Release 2.2

Version 2.2.0

Version Info

Release date: 21.09.2021
Release info: Feature Release

General

This is a minor feature release for VisionLib which:

• updates external dependencies to remove known vulnerabilities,
• restructured the API reference (see VisionLib.SDK.Unity),
• improves performance of internal image processing and
• adds several bug fixes and enhancements.

Thank you to all our users for giving feedback.

Features

Unity

• Added SetGlobalObjectPose command for TrackingAnchor.
• Added a new AddModel function to ModelTracker.cs and corresponding AnchorAddModel function to TrackingAnchor.cs. This function makes it possible to load additional models (model parts) for one tracking target during runtime.
• The DeviceInfo.Camera class now contains supportsDepthData and supportsSmoothedDepthData variables, which state whether the VisionLib supports the acquisition of (smoothed) depth data with the respective camera.
• Added possibility to set a rootTransform for a WorkSpace. This is only relevant for ARFoundation and HoloLens scenes using DynamicModelTracking (see Using Models from Unity Hierarchy for Tracking (Beta)) and is automatically determined if the Destination object of the WorkSpace points to a TrackingModel.
• Added a new SetModelPropertyURI function to ModelTracker.cs and corresponding AnchorSetModelPropertyURI function to TrackingAnchor.cs. This function makes it possible to load specific model, which is specified by an uri. This will remove all other models.

Native

• The JSON object returned by the vlWorker_GetDeviceInfo function now contains supportsDepthData and supportsSmoothedDepthData variables for each object in the availableCameras arrays. This variables state whether the VisionLib supports the acquisition of (smoothed) depth data with the respective camera.
• Updated external dependencies (curl, Eigen, ffmpeg, libJPEG, libPNG, libTIFF, OpenCV, OpenSSL) to remove known vulnerabilities.
• Removed external dependencies imagexx, jpegxx and pngxx.

Enhancements / Bugfixes

General

• Fixed a bug in AutoInit, which lead to an error when trying to learn with an empty WorkSpace.
• Fixed memory alignment bug in AutoInit on Windows 32 bit versions.
• Fixed a bug in internal generation of InitData. This might lead to a slightly different behaviour when learning initData during runtime (see /ref vlUnitySDK_Tutorial_InitData).
• Improved speed of some internal image processing functions.

Unity

• Fixed bug which prevents the orientation overwrite in ScreenOrientationObserver to work correctly.

Interface Changes

General

• The functions GetRadialDistortion() and SetRadialDistortion() from IntrinsicDataWrapper are deprecated. Please use the GetDistortionParameters() and SetDistortionParameters() instead.

  Unity

• The function SetModelPropertyAsync is obsolete. Please use the specific SetModelPropertyEnabledAsync or SetModelPropertyOccluderAsync function instead.

The function AnchorSetModelPropertyAsync is obsolete. Please use the specific AnchorSetModelPropertyEnabledAsync or AnchorSetModelPropertyOccluderAsync function instead.

• The TrackingState now contains the ratio of successfully learned poses.
• The function SetInitPose for TrackingAnchor is obsolete. Please use the SetGlobalInitPose instead.

Others

Documentation Changes

• Fixed a bug in the generation of the documentation for the API and restructured this documentation.