Vuforia SDK Release Notes

Vuforia SDK v6.5.x

Bugfixes / Features   
All
  • Text Recognition has been deprecated and will be removed in a future major SDK release
  • Smart Terrain API has been deprecated and will undergo significant changes in Vuforia 7
  • VuMark: Fixed an issue where VuMarks did not respect the persistent extended tracking mode (extended tracking map was reset internally even though persistent mode was enabled)
  • Fixed small memory leaks around Vuforia::init, Vuforia::onResume and DataSet::load

Unity

  • Vuforia is now natively integrated with Unity 2017.2
  • Vuforia can be now installed from the Unity Download Assistant or from the dedicated Vuforia Augmented Reality Target support installer.
  • Vuforia libraries and scripts no longer need to be imported into the Unity project. Instead, enable Vuforia in the XR panel in the Unity player settings after installing it with the Unity Editor (“Vuforia Augmented Reality supported”).
  • Vuforia prefabs have been replaced with Game Objects that can be created from the  /Game Object/Vuforia menu.
  • Please consult the migration guide for steps to upgrade legacy Vuforia projects to the integrated version.
  • Optimized stereo rendering for digital eyewear and viewer type devices by leveraging Unity’s stereo rendering optimizations
  • Integration with 3rd party SDKs such as Cardboard or HoloLens has been simplified, not longer requiring binding of their camera rigs. 
  • New Positional Device Tracker API. This feature can be activated and configured in the editor via the device tracker section of the Vuforia configuration. The mixed reality controller has also been extended to  support positional device tracker: added two new generic mode supporting both Rotational and Positional tracking: HANDHELD_AR_DEVICETRACKER and VIEWER_AR_DEVICETRACKER. This feature is currently only supported on Google Tango devices (Android).
  • Vuforia will leverage Tango capabilities to improve extended tracking on Tango enabled devices
  • Recommended rendering API: Metal (iOS), OpenGL ES 3.0 (Android), DX11 (UWP).
  • Vulkan Graphics API is not supported (Android).
  • The minimum supported Android version is 4.1
  • The minimum supported iOS version is 9.
  • The minimum supported Windows version is 10 (UWP only).
  • GearVR support has been deprecated and will be removed in a future major Vuforia release.
  • Epson BT-200 is not supported any longer in Unity 2017.2 (please use our native SDK or a previous version of Unity + Vuforia Unity Extension).
  • The ROTATIONAL_HANDHELD_AR and ROTATIONAL_VIEWER_AR modes in the MixedRealityController have been deprecated. They will be removed in a future Vuforia release.
  • Editor: fixed incorrect import of image preview for Object Recognition.
  • iOS: fixed upside down camera issue with the front camera.
  • Front camera image may freeze on Pixel C when toggling from front to back cams and then return to front cam
Android
  • Device Tracker: New PositionalDevice Tracker API exposes 6-DOF device pose in world coordinate system. Currently supported only on Google Tango devices. 
  • Support for Google Tango. Vuforia will leverage Tango capabilities to improve extended tracking and device tracking on Tango enabled devices.
  • This release includes a 64 bit ARM build of Vuforia.
  • Vuforia now uses Android Camera2 API for camera access by default on Android M and above. If you use the low-level advanced camera API you will need to update your code as the field names have changed. 
  • Vuforia::init will now request the camera permissions via a system dialog on applications targeting Android M and above.
  • Changes in Vuforia::init, Vuforia::onResume and CameraDevice::init to support the camera permissions dialog and Google Tango mean that these API can no longer be run from the UI thread, but must be called asynchronously.
  • Fixed small memory leaks around Vuforia::init, Vuforia::onResume and DataSet::load
  • The advanced camera API CameraDevice.getFieldString did not work in Java due to an issue in the Java bindings for that API.
  • GearVR support has been deprecated and will be removed in a future major SDK release
  • Removed the deprecated API Renderer::drawVideobackground
  • Black screen my occur on-resume when app is left in pause for long periods, observed on Galaxy S8 and Pixel.
iOS

  • Fixed small memory leaks around Vuforia::init, Vuforia::onResume and DataSet::load
  • Removed the deprecated API Renderer::drawVideobackground
UWP

  • Removed the deprecated API Renderer::drawVideobackground
  • Error message not shown when native camera apps conflict with Vuforia camera access. Camera apps must be closed, to unbind the camera, before running a Vuforia UWP app.


Known Issues  
All SDKs N/A

Unity
  • Camera clipping in stereo AR mode change slightly when pause-resuming the app on iOS.
  • Black screen in AR mode when launching a Vuforia + Cardboard VR SDK on iPhone 5
  • Incorrect view in VR mode with Vuforia + Cardboard VR SDK on iPhone SE and IPhone 5
  • Getting access to Camera Image with the GetCameraImage method successively from multiple scenes doesn’t return correct images
  • Using direct rendering mode in Unity (via DrawMeshNow or GL calls) will lag when displayed over video background on iOS with Metal API at 60fps
  • Duplicated target behaviors have the same preview image even if you change the target/dataset from the menu
  • Google Tango: Connecting to the Tango service can take multiple seconds which can delay Vuforia initialization
  • Video Background rendering in play mode doesn’t deallocate video memory when switching scenes
  • VR Mode from the Mixed Reality Controller (VIEWER_VR) is not supported on ODG R-7
  • AR Camera Lags when using input field front of the video background (Unity issue: https://fogbugz.unity3d.com/default.asp?934841_l6klsmjd5av7g5th )
Android

 

  • Google Tango: Connecting to the Tango service can take multiple seconds which can delay Vuforia initialization
  • Multiple rendering issues have been observed on the Nexus 7 (front camera only Android tablet).

 

iOS

  • The new Vuforia Positional Device Tracker API is not yet supported on iOS devices.

UWP

  • The new Positional Device Tracker API is not yet supported on UWP devices.

