In computer-aided design (CAD), hybrid modelling is a method of using different modeling techniques (with different rulesets) in the same workflow. Four types of modeling data can be represented in a hybrid modeling approach:
BRep (boundary representation) model data
Polygonal model data
Point cloud data (3d scanned objects)
Voxel data (3D pixels for volumetric imaging)
Most traditional 3D software packages specialize in working with one type of modeling data, and can be missing functionality to interpret other formats. Similarly, some CAD file formats are limited to a specific type of modeling data.
Before we discuss why this can create difficulties -- we need to understand each of different modeling formats.
Main benefits and disadvantages of BRep Modeling
BRep modeling, or boundary representation modeling, is the most common type of modeling in CAD applications.
In short, BRep is a mathematically precise representation of a 3D object that defines the geometric boundaries between solid and non-solid geometries.
The shape and contours of a BRep object are not built out of reducible objects like polygons or vertices. Instead, a BRep object is defined by the mathematical relationships between its surfaces.
Take a look at the image above, and focus on the bumper.
This bumper is not defined by any smaller components like polygons. Instead, it exists as its own discrete object. Its shape is defined by the position and curves of its surfaces in 3D space, relative to the XYZ axes.
There’s a mathematical formula describing the surface’s rotation and its slight S-like curve on one the Y-Axis. There’s another mathematical formula that describes the elbow-like bump at the bottom of the bumper on the Z axis. And when we combine a myriad of precise, mathematical formulas to describe all of its surfaces across all of the axes -- we get a BRep object.
BRep objects are mathematically precise, and allow designers and engineers to build “perfect” representations of their designs
Unlike other forms of modeling, BRep allows you to “zoom in” without losing “resolution.” A BRep curve will remain curved at every level of magnification
BRep’s mathematical precision makes it ideal for manufacturing applications
BRep file formats are inherently heavy, and store a lot of meta-data that takes up disk space
When an object needs to be visualized, rendered, or animated -- BRep takes too much processing power
Organic / natural objects are hard to recreate with the precise mathematical formulas of BRep
Overall, the features of BRep make it an ideal format for engineers and designers of manufactured goods -- but it’s severely limited for visualization or rendering purposes.
Main benefits and disadvantages of Polygonal Modeling
Polygonal (or polyhedral) modeling is the most common type of modeling for video games and animation studios.
This type of modeling builds 3D objects out of smaller components called “tris” (triangles) or “polys” (polygons).
Each poly or tri is a completely flat shape that is defined by the position of its vertices (or points) and its connecting edges.
Complicated models of any shape can be built completely out of tris or polys. As designers require more “high-fidelity” designs (smoother looking surfaces, more details, etc.), they can increase the number of polygons in their models.
Because modern computers are optimized to handle polygons, polygonal models are the easiest to render and visualize
Polygonal modeling allows designers to create more unique / organic looking designs (humans, animals, etc.)
Because polygonal models are created from smaller components (polys and tris), they can be deformed and animated more naturally
Building a model out of polygons is imprecise, and introduces human error
Polygonal models don’t hold up on all resolutions
Polygonal modeling is very time consuming, especially for intricate designs
The features of polygonal modeling make it ideal for use cases where precision isn’t vital, and where visualization is more important. This is why animation studios and video game studios use polygonal models almost exclusively.
Main benefits and disadvantages of Point-Cloud Modeling
Point-cloud modeling is typically used in the process of 3D scanning objects.
Rather than defining surfaces through mathematical formulas, or building them out of rudimentary shapes like triangles -- point cloud modeling creates a representation of a 3D object with densely placed vertices -- or “points” -- along its surface.
With a high enough resolution and point-density, point-cloud models can accurately represent the features of virtually any 3D object. In fact, point-cloud 3D scanning has been used to create 3D representations of highly complex objects like human faces.
Beyond scanning objects, point cloud data has a wealth of applications in simulation.
Point-clouds can be used to represent solid objects in a Finite Element Analysis context -- boiling down mathematically complex CAD surfaces into a relatively finite number of points. This allows engineers and scientists to simulate objects under stress, simulate deformation, etc.
Point cloud models can accurately represent relatively complex objects with a finite number of elements (points)
Point cloud models can be the quickest to create -- assuming you have access to 3D scanning technology or CAD conversion software (we don’t recommend building one from scratch)
Point-cloud models lack the precision of BRep, and cannot create mathematically perfect curves
Point-cloud data does not include information on surfaces, and therefore cannot be used natively for rendering or manufacturing
Point-cloud data is very difficult to translate into an accurate BRep or polygonal model
Ultimately, point-cloud modeling alone doesn’t have a lot of applications. 3D Scanned objects need to be converted to another type of model to be truly useful. In conjunction with other modeling types, however, point-cloud can be incredibly useful. It can be used with BRep to simulate stresses on engineered models. It can be used with polygonal modeling to create complicated 3D models of scanned objects.
