Precision 3D Object Alignment is a feature that significantly improves the accuracy of aligning to real world objects. Obtaining two additional angles can improve alignment such that the mean deviation is less than 3mm which is more than 2x more accurate than before.

Please note that enabling Precision 3D Object Alignment will automatically disable Object Active Tracking as these two placement modes cannot be used at the same time.

Step-by-Step Instructions

1. Enable Precision 3D Object Alignment in Settings:
   • Open the WorkLink App.
   • While viewing the list of scenarios, tap on the Settings icon (⛭) in the upper-right corner of the screen.
   • Toggle on the Precision 3D Object Alignment option to enable this feature. Note that enabling this feature will disable Active Tracking on mobile devices.
     
   • In case you want to use 25 segments instead of 9, be sure Object Tracking Debug is enabled as well, so you can turn on Increase Precision Alignment Segments in debug mode while aligning.
   • Exit the Settings menu to return to the Scenario List.
2. Load a Scenario with AR Object Trackers
   • From the Scenario List, load a scenario that includes at least one AR object tracker.
3. Start Alignment
   • During content placement, select any AR object tracker to start the alignment process.
   • When the real-world object is initially aligned, the app will prompt you to obtain more angles to improve alignment.
4. Precision Alignment Segments
   • The angle capture sphere has 25 areas or segments.
   • 
5. Obtain Additional Angles
   • It is highly recommended to capture two additional angles that are furthest away from the center for optimal accuracy. Ideally use angles from which you will view the content throughout the scenario for optimal perceived accuracy.
   • If you do not need additional angles and are satisfied with the initial alignment, tap on Skip.

• For each additional angle, use the sphere gizmo to guide you. Tap on the screen to confirm each angle. Hold still while capturing the additional angle. Each section of the sphere will turn green as you capture an angle.
  
• Usually 1 or 2 additional angles should be sufficient. Use angles that are furthest away from the center for optimal accuracy.
• It is not needed to capture all angles, you can confirm after you're satisfied with the accuracy you perceive.
• If the last angle does not improve alignment based on your visual inspection, tap on Remove Last to discard it.
• To restart the alignment process from the beginning, tap on Start Over.

1. Finish Alignment
   • Once you are satisfied with the alignment, tap on Confirm to finalize the placement of the AR content.

FAQ

Q: Why doesn't my precision alignment seem to get better?

The WorkLink App provides debugging tools to help diagnose alignment issues. Several factors may affect precision alignment.

Segments consistently turning yellow after capturing several angles, are a good indication that your setup might not be ideal and needs to be tweaked a bit more. Essentially it means that the measurements taken in orange/yellow segments, deviate strongly from the last taken segment. The origin of this symptom can have different reasons. Those potential reasons are explained below.

 

Understanding Precision Alignment

Notification displays:
Matching Measurements: X, Recommended: 3.

• Matching measurements = the total amount of measurements the algorithm is using to come to a better precision alignment result.
• Recommended = 3, we recommend confirming precision alignment after capturing at least 3 matching measurements.

Precision alignment captures multiple SLAM position measurements and adds them to a measurement pool. An algorithm evaluates these positions and selects plausible ones to average them into a more accurate SLAM position, improving tracking precision.

Compared to single-measurement tracking, precision alignment helps reduce axis-based offsets, which commonly occur when confirming an object tracking angle, placing an object’s position from a single viewpoint in SLAM. These offsets become noticeable when viewing the object from a different angle.

Common Reasons for Precision Alignment Issues

1. False Positives (Incorrect Tracking Data)

• Cause: If inlier ratio values are set too low during authoring, the tracking algorithm may recognize incorrect features as part of the object, leading to unstable positioning.
• Symptoms: The object appears to "fly around" irratically or shift unpredictably.
• Mitigation:
  • Use the app’s debugging tools with active tracking to observe false positives.
  • Increase the inlier ratio threshold to prevent incorrect matches while ensuring it is not too high, which could prevent tracking altogether.

2. Object Movement Between Measurements

• Cause: If the tracked object is moved between precision alignment measurements, SLAM positions will vary significantly, and previous valid positions will be discarded.
• Mitigation:
  • Keep the object stationary while capturing measurements.

