Blog04 - Exploring Marker-Based and Markerless Augmented Reality

Introduction:

Augmented reality (AR) overlays digital information and objects onto the real world. In this post, we’ll explore two main approaches for developing AR apps - marker-based and markerless AR. We will explain the difference between the two and how our project uses both techniques.

Marker-Based AR:

Marker-based AR relies on recognizable images or QR codes called "markers" to trigger the appearance of virtual objects. The camera detects the marker, allowing the AR app to determine the marker's position and orientation. This allows the virtual content to track the marker as it moves.

To set up marker-based AR, you need to create and print a marker image. Then use an AR SDK like ARCore or ARKit to detect the marker and render 3D models over it. Defining the marker size and adding a border helps improve tracking accuracy. Overall, marker-based AR offers a straightforward way to anchor virtual objects in real-world spaces.

Markerless AR:

Markerless AR does not require special markers or images. Instead, it uses the device's camera and sensors to understand its environment. ARKit and ARCore use visual inertial odometry algorithms to achieve 6DOF tracking, allowing virtual objects to be pinned to surfaces more realistically.

The setup for markerless AR is enabled through the SDK capabilities. ARCore and ARKit perform plane detection to identify surfaces like tables and floors to place objects on. This allows for more flexible, markerless experiences. However, tracking can be less accurate without a fixed marker.

Why we used both for our project:

The markerless AR seemed as a better option in cases of layout setup on a physical table, as the size of the table can always vary. The marker based AR seemed more convenient in the case of displaying character sheets, as the type and amount of characters in the game is predefined. Having markers representing these characters seems more convenient as well as more easily understandable for the users.

Conclusion:

AR continues to evolve rapidly, with markerless tracking accuracy improving all the time. But for precise control over virtual object placement, marker-based AR still plays an important role. Knowing when to use each approach helped strengthen our AR development skills. I look forward to experimenting more as AR headsets and capabilities advance.