Main benefits and disadvantages of Voxel Modeling
Voxels are to 3D what pixels are to 2D.
Firstly -- let’s examine what pixels actually are. Everything you see on your computer screen is made up of very small squares called “pixels.”
If you’re on a fairly new computer, you most likely can’t make any pixels out, because your display is “high-resolution.” These pixels are so tiny, and there are so many of them, that you can’t actually see them. Instead, you see words, pictures, and symbols that appear smooth.
Voxels are essentially 3D pixels, but instead of being squares, they are perfect cubes.
In theory, voxels are the perfect modeling technique for replicating reality.
After-all, our world is made of something akin to voxels (but they are much smaller, and we call them “sub-atomic particles”). If you have a high enough density (or “resolution”) and the proper rendering techniques, you can use voxels to replicate real-world objects that would be impossible to differentiate from the real thing -- in appearance and behavior.
In practice, however, there are no mainstream methods for easily building out complex, high-resolution objects using voxels. There are some promising attempts (like Atomontage, referenced above) -- but all the other modeling methods listed above are still quicker and easier for truly complex designs.
Furthermore, modern computers are simply not optimized to handle rendering voxels. Most of our hardware is meant to render polygons -- so high-resolution voxel objects can take a serious toll on current hardware.
Despite not being ready for mainstream adoption, voxel modeling has a few very specific use-cases today.
Currently, voxels are used in many scientific disciplines to quickly determine volumetric data. For example, in Voxel-based morphometry, researchers can compare the differences in concentration of brain tissue using voxels. Geologists often use voxel modeling techniques to model geological features like terrain and elevation. More broadly, scientists can use voxel-based modeling to visualize and measure the volume of anything from fluids to green spaces in urban centers. Voxels are also useful in simulation techniques that require modeling of individual particles, as is the case of smart material simulation.
And that’s where voxels truly shine.
Because they can represent complex objects in reducible, discrete units (like particles), they can be incredibly powerful for simulating real world behaviour of complex objects.
Voxels are a more “accurate” 3D building block than any other modeling type, as they mimic particles
Voxels unlock new simulation techniques that would be impossible with other modeling methods
Voxels are the quickest way to quickly model and visualize volumetric data (especially in natural or organic formations)
Without using prohibitively expensive techniques like 3D scanning, it is much harder to build complex objects using voxels
Voxel-modeling lacks the mathematical precision of BRep modeling
Current computer hardware is optimized for rendering polygons, and we don’t have specialized hardware to efficiently render high-resolution voxels
The Process of Hybrid Modeling
Simply put, hybrid modeling combines the benefits of BRep, polygonal, point-cloud, and voxel-based modeling -- the four main types of 3D modeling -- into one workflow.
Traditionally, you would need 4 different software suites to handle all of these model types. You would create a model using one technique (like BRep), and use a CAD conversion software to translate the BRep data into another format. Once translated, you would open your file up in another software to make changes.
This is incredibly cumbersome for a number of reasons.
What if you engineered a new product using BRep, and needed to quickly visualize or render it for promotional purposes? And what if, once visualized, your creative team had recommended edits to your BRep?
What if you wanted to run a simulation on your BRep model using point-cloud data, and then apply the insights of that simulation to make changes to the original BRep?
You’d need to jump from software to software, converting your files back and forth, manually fixing them at every step of the process. This is not only time-consuming, but creates ample opportunity for human error. And in industries like CAM (computer aided manufacturing) and CAE (computer aided engineering) -- this can be quite costly.
This is where “hybrid modeling” comes in.
Software development kits like Spatial’s CGM allow 3D developers to build hybrid-modeling features natively into their toolset. This allows designers and engineers to seamlessly hop back-and-forth between modeling types without losing any data in the process. It allows engineers to run any simulation on their BRep files, and take those insights back to automatically tweak their models.
This “back and forth” between environments and modeling techniques is the process of hybrid modeling.
Powerful hybrid modeling software means you can shorten the process of model preparation and fill in the gaps that are typical of 3D models -- like simulating the behaviour of engineered models on a particle-basis.
Hybrid software can recognize these gaps and allow the user to make adjustments along the way.
Hybrid Modeling as the Ultimate Solution
Good news, designers and engineers - it’s no longer a hard and fast decision about what type of modeling is right for you.
Engineers, architects and designers in different industries can seamlessly migrate between BRep, polygons, point-clouds, and voxels -- to leverage the benefits of the various techniques and minimize the downsides.
Now, who said you couldn’t “have your cake and eat it too?”