 

 

Vuforia SDK v6.2.x



Bugfixes / Features 
All
  • Vuforia now supports tracking of an unrestricted number of Image Targets simultaneously 
  • Fixed performance issue with VuMarks when using the “track from appearance” option
Unity 6.2.10
  • Fixed an issue causing Vuforia UWP apps to fail the Windows App Certification Kit test when submitting to the Windows Store
Android
6.2.10
  • Fixed an issue causing a dark camera preview image on some Nexus and LG devices 
  • Fixed a memory leak on Android when  passing the first Activity during initialization
iOS
6.2.9
  • Fixed an issue causing the SDK to hang on start-up if the camera is disabled in Settings > General > Restrictions.
UWP
6.2.9
N/A
Known Issues
All SDKs N/A

Vuforia SDK v6.2.6

Bugfixes / Features 
All
  • Vuforia now supports the OpenGL ES 3.x graphics API
  • The key types for devices ( Mobile vs Digital Eyewear ) have been unified so that a license key can support both mobile and digital eyewear apps. Existing Vuforia Digital Eyewear keys will continue to work for new digital eyewear apps created with Vuforia 6.2. Use of a newly generated unified license key to build a digital eyewear app will require an upgrade to Vuforia 6.2
  • Fixed a rare issue where detection performance was poor for some ObjectTargets
  • Fixed a rare issue where tracking performance was poor when multiple ObjectTargets were tracked simultaneously
  • Fixed an issue where Extended Tracking or Smart Terrain sometimes failed to initialize on low feature targets
  • Fixed an issue where DataSet::exist() API would take very long to complete when an app had a large amount of files in the Assets folder
  • Fixed an issue where TargetFinder.getInitState() wrongly returns INIT_SUCCESS after it has been deinitialized
  • Fixed an issue where the DeviceTracker would produce jittery poses after resuming the application following a long pause/resume cycle
Unity
  • A new configuration window replaces many ARCamera-inspectors. The configuration window can be opened with Vuforia/Configuration in the Unity Editor Menu and affects all scenes in the project. Please see the migration guide for more information
  • Vuforia is initialized and deinitialized only once per application instead of every scene
  • Fixed a rare issue where stereo rendering was incorrect when switching from VR to AR
  • Fixed an issue in playmode where the CurrentFrameIndex was wrong after switching scenes
  • Fixed an issue with dynamic shadows on ODG R7
  • Fixed an issue where changing the video mode at runtime was not working
Android
  • Added C++ Vuforia::init() method on Android to match Java version of the same
iOS N/A
UWP N/A
Known Issues
All SDKs
  • FrameMarkers are no longer supported ( deprecated as of 6.0.112 ). We recommend using VuMarks as an alternative.
Unity
  • The minimum supported Unity version is now 5.3.6
  • In Unity 5.5, no preview image is shown for new object targets in the editor. The preview image can be enabled by changing the image settings to type “Default” and shape “2D”
Android
  • ODG X6 AR glasses are no longer supported
  • Augmentations inverted on Nexus 5X
  • Eyewear calibrated adjustment and projection matrices are inconsistent with VR
  • On some device running Android N performance and framerate issues have been observed, in particular when using Extended Tracking
iOS
  • Vuforia is not working with Metal API validation on iOS 10 in Unity
  • Video background stuttering seen when moving the camera at 1080p60 on iPhone 7 Plus
UWP N/A

Vuforia SDK v6.1.17

Bugfixes / Features 
Unity
  • Introduced support for Microsoft HoloLens
Known Issues
All SDKs N/A


Vuforia SDK v6.0.117

Bugfixes / Features 
Unity
  • updated to libPNG 1.6.25 containing fix to resolve known vulnerability issue.
Android
  • updated to libPNG 1.6.25 containing fix to resolve known vulnerability issue.
UWP
  • updated to libPNG 1.6.25 containing fix to resolve known vulnerability issue.
Known Issues
All SDKs N/A

Vuforia SDK v6.0.114

Bugfixes / Features 
Unity
  • This release introduces a new Find The Penguin sample app
  • Unnecessary warnings removed from C# scripts
Known Issues
All SDKs N/A

Vuforia SDK v6.0.112

Bugfixes / Features 
All
SDKs
  • Introducing support for VuMark
  • Improvements in Extended Tracking performance
  • Deprecating Frame Marker APIs, these will not be supported in future releases
UWP
  • Premier of public UWP SDK for Windows 10 devices
  • Public Early Access Program (EAP) UWP SDK version available for HoloLens
Android
  • Introducing Advanced Camera API
iOS
  • Introducing support for 1080p @ 60fps
Unity
  • Introducing 64bit Editor support for Windows
Known Issues
All SDKs N/A

Vuforia SDK v5.5.9

Bugfixes / Features 
All
SDKs
  • Vuforia now supports rotational head tracking and distortion correction natively. Development of Mixed Reality AR/VR apps no longer requires the integration of a 3rd party VR SDK.
  • Rotational head and hand tracking in 3 degrees of freedom (3 DoF) is supported along with support for head and hand models to define rotational pivot offsets
  • Portable AR/VR rendering for both mono and stereo viewports is supported using Render Primitives
  • Viewer management is supported to configure Vuforia for drop-in VR headsets, such as Cardboard and Gear VR, using Viewer Parameters
  • Changes to legacy APIs, specifically around rendering and device management, as well as renaming of core classes, the QCAR namespace and include directory from QCAR to Vuforia. See: Changes in Vuforia 5.5
  • Introduced a new Device class to represent device hardware properties
  • The Eyewear class has been replaced with a specialization of the Device base class and renamed EyewearDevice class. This class only represents Optical See-Through Devices. Video See-Through devices are represented by the base Device class.
  • Extended the State and Trackable Result classes to utilize multiple coordinate systems. A Trackable pose in Vuforia can now be in either the camera or world coordinate system.
  • Trackable poses now have an independent timestamp property to support on-demand pose updating and pose prediction
  • All physical measurements in the SDK have been harmonized to meters
  • New Camera Direction enum field is provided to define camera direction on devices with only a front facing camera
  • The Vuforia Watermark now appears persistently for apps that are on the Starter plan
Android
  • Eyewear User Calibration now returns two Matrix34F matrices instead of a single Matrix44F for the camera-to-eye pose and the calibrated eye projection matrix.
  • Eyewear User Calibration API has been extended with a consistency qualifier that can be used to improve the user calibration
  • Fixed a bug where the camera image appeared flipped on the Nexus 5X back camera and Nexus 6 front camera
iOS
  • The Apple Metal API is supported for iOS 8+, though Cloud Reco requires the use of iOS 8.3 and above
Unity
  • The VideoTextureRenderer script on the ARCamera prefab has been renamed to VideoBackgroundManager and extended to allow configuration for the video background behavior
  • A new utility class named MixedRealityController has been introduced that provides a simple runtime interface to switch between different mixed reality modes.
  • The minimum supported Unity version has been updated to 5.2.4
Known Issues
All SDKs
  • The RotationalDeviceTracker only works on devices with a built-in gyroscope
  • Chromatic aberrations may be visible in the periphery of the lens when using distortion correction
Android
  • Visible latency observed on the Samsung Galaxy S5 when using Rotational Device Tracking
iOS
  • Viewport offsets greater than zero differ in the Y direction between Open GL ES 2.0 and Metal
  • Applications in VR mode are required to start in landscape mode
Unity
  • When transitioning from mono to stereo AR, the left eye view may remain dark for a split second. It is recommended to mask these transitions to conceal this issue.

