. However, there is currently no way to just take a scan and convert it directly into a CAD. There are some assisting tools to help us get there and make it a lot easier, but most of the time CAD models are created by a designer or an engineer from scratch, inside the software. It requires effort and time, but it’s also widely used in manufacturing today.
SubD surfaces and voxels
SubD (subdivisional) surfaces are often used in animation, and in the AR and VR world for different types of representations. They’re a hybrid of a NURB surface that’s controlled with a polygon armature.
Voxels, which stands for volumetric pixels, are the 3D equivalent of pixels. They’re small cubes, tantamount to grains of sand. Voxels are often used in healthcare, to model prostheses or braces.
When to use which?
Not all 3D models are made equal. Getting from a point cloud to a CAD model is more laborious than getting to a polygonal mesh. But whether you want to put in the effort to get a CAD will depend entirely on what you need from your model. Do you require an exact duplicate, manufacturing flaws and all, or a perfect model to manufacture? Here are a few examples of their uses.
As-Built or exact duplicates are models created to get an exact representation. For example, if you want to make a brace or a device that goes around someone’s hand, you would need a representation of that person’s anatomy with very little deviation. Our tools allow you to wrap NURB surfaces directly to the 3D scan of an arm and create an almost exact representation, within microns of the original shape.
Design Intent – 3D models of industrial parts shouldn’t be an exact replica because fabricated parts come with deviations. There is a leeway between the design and the actual product called a manufacturing tolerance. While tolerances are accepted, they shouldn’t appear on a digital model.
Why Not Both – There are instances where you want most of the shape to be as-built, and some of the shape to correspond to design intent, like the topologically optimised bracket in the above picture. This organic shape was calculated with simulation software, which often uses mesh models, and was subsequently wrapped in CAD surfaces. However, to have the holes defined very precisely, we kept them as mesh.
Where does Design X come into your workflow?
Now that we understand the 3D model types, it’s time to get granular with the data processing workflow.
If you need a mesh model, the process is simple. With Design X, you can scan something and create the mesh by connecting the vertices from a point cloud. 3D printers and slicing software for additive manufacturing will read mesh files like STL, PLY, and OBJ. This simple workflow gives you an exact duplicate but with one important drawback – the features are difficult to edit.
However, when it comes to milling, a mesh model doesn’t cut it. If you need to run a CNC tool path across a shape, most Computer Aided Manufacturing (CAM) packages will yield a smoother toolpath on a NURB surface than on a mesh. Geomagic Design X can quickly wrap surfaces around a mesh to make toolpath programming more efficient.
Finally, often you want to scan an object to create a fully featured, editable CAD model. This requires the most effort, but it’s where Geomagic Design X has the biggest impact. It processes large scan data sets with millions of points faster than any other software, enabling you to quickly reverse-engineer physical parts from 3D scan-based data into digital parametric CAD models.
