AR has penetrated just about every industry in 2020, and if you have decided to take your business to the next level by offering augmented reality (AR) experiences to consumers, chances are you may have been or will be stumped by the concept of “AR-ready 3D models”. What does it mean, is there a list of criteria for constructing good AR 3D models, and where or how can I get them?
By the end of this article, you will have a clearer understanding of how 3D models are built (for AR use) and be able to assess if your existing 3D models are AR-ready based on the following factors:
In the least amount of technical jargon possible, here are some of the things to take into account when it comes to making AR-ready models.
3D modelling serves vast and varied purposes, such as gaming, movies, architecture, and engineering, and with that comes different sets of requirements for what makes a “good” 3D model. Having an AR-ready model simply means that the models are built and optimized for use with AR technology, utilizing the right tools and assets in supported formats.
At this point, you can probably guess as much — in order to know what you need, you have to understand what you are trying to achieve. It sounds rudimentary, but it has a direct impact on the way that your AR experiences are produced.
For instance, brands and retailers who want to provide a contactless way for customers to interact with products and visualize the way they look in a real environment are trying to create digital twins of their products, so life-like textures and movements will be prioritized. The AR content will then be deployed via WebAR for convenient access.
On the other hand, businesses seeking new ways to excite and engage customers will be more interested in games and storytelling. The AR content needs to be highly interactive and immersive, and may be better off hosted on an app.
Building a solid foundation with good topology and low polygon count (“low poly”) comes first.
Polygon: Straight-sided shape with three or more sides. An individual polygon is also commonly referred to as a "face". 3D models are formed by many polygons connecting and forming a "polygon mesh".
Edge: Each straight side of a polygon is an edge.
Vertex: A point at which two or more edges meet. More often used in its plural form, "vertices".
A model’s topology depends on the way polygons are arranged so that they form vertices and edges in places that make sense for the structure of the model. Furthermore, clean topology is especially crucial if your goal is to animate the model, as vertices have to be organized in a way that allows the object to bend and deform smoothly and realistically. (No overlapping polygons to avoid errors in the appearance of the model.)
The optimal poly count depends on the intended use of the 3D model — for example, models for WebAR should keep to a poly count of not more than 100,000. In any context, poly count that is way too low will affect the visual quality of the model; too high and it affects performance and hogs up memory space. As both are just as important, the key is to work with skilled 3D artists as they know best how to strike a balance between both (as well as how to compensate for lower poly count) based on the model’s final application.
When creating 3D models to mimic the experience of in-store shopping, the models need to look and move as realistically as possible. This is where UV unwrapping and texturing come into play.
The most straightforward way to understand UV unwrapping is to imagine cutting along the edges of a cube so that you can flatly lay out all the faces; this flat representation of the object is its UV map. How a model is unwrapped determines how well you can “texture” it.
Textures are 2D images applied onto a UV map that give the object its colors, materials finishes, etc.
Neat unwrapping along with high-resolution and well-defined textures are especially important for materials like wood, with its grains and knots, or glossy surfaces in the way it reflects light. It is with these minute details that 3D models are brought to life. In fact, good textures can help to bring out the finer details of a model that may have been “missed out” in its topology in a bid to keep poly count low.
Let’s take a look at some examples of bad and good 3D models, based on what you have learnt so far.
Other phases that are also essential for animated models are rigging and skinning which affects movement. For a slightly deeper understanding, take a look at these resources.
There are several methods that 3D artists use to create models, depending on the materials available to work with and the desired outcome.
Modelling from scratch: This is exactly what it sounds like — building the models from ground up based on images and videos. It can be a tedious process with no “shortcuts'', but it allows for greater control over the entire modelling process.
Modifying existing models: 3D artists can tap into CAD libraries, where a vast source of 3D models are available (in various formats, such as .dwg, .blend, and .skp) and can be used as “templates”. Brands and retailers can also provide their current 3D models for artists to modify — this also means that specifications of the models will be highly accurate as it follows the particular product’s technical drawing. Depending on the usability of the existing models and number of details to adapt, this method could save time or use up more time to perfect the models to AR requirements.
Photogrammetry: This is the process of using photographic images to map objects and obtain measurements such as height, width, depth, angles, and so on, which are then used to create 3D renderings. This means that the models will be a highly accurate replica of the photographed object, but it typically requires a lot of modification to ensure that the models are AR-ready.
Omniaz’s 3D artists work with a combination of any or all of the methods on a case-by-case basis, first and foremost determining the usability of a client’s existing models (if any). This versatile approach allows for a tailored solution to achieve each client’s needs, no matter their level of familiarity with AR or the resources they have on hand.
Different tools are also used for different stages of the modeling process. Blender, a free (yet comprehensive) software, is frequently used for building the model itself and UV unwrapping. Many 3D artists also use Autodesk Maya or “Maya” for short, and 3ds Max. To add texture, Substance Painter and 3D-Coat are some of the widely used applications.
As with everything in the digital space, the right file formats are required for things to run smoothly. Typically, GLB and USDZ formats are used, as those are the standards set by Android and iOS respectively. FBX files are also common.
File size limitations are also in place for optimum performance — ensuring that the AR runs smoothly with reasonable loading time. While it is impressive to have AR models that look and behave as much as the real products as possible, getting carried away with animations can lead to a heavier file, which could lead to bad user experience for lower-end or older smartphones. This, again, is where it is important to work alongside 3D artists who can make informed decisions.
As a rule of thumb, keeping the file under 5mb is good.
And there you have it, the makings of a good AR-ready 3D model. Once ready, you can integrate your models onto your e-commerce site with a simple plug-in, use them in online ads on social media and Google, or even in physical stores (triggered by QR code). The endless capabilities of AR can be harnessed in many ways to elevate your business; it all goes back to the goals you are trying to accomplish and working with 3D artists and an AR solution provider that can guide you along the way.
Check out the video below to witness the process of modeling a TV console from start to finish.