Vuforia SDK v5.010

Bugfixes / Features 

Android

  • Fixed an issue in which a graphics layer error could cause video background rendering to stop

  • Improved the accuracy of eye calibration accuracy for digital eyewear apps

  • Fixed an issue on the ODG R-7 in which the stereo rendering could revert to mono when running an app a second time

Unity

  • Added support for Unity 5.3
  • Compatibility with Unity 5.3 built-in VR support for GearVR
  • Improved rendering performance for stereo AR apps that render the video background
  • Fixed an issue in which the camera image was not rendering on select versions of the Galaxy S3
  • Fixed a permission issue on Android M. The target SDK version is now set to 22 in the AndroidManifest 
Known Issues

All SDKs

  • No Known Issues


Vuforia SDK v5.0.6

Bugfixes / Features 

iOS

  • Fixed an issue where devices running iOS 9 experience a black screen on resume after pressing the sleep/wake button

Unity

  • Fixed an issue where devices running iOS 9 experience a black screen on resume after pressing the sleep/wake button
Known Issues

iOS

  • Sample apps may experience a crash after pressing the sleep/wake button

Vuforia SDK v5.0.5

Bugfixes / Features 

All SDKs

  • The SDK now supports mobile, video see-through digital eyewear and optical-see-through digital eyewear devices from one SDK package and single set of APIs.
  • Support for mixed reality applications built to function with Google Cardboard and the Oculus Mobile SDK ( Gear VR ).
  • The local device database target limit has been increased. See: Best practices for large device databases
  • Provided the ability to control the application render frame rate. There is a new Utility API to get the recommended render frame rate settings based on application hints. Removed the flag for camera-synched rendering in VideoBackgroundConfig.
  • Extended CameraCalibration class to expose field of view. This dimension is use to perform scene scale calculations for digital eyewear.
  • Cloud Reco APIs have been introduced to control filtering behavior for matches.
  • Missing classes for Smart Terrain have been exposed in the Vuforia Java API bindings.
  • Fixed an issue where MultiTargets were occasionally failing to detect.
  • Eyewear: Viewer can be explicitly set by name if it is not automatically detected
  • Eyewear: Clipping plane distances can be set explicitly if they are not automatically detected
  • Eyewear: ODG R-7 supports predictive tracking.
  • Eyewear: ODG R-7 supports on-demand target poses based on latest sensor-measurements via the new StateUpdater class.
Android
  • The minimum supported Android platform version is Android 4.0.3 (Ice Cream Sandwich)
Unity
  • Vuforia 5 supports Unity versions 4.6.7 and 5.1.3p1 for Google Cardboard and Oculus Mobile SDK integration. Unity 5.1.3p1 resolves performance issues experienced with earlier 5.x releases. You can download 5.1.3p1 at https://unity3d.com/unity/qa/patch-releases?version=5.1
  • Eyewear: Vuforia 5 supports Oculus Mobile SDK versions 0.5.0 and 0.6.0.1
  • Eyewear: Vuforia 5 supports Cardboard SDK for Android version 0.5.5 and Cardboard SDK for Unity version 0.5.0.
  • ​The Qualcomm Augmented Reality folder has been renamed to Vuforia. Various classes and methods containing “QCAR” have been renamed as well. See: Migrating Unity Projects to Vuforia 5
  • A new VR camera binding workflow has been provided to facilitate the use of alternate head and camera rigs with the Vuforia ARCamera.
  • For Digital Eyewear apps, the Field of View of the video background is now inherited from 3rd party SDKs. The scene scale parameter has been deprecated. 
  • Fixed an issue where the Multi Target preview in the Unity editor did not support targets with consecutive part rotations
  • Fixed various issues related to camera mirroring. Changing mirroring at runtime can be done by changing the reflection flag of the VideoBackgroundConfig.
  • Fixed an issue where a workaround was required to enable smart terrain on optical see-through digital eyewear devices.
Known Issues

All SDKs

  • Mixed Reality apps are very demanding and performance may differ across devices. High resource demands may result in a drop in render frame over time. See: Framerate Optimization for Mixed Reality Apps
  • Augmentation position for UserDefinedTargets cannot be accurately determined on optical-see-through eyewear. 
Android
  • Samsung Galaxy S6 may exhibit undesirable video artifacts when operating at its thermal limit.
Unity
  • Eyewear: When using Cardboard SDK for Unity version 0.5.1, iOS apps will crash when scanning a viewer QR code.
from 4.2.3
  • Object Recognition does not work well with objects that have very little texture (e.g. a cup, a TV). Please see the developer guide for supported object types.
  • Object Recognition: Possible poor detection of low textured targets in database that contains other high textured targets 
  • Object Recognition: Some low texture objects like toy action figures may not  detect well in a cluttered background environment 
  • Vuforia Object Scanner: Application may run with low FPS after longer scanning sessions. 
  • Vuforia Target Manager: Upload of a large OD file may take multiple minutes 
  • Extended Tracking does not work after using SmartTerrain and vice-versa. You need to re-init Vuforia first. 
  • Extended Tracking: Object Targets are reported as TRACKED instead of EXTENDED_TRACKED
  • Nexus 5 with Android L: Camera view is very dark. This is a known issue in the device driver.  
  • The minimum supported Unity version is 4.6.7
  • A crash may occur when changing scenes if the Text Tracker has been started. This is only observed on iOS devices when the IL2CPP scripting backend is used.
    Before loading a new scene, stop the TextTracker, then, load the new scene from a Coroutine and call “yield return new WaitForEndOfFrame()” before loading the next scene.
  • Vuforia does not support the Metal rendering API on iOS. Be sure that OpenGL 2.0 is selected as the Graphics API when building for iOS in Unity 4.6.3 or later. 
from
3.0.9
  • The Vuforia SDK only supports OpenGL ES 2.0.
  • Extended Tracking and Smart Terrain are only supported on multi-core devices.
  • Extended Tracking and Smart Terrain may fail during very fast motions.
  • Extended Tracking may fail in very plain environments (e.g. blank wall)
  • Extended Tracking may result in more jitter and a visual snap back on the augmentation when the target re-enters the field of view
  • Extended Tracking and Smart Terrain may fail in non-static environments (i.e. large motions or environment changes)
  • Extended Tracking can only be used for a single UDT target at a time
  • Extended Tracking targets must have mutually consistent sizes/scales if they are used together in the same scene
  • Extended Tracking may fail to relocalize when the phone is moved in pure rotational motion around the scene.
  • Extended Tracking and Smart Terrain are not supported on front cameras
  • Smart Terrain and Extended Tracking cannot be used at the same time. You need to stop SmartTerrain before using Extended Tracking and vice versa.
  • Smart Terrain may occasionally reconstruct flat objects (such as a drink coaster) as a 3D object and return it as a Prop
  • Smart Terrain may occasionally reconstruct two nearby objects as a single object
  • Smart Terrain may fail to reconstruct the ground plane and objects if the initialization target is placed on the border of the primary surface (e.g. table). The initialization target needs to be placed well within the primary surface to avoid this issue.
  • Smart Terrain may in some occasions wrongly reconstruct the primary surface such that it appears to extend into the background. This may happen if the primary surface and background are visually not distinguishable (e.g. white table in front of a white wall)
  • Smart Terrain may not work well on play surfaces that are non-uniformly textured (E.g. large tiled floor with different colored tiles)
  • Smart Terrain cannot be initialized with User Defined Targets
  • Text recognition may occasionally mis-detect a word or detect a partial word
  • Time to recognize a word depends on number of words and the length of words. Specifically, Text Recognition may take longer than 2 seconds on some single core devices, as well as on some devices when more than 3 words are in the recognition area or for words that are longer than 24 letters.
  • Text recognition may fail to track text well if the page of text and the background are not on the same plane.
  • Low contrast targets may not be detected well in highly cluttered backgrounds
  • Text recognition does not support the front camera.
  • Enabling additional words in the Text recognition feature may slightly increase recognition time
  • TextRecognition may not work if another Tracker is used before.
  • TextRecognition may stop working after scanning a 45 letter word. 
  • Cloud recognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • The Vuforia SDK only support armeabi-v7a devices. Devices with no VFP unit or armv6 are not supported.
  • Not all focus modes are supported on all devices
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.
  • Multiple simultaneous Multi Targets are not supported