3. Poor SLAM Conditions

Unstable SLAM tracking can cause misalignment and inconsistent measurements. Potential causes include:

Issue	Description
Poor lighting	Too dark or overly bright conditions.
Lack of visual features	Plain surfaces (like a white featureless floor) or highly uniform areas.
Reflective surfaces	Reflective/glass materials can be problematic for SLAM tracking, as it can confuse the reflective surface with the reflected features, creating incorrect depth assessment.
Motion blur	Fast movements affecting SLAM tracking stability.
Occlusions	Objects blocking key reference points. Ensuring the camera captures enough environmental details can help, for example by showing it the areas you will be using the AR experience in, so additional reference points can be gathered.
Repetitive textures	Identical patterns (e.g., repetitive carpet patterns) can confuse the tracking system.
Environmental changes	Moving objects (e.g., tracking near a window on a moving train) or shifting lighting (e.g., changing shadows).

• Mitigation:
  • Ensure the device has clear, well-lit, and feature-rich surroundings for tracking.
  • Minimize sudden movements and occlusions.
  • If necessary, reposition the camera to include more environmental reference points.

4. Spatial Drift on HoloLens 2 (HL2)

Cause: Over time or across certain distances, the HoloLens 2 may accumulate spatial mapping errors, especially in environments with few distinct features or inconsistent lighting. These errors can cause holograms to appear offset from their original position, which negatively affects precision alignment.
Symptoms:

• Holograms appear to "drift" or misalign when walking around a large object, even after multiple measurements.

Mitigation:

• Use the HoloLens "Reset Holograms" function (found under Settings > System > Holograms) to clear the current spatial map and force the device to re-scan the environment. This can resolve corrupted or outdated spatial mapping data.

• Ensure the environment has sufficient visual features and stable lighting.

• Avoid environments with poor SLAM conditions (as detailed above).

• Consider guiding the user to walk around the environment slowly before performing alignment, so the device collects richer spatial data.

5. Spatial Drift on Android and iOS

Cause: Unlike HoloLens 2, which relies on a high-resolution spatial map, mobile AR on Android and iOS (typically using ARCore and ARKit respectively) maintains SLAM tracking via visual-inertial odometry without persistent or large-scale environmental mapping. This makes drift more pronounced as the user walks longer distances or moves between visually distinct areas.

Symptoms:

• Holograms gradually appear misaligned as the user walks across larger spaces (e.g., >3–5 meters).

Explanation:
Mobile AR platforms are optimized for short- to mid-range experiences (typically within a few meters of the starting point). As distance from the origin increases, accumulated errors in device pose estimation can cause drift, especially in:

• Large featureless areas (e.g., warehouse floors or long hallways).

• Environments with changing lighting or visual occlusions.

• Sessions exceeding several minutes without reorientation.

Mitigation & Best Practices:

• Design experiences within constrained spatial boundaries, ideally within a 3–5 meter radius of the initial alignment area.

• Avoid frequent transitions between visually distinct areas (e.g., from a brightly lit room to a dim corridor).

• If walking longer distances is required, consider using intermediate visual checkpoints or instructing the user to re-align at specific locations.

• Provide clear guidance to end users on how far they can reasonably expect accurate tracking.

• Note: This is a platform limitation rather than a bug. Understanding and communicating the expected accuracy envelope to end users is critical when authoring or deploying AR experiences on Android or iOS.

6. Hardware calibration on HoloLens 2

Cause: On large objects a minor hardware calibration issue becomes visible that is typical to HoloLens 2. Usually the HL2 has a slight miscalculation between the physical size of the 3D content and the real world's physical size, of about 0.5% scaling. This is not visible on small objects, but when aligning to large objects, it becomes more apparent.

Mitigation: To counter this factory calibration issue, we built in a HoloLens 2 calibration feature, defaulting to 100.5% when enabled.

You can find it under the settings menu: "Scale Calibration (beta)": Scales the size of the content by 100.5%

Q: What is SLAM?

A: Simultaneous Localization and Mapping: SLAM in AR refers to a technology that enables real-time tracking of a device's position and orientation while simultaneously constructing a map of the environment, allowing for accurate placement of virtual objects in the real world.

Q: What is the inlier ratio?

A: The inlier ratio in object tracking is a percentage-based match that compares a line model, generated from a 3D model, with hard contrast lines in the video feed. This comparison is used to measure the success of object tracking. 

Learn More

• WorkLink Create: Object Tracking