Optimizing Model Target Tracking

Model Targets can be configured with a motion hint and model type that significantly improve the tracking performance in different use cases. Specify if you expect your Model Target to be in motion or stationary. Set the tracking mode to fit the physical object of your Model Target to improve tracking stability and improved detection on reflective or scanned objects.

For a general introduction to motion hint and setting the hint on other Vuforia object targets please see Motion hints.

Use the Model Target Generator (MTG) to set the default motion hint and model type. The hint and tracking mode can later be changed in Unity and through the Model Target Observer.

Overview Chart

The table below lets you get a quick overview of the effects the combinations of motion hints and tracking modes can be combined to. See the Model Target Motion Hint and Model Target Tracking Mode for more details.


Tracking Modes

Has moderate power consumption.

Provides good results with most objects.

Optimized for large reflective objects with poor features.


Motion Hints

Object is not move during tracking*.

Implements power saving mode. Tracking happens at a lower framerate.

Uses AI-powered tracking to significantly improves tracking for reflective objects when they are not moving. Not available on all devices, e.g., HL will fall back to below behavior.

Object can be moved during tracking.

Standard mode. Moderate power consumption with stable tracking for most objects.

Improved tracking for challenging, reflective objects at a higher performance cost.
Allows for limited object motion.

Object is in constant motion, e.g., for handheld interactions.

Only tracked while in the camera view. Only reports the latest position. Jitter might be visible if the object is not moving or if it is occluded.

* Device Tracking is required to get the behavior described for STATIC.

Model Target Motion Hint

The motion hint can be set to three values depending on the user scenario.

A Model Target can be set to:

  • STATIC that is ideal for tracking stationary Model Targets.
  • ADAPTIVE which allows for limited motion of Model Targets.
  • DYNAMIC that tracks a Model Target in constant motion.

STATIC Model Targets

STATIC should be used with the Device Tracking enabled for the best performance.

STATIC works best with Model Targets that are stationary, such as cars and large machinery. If you expect the object to be moved during tracking, select the ADAPTIVE motion hint as moving an object that is set to STATIC will likely break tracking, forcing the user to wait for the object to be recognized again.

STATIC is especially beneficial in combination with the CAR mode. Using this hint with the Tracking Mode for cars will enable an AI powered tracking technology that greatly improves tracking quality and robustness as long as the car is not moving.

The STATIC AI tracking technology is supported on most iOS devices and on the most recent Android Samsung flagship devices:

  • iPhone 8 and above.
  • iPad Pro (2nd gen) and above (i.e iPad 7,1 and above).
  • Samsung Galaxy S series : S10 and above.
  • Samsung Galaxy Note series : Note10 and above.
  • Google Pixel 6 and above.

On all other devices, tracking will automatically fall back to the same technology and performance as seen with ADAPTIVE in CAR mode.

With other model tracking modes than CAR, the STATIC motion hint will save battery power and enable longer AR sessions.

ADAPTIVE Model Targets

ADAPTIVE will automatically select the optimal tracking behavior, depending on the current situation. However, ADAPTIVE does not allow the same power optimizations compared to STATIC.

DYNAMIC Model Targets

DYNAMIC mode tells Vuforia Engine to expect the Model Target to be in constant motion. It will therefore not use Device Tracking to stabilize stationary targets to support Extended Tracking. Occluding the Model Target or losing it in the camera view will result in loss of tracking and leaving the Model Target at rest will result in reduced tracking quality compared to other motion hints. Therefore, use DYNAMIC only in scenarios with handheld Model Targets that are highly interactive, as in some use cases with toys.

Set the motion hint in the Model Target Generator

In the MTG, the motion hint can be set to STATIC, ADAPTIVE or DYNAMIC from the Motion Hint tab. During the creation of a Model Target database, the MTG stores the value in an .xml database file. The motion hint value can be changed later in Unity or with a get/set call. See Motion Hints on how to do so.

Model Target Tracking Mode

Model Targets can be configured in 3 different modes for different model types. Choose between modes that are optimized for different scenarios and for applications that, for example, track car exteriors or reflective objects where the user will inspect the object from different views, or for Model Targets created from 3D scans.

Choosing the right tracking mode for your use case can help reduce drifting of the digital content as well as improving the alignment between the Model Target and the real-life object. For Advanced Model Targets that are created from 3D scans, selecting the 3D Scan mode can also improve target recognition.

  • DEFAULT mode can be used in many scenarios and will deliver consistent tracking on a broad range of object types.
  • CAR mode is useful for car exteriors and similar highly reflective large objects.
  • SCAN mode should be set for Model Targets created from 3D scanned objects.

DEFAULT is suitable for a variety of models. It provides robust tracking on most objects and can be freely used in combination with the different motion hints.

CAR greatly improves the tracking performance on cars and large reflective objects and will reduce drift in certain situations at the cost of higher CPU load. For optimal performance on cars that do not move while tracking, we recommend using this mode together with the STATIC motion hint.

NOTE: CAR mode works best on high-end devices with larger processing power.

SCAN should be used when the digital model originates from a 3D scan, e.g. if the model was created from scans with the Structure Sensor or using Photogrammetry. In particular, this mode improves recognition for Model Targets created from 3D scans that include a texture. For untextured scans, this mode will not have a strong effect.

NOTE: If the scan includes a photorealistic texture, make sure to select the Realistic Appearance Mode while training an Advanced Model Target database.

Set the tracking mode

The Model Target Tracking Mode can be configured for each model in the MTG in the Model Type tab.

Select one of the following model types from the menu:

Note that the 3D Scan mode can only be selected in the MTG. This mode cannot be set or changed later in Unity or at application runtime.

Set Tracking Mode in Unity

The Model Target Tracking Mode can be changed after creating the Model Target database. Model Targets created with Mode DEFAULT can be changed to mode CAR and vice versa. The 3D SCAN mode can only be set before creating and training a database and cannot be changed in Unity.

  1. In Unity, a dropdown is available in the Model Target Inspector window that can be used to select the tracking mode and motion hint:

Tracking Mode in Native

Use the vuModelTargetObserverSetTrackingMode call before activating the corresponding Observer to change the tracking mode for the Model Target.

Please note that the same restrictions apply for changing the mode as in the Unity inspector. Model Targets created with mode  _DEFAULT can be changed to mode  _CARS and vice versa, but the  _SCAN mode cannot be changed or set at runtime.

Calling vuModelTargetObserverSetTrackingMode with an incompatible mode or after the Model Target Observer has been activated will result in an error. See Model Target API Overview for details.

Camera and Lighting Considerations

Expect to see better performance on high-end devices in good lighting conditions when tracking a Model Target. See Camera Focus Modes to ensure the camera have the object in focus. Some devices also have cameras that work well low light settings. For older devices that show low tracking performance, the lighting conditions can be improved by turning on the flashlight on a mobile device. However, be aware that using the flashlight on reflective objects such as industrial machinery can do more harm than good.