Vuforia SDK v4.2.3

Bugfixes / Features 

All SDKs

  • The SDK now supports offline license verification, internet connectivity is no longer required to authorize a license key. Existing apps will need to obtain an updated license key from the License Manager to enabled this functionality. You can find your updated keys by clicking on the app's name in the License Manager UI.
  • The Vuforia watermark is displayed once daily during the initial app session. 
Known Issues

Unity

  • Unity 5 scenes that use a non-legacy Specular or Normal Mapping shader and a Canvas UI element with the 4.2.3 ARCamera will not render the UI element on iOS and will show a white screen on Android


Vuforia SDK v4.0.105

Bugfixes / Features 

Unity

  • Added support for Unity 5.0. Supported platforms are 64bit & 32bit OSX and 32bit Windows systems.
  • Added support for 64bit Play Mode on OSX
  • Added support for 64bit iOS builds on OSX on OS X 64bit using Unity 5.0
  • Fixed an bug that was causing a false network connection error to be thrown during QCAR initialization in Windows Play Mode.
Known Issues

Unity

  • A crash may occur when using an external webcam for Play Mode on OSX with Unity 5.
    This crash is always seen in OS X 10.9 Mavericks and the first time Unity is launched after rebooting OS X 10.10 Yosemite.
  • The Unity 5.0 editor crashes when an ImageTarget prefab instance in the ImageTargets sample is selected prior to selecting another prefab in the hierarchy when the sample is first loaded. Selecting an alternate prefab, such as the ARCamera, after initially loading the sample scene will prevent this crash from ever occurring. 
  • Play Mode is not available in the 64bit version of the Unity 5 Windows editor, we recommend using the 32bit version.
  • The minimum supported Unity version is 4.6.x.

Vuforia SDK v4.1.3-beta

Bugfixes / Features 

All SDKs

  • The following devices and tested firmware versions are supported by this release:
    • Epson BT-200
                  Firmware 1.1.2, Android 4.0.4
    • ODG R-6
                  Firmware 4.2.2, 23-Sep-2014, Android 4.2.2
    • Samsung Gear VR Innovator Edition on Note4
                  Supported models:      SM-N910F, SM-N910G, SM-N910T, SM-N910V,
                                                      SAMSUNG-SM-N910A
                  Firmware KTU84P.N910FXXU1ANK3_VR, Android 4.0.4
  • Added the ability to use personalized calibration profiles through a new Eyewear API
  • Added support for stereo rendering for optical and camera see-through digital eyewear devices
  • Added the ability to scale the scene and video background rendering to adjust for real-world scale on camera see-through devices
  • Added the ability to render scenes with no video background to support optical see-through devices

Unity

  • Added support for integrating the Oculus Mobile SDK (version 0.4.2) enabling use of the Oculus distortion shader and timewarping.
  • Added automatic scene scaling for GearVR. This value can be overridden in the inspector.
Known Issues

All SDKs

  • Epson BT-200: Due to variability in the camera the default user calibration is not well registered. Performing user calibration will resolve this.

Unity

  • This release does not support the latest Oculus Mobile SDK 0.4.3.1.
  • No API is provided to request a specific user calibration profile. The active profile is always used, which is Profile 1 in the Calibration App UI.
  • It is not possible to request that a specific user calibration profile is used when rendering, the currently active profile is always used. You can work around this by setting the active profile.
  • GearVR: Exiting an app using the back button when the app uses the Oculus Mobile SDK sometimes results in a crash.
  • GearVR: When integrating with the Oculus Mobile SDK if multi-threaded rendering is disabled the right eye does not render correctly.
  • Apps built with Unity 5 crash on exit when multi-threaded rendering has been enabled on Android.
  • Epson BT-200: If an app exits with Application.Quit directly from stereo mode (3D) on an Epson BT-200 the device will not revert to ‘2D’ mode. You can workaround this by calling Eyewear.SetStereo(false) before Application.Quit.
  • To use Oculus SDK 0.4.2 with Unity 5 ensure you have Unity 5 patch 2 installed. This can be obtained from https://unity3d.com/unity/qa/patch-releases

Vuforia SDK v4.0.103

Bugfixes / Features 

All SDKs

  • Improved detection times and 3D Object Target database sizes.  Developers will need to re-upload their Object Datafiles to get new databases with these improvements.
  • Added a new API hint for controlling maximum simultaneous 3D object targets (QCAR:: HINT_MAX_SIMULTANEOUS_OBJECT_TARGETS). This allows developers to optimize performance for use cases where only 1 target needs to be tracked at any time. 
  • Added a new API hint for partially loading object datasets (QCAR::HINT_DELAYED_LOADING_OBJECT_DATASETS). Enabling this behavior reduces memory consumption but delays detection results and is recommended only for large datasets with many 3D object targets. 
  • Fixed a bug where on rare occasions detection can fail after switching 3D ObjectTarget datasets. 
  • Fixed an issue occurring when a license key with leading or trailing white space could result in a "no network available" error rather than the expected "invalid key" error.
  • Fixed incorrect video rendering in portrait mode for some reflection settings and made a change in video background bind to clear viewport settings.
  • Fixed an bug that could crash apps when activating a dataset while the tracker is running.
  • The Vuforia SDK is now enforcing the OpenGL ES 2.0 rendering pipeline and must be initialized using  QCAR::setInitParameters(QCAR::GL_20) ( support for OpenGL ES 1.1 was deprecated in SDK v2.8.7 ).

Unity

  • Fixed a minor memory leak caused by pausing and resuming an app.
  • Fixed an issue causing “Apply Dataset Properties” to ignore changes to the XML configuration files for 3D object targets.
Known Issues

All SDKs

  • Device Databases created prior to this release will need to be assigned a license key and cannot contain more than 100 targets. You will need to remove targets from the database in the Target Manager before downloading a database and all targets will be included in the download.

Unity

  • A crash may occur when changing scenes if the Text Tracker has been started. This is only observed on iOS devices when the IL2CPP scripting backend is used.
Before loading a new scene, stop the TextTracker, then, load the new scene from a Coroutine and call “yield return new WaitForEndOfFrame()” before loading the next scene.
  • Vuforia does not support the Metal rendering API on iOS. Be sure that OpenGL 2.0 is selected as the Graphics API when building for iOS in Unity 4.6.3 or later.

 


Vuforia SDK v4.0.11 

Bugfixes / Features 

All SDKs

  • Fixed out of memory crash when loading Device Databases containing large Object Targets on low end devices.
  • ObjectTarget::setSize() is now supported for Object Recognition. 
  • Cloud Recognition is now supported for the Vuforia 4.0 Platform. 
  • Fixed a bug that could cause Object Data files to be corrupted if the Vuforia Object Scanner app was paused using the Home button.
  • Fixed a bug that slowed redetection of objects during an object scanning session.
  • Fixed a bug causing Smart Terrain reconstructions to restart if reset() is called after the reconstruction was stopped.  
  • Fixed a bug preventing Smart Terrain initialization targets from being reset programmatically.  
  • Fixed a memory leak arising from restarting the device camera. 
  • Fixed a JNI reference leak preventing Device Database references from being destroyed when unloaded. 
  • The Cloud Recognition TargetFinder class will no longer return a valid pointer if its enableTracking() method fails. 

iOS

  • 64-bit support has been added to the Vuforia SDK - libVuforia.a now contains armv7, armv7s and arm64 slices.
  • Sample apps have been updated to build for 32 and 64-bit iOS devices.
  • Introduced Object Recognition for the Vuforia iOS SDK.
  • Introduced Smart Terrain for the Vuforia iOS SDK.

Unity

  • iOS 64-bit support has been added to the Vuforia Unity extension.  libVuforia.a now contains armv7, armv7s and arm64 slices. Please make sure to use Unity 4.6.2 or later and use the IL2CPP compiler backend to get native 64bit support.
  • Fixed an issue seen on iOS 8 where the app could get stuck or show a black screen after pausing and resuming. 
Known Issues

Unity

  • A memory leak of ~1kb can occur when pausing and resuming a Unity app. 
 

Vuforia SDK v4.0.5 

Bugfixes / Features

All SDKs

  • Introduced Object Recognition, a feature than enables the detection and tracking of intricate 3D objects.
  • Introduced the Vuforia Object Scanner, an app that captures the features of 3D objects and produces an Object Data file to generate Object Targets in the Target Manager.
  • Introduced a new License Manager tool in the portal to help manage app license keys. Each Vuforia 4.0 app needs a unique license key obtained from the License Manager.
  • Refactored the APIs for ImageTracker and Smart Terrain: Please see the migration guide.
  • Introduced a native API for Smart Terrain.

Unity

  • Refactored QCAR initialization to simplify embedding Vuforia Unity apps into native Android and iOS containers.
  • Refactored the ARCamera to use a Unity texture for the video background.

 

Known Issues

All SDKs

  • Cloud Recognition is not available in the Vuforia 4.0 SDK Beta
  • Object Recognition does not work well with objects that have few distinct features, especially when these objects are placed against a cluttered background. .
  • Object Recognition apps may crash when loading Device Databases containing large Object Targets on low end devices.
  • Object Recognition apps may experience slower detection times when using Device Databases with multiple Object Targets.
  • Extended Tracking is not compatible with Smart Terrain. QCAR must be reinitialized if enabling either after the other has been enabled in an app.
  • Targets that use Extended Tracking may report a status of TRACKED instead of EXTENDED_TRACKED
  • Smart Terrain reconstructions are restarted if reset() is called, even if the reconstruction was previously stopped()  
  • Nexus 5 with Android L: Camera view is very dark. This is a known issue in the device driver
  • Object Recognition apps may crash when loading Device Databases containing large Object Targets on low end devices.
  • On Android, you can work around this issue by adding android:largeHeap=”true” to the AndroidManifest.xml of your application. 
  • Minimum supported Unity version has changed to 4.5
  • Apply Dataset Properties is not supported for Object Target datasets.

Unity

  • The Unity device filter, in build settings, must be set to ARM v7 in Unity 4.6 or later to support ARM emulation on x86 device.
  • 64 bit devices running iOS 8: Unity apps using Enabling Extended Tracking with Object Recognition will experience a reduced frame rate  


Vuforia SDK  v3.0.9 

 
Bugfixes / Features

All SDKs

  • Fixed a bug where visual artifacts were distorting the video background on some devices
  • Fixed a bug with Extended Tracking not reseting correctly causing it to fail when the environment changes significantly

Unity

  • Fixed a bug where datasets could not be loaded from an OBB file.
Known Issues  

All SDKs

  • The Vuforia SDK only supports OpenGL ES 2.0.
  • The minimum supported Android platform version is Android 2.3 (Gingerbread).
  • Extended Tracking and Smart Terrain are only supported on multi-core devices.
  • Extended Tracking and Smart Terrain may fail during very fast motions.
  • Extended Tracking may fail in very plain environments (e.g. blank wall)
  • Extended Tracking may result in more jitter and a visual snap back on the augmentation when the target re-enters the field of view
  • Extended Tracking and Smart Terrain may fail in non-static environments (i.e. large motions or environment changes)
  • Extended Tracking can only be used for a single UDT target at a time
  • Extended Tracking targets must have mutually consistent sizes/scales if they are used together in the same scene
  • Extended Tracking may fail to relocalize when the phone is moved in pure rotational motion around the scene.
  • Extended Tracking and Smart Terrain are not supported on front cameras
  • Smart Terrain is only enabled in the Vuforia Unity extension
  • Smart Terrain and Extended Tracking cannot be used at the same time. You need to stop SmartTerrain before using Extended Tracking and vice versa.
  • Smart Terrain may occasionally reconstruct flat objects (such as a drink coaster) as a 3D object and return it as a Prop
  • Smart Terrain may occasionally reconstruct two nearby objects as a single object
  • Smart Terrain may fail to reconstruct the ground plane and objects if the initialization target is placed on the border of the primary surface (e.g. table). The initialization target needs to be placed well within the primary surface to avoid this issue.
  • Smart Terrain may in some occasions wrongly reconstruct the primary surface such that it appears to extend into the background. This may happen if the primary surface and background are visually not distinguishable (e.g. white table in front of a white wall)
  • Smart Terrain may not work well on play surfaces that are non-uniformly textured (E.g. large tiled floor with different colored tiles)
  • Smart Terrain cannot be initialized with User Defined Targets
  • Text recognition may occasionally mis-detect a word or detect a partial word
  • Time to recognize a word depends on number of words and the length of words. Specifically, Text Recognition may take longer than 2 seconds on some single core devices, as well as on some devices when more than 3 words are in the recognition area or for words that are longer than 24 letters.
  • Text recognition may fail to track text well if the page of text and the background are not on the same plane.
  • Low contrast targets may not be detected well in highly cluttered backgrounds
  • Text recognition does not support the front camera.
  • Enabling additional words in the Text recognition feature may slightly increase recognition time
  • TextRecognition may not work if another Tracker is used before.
  • Cloud recognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • The Vuforia SDK only support armeabi-v7a devices. Devices with no VFP unit or armv6 are not supported.
  • Not all focus modes are supported on all devices
  • On certain older devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and  may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.
  • Multiple simultaneous Multi Targets are not supported
  • A known memory leak exists in the CameraDevice class. For each cycle of calling CameraDevice.start() / CameraDevice.stop() memory consumption grows. Note that this memory is released if CameraDevice.deinit() is called immediately after calling stop(), which is the recommended way of using the camera to also ensure other applications can use the camera.
 


Vuforia SDK v3.0.6 

 
Bugfixes / Features

Unity

  • Fixed a bug where the video background was wrongly oriented and distorted in iOS apps built with Unity 4.5.1.

 

Vuforia SDK  v3.0.5 

 
Bugfixes / Features

All SDKs

  • Introduced the SmartTerrain feature: ImageTargets, MultiTargets and CylinderTargets can now be used to reconstruct the surface and objects that are placed around them. Please see the Vuforia sample applications and developer guide for details. Note that Smart Terrain is available only through the Vuforia Unity Extension
  • Added support for high resolution video background rendering of up to 720p on supported high end devices.
  • Consolidated multiple definitions of STORAGE_TYPE in both DataSet and WordList classes into a single definition QCAR::STORAGE_TYPE. QCAR::DataSet::STORAGE_TYPE and QCAR::WordList::STORAGE_TYPE are now deprecated.
  • Fixed a crash when calling QCAR::ImageTracker::resetExtendedTracking(). The API has changed in behavior as the ImageTracker now needs to be stopped first before calling resetExtendedTracking()
  • Fixed a crash caused by repeatedly initializing and deinitializing the CameraDevice
  • Fixed an issue where TextReco stops working after calling selectVideoMode()

Unity

  • Added support to load device databases from APK Expansion files on Android. The Unity Extension will now associate targets authored in the scene using the dataset name, not the full path, so that databases can be loaded from a .obb file. The DatasetLoadBehaviour will also try to load databases in the StreamingAssets folder from common APK extension directories automatically.
  • Moved Dataset.StorageType enum to QCARUnity.StorageType to correspond to the QCAR::STORAGE_TYPE changes in the native API. Dataset.StorageType is now deprecated.
  • Fixed a crash when calling ImageTracker.ResetExtendedTracking(). The API has changed in behavior as the ImageTracker now needs to be stopped first before calling ResetExtendedTracking()
  • Fixed a crash caused by repeatedly initializing and deinitializing the CameraDevice
  • Fixed an issue where TextReco stops working after calling SelectVideoMode()
  • Fixed multiple minor memory leaks
  • Fixed an issue where an application could get into an inconsistent state if the ARCamera was disabled before pausing and resuming the application.
  • Fixed an issue with distorted web cam views in Mac Play Mode
  • QCARBehaviour.GetVideoMode() has been replaced with CameraDevice.Instance.GetVideoMode()
  • Some previously public classes and methods of the Unity Extension that were never part of the official public API have been made internal with this release.
  • WebCamBehaviour.StartCamera and StopCamera have been removed.
  • Use CameraDevice.Start() and CameraDevice.Stop() instead.
  • Added Tracker.IsActive API
  • Added XMLdoc API documentation for compiled Unity Extension dlls that can be used in Visual Studio (Intellisense) or MonoDevelop
Known Issues  

All SDKs

 
  • The Vuforia SDK only supports OpenGL ES 2.0.
  • The minimum supported Android platform version is Android 2.3 (Gingerbread).
  • Extended Tracking and Smart Terrain are only supported on multi-core devices.
  • Extended Tracking and Smart Terrain may fail during very fast motions.
  • Extended Tracking may fail in very plain environments (e.g. blank wall)
  • Extended Tracking may result in more jitter and a visual snap back on the augmentation when the target re-enters the field of view
  • Extended Tracking and Smart Terrain may fail in non-static environments (i.e. large motions or environment changes)
  • Extended Tracking can only be used for a single UDT target at a time
  • Extended Tracking targets must have mutually consistent sizes/scales if they are used together in the same scene
  • Extended Tracking may fail to relocalize when the phone is moved in pure rotational motion around the scene.
  • Extended Tracking and Smart Terrain are not supported on front cameras
  • Smart Terrain is only enabled in the Vuforia Unity extension
  • Smart Terrain and Extended Tracking cannot be used at the same time. You need to stop SmartTerrain before using Extended Tracking and vice versa.
  • Smart Terrain may occasionally reconstruct flat objects (such as a drink coaster) as a 3D object and return it as a Prop
  • Smart Terrain may occasionally reconstruct two nearby objects as a single object
  • Smart Terrain may fail to reconstruct the ground plane and objects if the initialization target is placed on the border of the primary surface (e.g. table). The initialization target needs to be placed well within the primary surface to avoid this issue.
  • Smart Terrain may in some occasions wrongly reconstruct the primary surface such that it appears to extend into the background. This may happen if the primary surface and background are visually not distinguishable (e.g. white table in front of a white wall)
  • Smart Terrain may not work well on play surfaces that are non-uniformly textured (E.g. large tiled floor with different colored tiles)
  • Smart Terrain cannot be initialized with User Defined Targets
  • Text recognition may occasionally mis-detect a word or detect a partial word
  • Time to recognize a word depends on number of words and the length of words. Specifically, Text Recognition may take longer than 2 seconds on some single core devices, as well as on some devices when more than 3 words are in the recognition area or for words that are longer than 24 letters.
  • Text recognition may fail to track text well if the page of text and the background are not on the same plane.
  • Low contrast targets may not be detected well in highly cluttered backgrounds
  • Text recognition does not support the front camera.
  • Enabling additional words in the Text recognition feature may slightly increase recognition time
  • TextRecognition may not work if another Tracker is used before.
  • Cloud recognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • The Vuforia SDK only support armeabi-v7a devices. Devices with no VFP unit or armv6 are not supported.
  • Not all focus modes are supported on all devices
  • On certain older devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and  may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.
  • Multiple simultaneous Multi Targets are not supported
  • A known memory leak exists in the CameraDevice class. For each cycle of calling CameraDevice.start() / CameraDevice.stop() memory consumption grows. Note that this memory is released if CameraDevice.deinit() is called immediately after calling stop(), which is the recommended way of using the camera to also ensure other applications can use the camera.
 


Vuforia SDK  v2.8.7

 
Bugfixes / Features

All SDKs

 
  • Added support for Extended Tracking: ImageTargets, MultiTargets and CylinderTargets can now continue to be tracked even when the  target is not visible. Please see the Vuforia sample applications and developer guide for details.
  • Added a new Java language binding to enable Vuforia application development without the NDK. See the Vuforia sample applications for details. Support for C++ APIs continues.
  • The enum based type system has been replaced with the concept of the QCAR::Type class. Please refer to the 2.8 migration guide how to upgrade your application.
  • The release artifact names have changed: libQCAR.so/libQCAR.jar have been renamed respectively to libVuforia.so/libVuforia.jar.
  • Fixed crash on Samsung Galaxy Nexus when an ImageTarget is detected.
  • Fixed an issue in TextReco where QCAR::WordList::addWordU and removeWordU failed to add/remove open compound words.
  • Fixed an issue in TextReco where invalid characters could be added to the word list.
  • Fixed an issue where MultiTarget::getPartOffset incorrectly returned false although the index was valid.
Known Issues

All SDKs

  • The Vuforia SDK only supports OpenGL ES 2.0.
  • The minimum supported Android platform version is Android 2.3 (Gingerbread).
  • Extended Tracking is only supported on multi-core devices.
  • Extended Tracking may fail during very fast motions.
  • Extended Tracking may fail in very plain environments (e.g. blank wall)
  • Extended Tracking may result in more Jitter and a visual snap back on the augmentation when the target re-enters the field of view
  • Extended Tracking may fail in non-static environments (i.e. large motions or environment changes)
  • Extended Tracking can only be used for a single UDT target at a time
  • Extended Tracking targets must have mutually consistent sizes/scales if they are used together in the same scene
  • Extended Tracking may fail to relocalize when the phone is moved in pure rotational motion around the scene.
  • Extended Tracking not supported on front cameras
  • Text recognition may occasionally mis-detect a word or detect a partial word
  • Time to recognize a word depends on number of words and the length of words. Specifically, Text Recognition may take longer than 2 seconds on some single core devices, as well as on some devices when more than 3 words are in the recognition area or for words that are longer than 24 letters.
  • Text recognition may fail to track text well if the page of text and the background are not on the same plane.
  • Low contrast targets may not be detected well in very cluttered backgrounds
  • Text recognition does not support the front camera.
  • Enabling additional words in the Text recognition feature may slightly increase recognition time
  • Cloud recognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • The Vuforia SDK does not support devices that have no VFP unit. Only devices with armv6 + VFP and above are supported.
  • Not all focus modes are supported on all devices
  • On certain older devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and  may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.

Unity

  • Added support for Unity 4.3 and later
  • Vuforia Play Mode can now be used with the free Unity license in Unity 4.3 or later.
  • The enum based type system has been replaced with the usage of C# generics. Please refer to the 2.8 migration guide
  • The names of the plugin libraries have changed: libQCAR.so/libQCAR.jar have been renamed respectively to libVuforia.so/libVuforia.jar (no changes required for Unity applications).
  • Fixed crash on Samsung Galaxy Nexus when an ImageTarget is detected.
  • The Java source code for the QCARPlayerNativeActivity can be found in the Assets/Plugins/Android/src folder. It can be modified to initialize other Unity plugins as well. Modifications need to be compiled to replace Assets/Plugins/Android/QCARUnityPlayer.jar.
  • Fixed a crash when VirtualButtons are used in more than one scene.
  • Fixed an issue in TextReco where WordList.AddWord and RemoveWord failed to add/remove open compound words.
  • Fixed an issue in TextReco where invalid characters could be added to the word list.
  • Fixed an issue where Vuforia was clamping the rendering frame rate to 30fps even after deinitialization when the VideoBackgroundConfig is set to synchronous.
  • A SetSize() method has been added to the Marker class.
  • Fixed a bug where ImageTarget.GetSize() would return a wrong aspect ratio for cloud reco targets.
  • Fixed a bug where “Screen position out of view frustum” errors were logged in Play Mode.

Vuforia SDK  v2.6.10

 
Bugfixes / Features

All SDKs

 
  • Added support for Cylinder targets
  • Text recognition - Up to 10,000 additional words may be added to the default word list
  • Text recognition speed has been improved significantly especially for mid and low tier devices
  • Vuforia no longer collects the hashed MAC address or other permanent device-based identifiers for purposes of identifying an end user’s device. A hashed app runtime timestamp is now used as an app identifier
Known Issues

All SDKs

  • The minimum supported Android platform version is Android 2.3 (Gingerbread).
  • Text recognition may occasionally mis-detect a word or detect a partial word
  • Time to recognize a word depends on number of words and the length of words. Specifically, Text Recognition may take longer than 2 seconds on some single core devices, as well as on some devices when more than 3 words are in the recognition area or for words that are longer than 24 letters.
  • Text recognition may fail to track text well if the page of text and the background are not on the same plane.
  • Low contrast targets may not be detected well in very cluttered backgrounds
  • Text recognition does not supp ort the front camera.
  • Enabling additional words in the Text recognition feature may slightly increase recognition time
  • Cloud recognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • The Vuforia SDK does not support devices that have no VFP unit. Only devices with armv6 + VFP and above are supported.
  • Not all focus modes are supported on all devices
  • On certain devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and  may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.

Vuforia SDK  v2.6.7

Bugfixes / Features

Unity

 
  • Added support for sharing Vuforia objects between multiple Unity scenes without re-loading device databases
  • Fixed an issue where the orientation setting for Android was not applied correctly when using Unity 4.2
  • Vuforia no longer collects the hashed MAC address or other permanent device-based identifiers for purposes of identifying an end user’s device. A hashed app runtime timestamp is now used as an app identifier


Vuforia SDK  v2.5.8

Bugfixes / Features

Unity

 
  • New TextRecognition feature introduced with support for English – please see Dev guide for details
  • Tracking improvements especially for low-feature targets such as logos or line art
  • The world center modes have been renamed to be more descriptive
  • MultiTarget textures are now rendered in non-webcam Play Mode.
  • MultiTargets are now created without colliders by default
  • The size of an ImageTarget can now be explicitly controlled in the inspector window
  • Fixed a bug where Camera.Instance.GetCameraImage was not updated after restarting Vuforia
  • Fixed a bug where Adding the QCARBehaviour and WebCamBehaviour scripts to a camera object would result in a crash in Play Mode


Vuforia SDK v2.5.7

 
Bugfixes / Features

All SDKs

 
  • New TextRecognition feature introduced with support for English – please see Dev Guide for details
  • Tracking improvements especially for low-feature targets such as logos or line art .
Known Issues

All SDKs

  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.
  • TextRecognition may occasionally misdetect a word or detect a partial word
  • TextRecognition is CPU intensive and time to reco may depend on number of words and length of words. Specifically, TextRecognition may take longer than 2 seconds on some single core devices, as well as on some devices when more than 3 words are in the recognition area or for words that are longer than 24 letters.
  • TextRecognition: unloading the word list may take several seconds
  • TextRecognition may fail to track text well if the page of text and the background are not on the same plane.
  • Low contrast targets may not be detected well in very cluttered backgrounds
  • TextRecognition does not support the front camera.
  • Enabling additional custom words in the TextFeature may slightly increase recognition time
  • CloudRecognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • The Vuforia SDK does not support devices that have no VFP unit. Only devices with armv6 + VFP and above are supported.
  • Not all focus modes are supported on all devices
  • On certain devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs


Vuforia SDK v2.0.32

 
Bugfixes / Features

All SDKs

 
  • No longer accesses the iOS unique device identifier (UDID).  Uses the MAC to generate a device identifier


Vuforia SDK v2.0.31

 
Bugfixes / Features

All SDKs

  • Fixed a bug where target detection performance was reduced in large datasets.
  • Fixed an issue where the tracked target pose was drifting on some devices.New features include

Unity

  • Added support for Unity 4.1 (beta) on iOS.
  • Fixed a bug where device database loading from absolute paths failed in Play Mode.
  • Fixed an issue where the StateManager API did not return MarkerBehaviours created at runtime.
  • Fixed an issue where TrackableBehaviours created at runtime would be initialized to default values when selected in the hierarchy during Play Mode.
Known Issues

All SDKs

  • On Windows 8, the installer needs to be run in Windows 7 compatibility mode:
1. Right-mouse click on the installer executable
2. Select Properites
3. Select the Compatibility tab
4. Check the box for "Run this program in compatibility mode for:", and select "Windows 7"
  • CloudRecognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • Virtual button presses are detected unreliably on the HTC Aria
  • Stability issues are observed on the Motorola Defy
  • The Vuforia SDK does not support devices that have no VFP unit. Only devices with armv6 + VFP and above are supported.
  • The MACRO and INFINITY focus modes do not work on the Galaxy Nexus
  • On certain devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • “CPU Pegged” shows up randomly and causes a crash on certain devices
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.
 

Vuforia SDK v2.0.30

 
Bugfixes / Features

All SDKs

  • New features include User-defined Targets and Cloud Recognition.
  • Introduced support for using the front camera.
  • Introduced support for multiple active datasets.
  • Improved tracking robustness.
  • Removed sample applications from Vuforia SDK Installation package. Sample applications are now a separate download.

Unity

  • Introduced Play Mode support in Unity for faster application development and testing within the  Unity Editor.
  • Introduced support for using the front camera.
  • Introduced support for multiple active datasets.
  • The Vuforia AR Extension for Unity now contains the Plugins for both Android and iOS.
  • Improved tracking robustness.
  • Removed sample applications from Vuforia Extension for Unity Installation package. Sample applications are now a separate download.
Known Issues

All SDKs

  • On Windows 8, the installer needs to be run in Windows 7 compatibility mode:
  1. Right-mouse click on the installer executable
  2. Select Properites
  3. Select the Compatibility tab
  4. Check the box for "Run this program in compatibility mode for:", and select "Windows 7"
  • CloudRecognition requires the device time to be accurately set (the tolerance is 5 minutes). If the device time is off, the SDK will return an ‘TIMESTAMP_OUT_OF_RANGE’ error. However due to a limitation in the Android SSL stack if the time is significantly off (more than a year), then the above error is not reported. Instead the error ‘SERVICE_NOT_AVAILABLE’ is reported.
  • Virtual button presses are detected unreliably on the HTC Aria
  • Stability issues are observed on the Motorola Defy
  • The Vuforia SDK does not support devices that have no VFP unit. Only devices with armv6 + VFP and above are supported.
  • The MACRO and INFINITY focus modes do not work on the Galaxy Nexus
  • On certain devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • “CPU Pegged” shows up randomly and causes a crash on certain devices
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • Stability issues have been observed when the “face unlock” feature is used while a Vuforia application is running.
 

Vuforia SDK  v1.5.9

 
Bugfixes / Features

All SDKs

  • Vuforia branding introduced to SDK
  • Fixed memory leaks
  • Fixed issue where frame and augmentation of previous run is visible for a few frames
  • Added OcclusionManagement sample app showing how to create an occlusion effect using the video texture feature
  • The android:minSdkVersion has been updated to 8 – this means QCAR can only be installed on devices running Android 2.2 Froyo or above. Android 2.1 Éclair is no longer supported
  • Detection and tracking speed has been improved
  • Added new CONTINUOUS_AUTOFOCUS mode to QCAR and exposed it in samples via GUI
  • Note:Continuous auto focus is only available on select Android devices, e.g. Galaxy Nexus S, Pantech Vega Racer
  • Fixed an issue where the reordering of trackables in the dataset XML file causes tracking to fail
  • Fixed an issue where the registering for YUV pixel format images fails
  • Added support for ICS and Honeycomb devices.
Known Issues

All SDKs

  • The MACRO and INFINITY focus modes don’t work on the Galaxy Nexus (driver seems to ignore this setting)
  • The CONTINUOUS_AUTOFOCUS mode stops working after triggering a single autofocus action and can only be activated again when you stop and restart the camera (e.g. by pausing and resuming a QCAR sample application)
  • On certain devices, the video preview rendering and the rendering layer order may be mismatched after another application overlays the QCAR rendering window. Apps may optionally destroy the Android activity with a call to Activity.finish() in your main Activity‟s onPause() and may reinitialize it onResume(). This no longer happens on devices running Honeycomb or above.
  • “CPU Pegged” issue is showing up randomly and causing a crash on certain devices
  • The Motorola Flipout is currently not supported
  • On rare occasions under very low light conditions, FrameMarkers may detect the same marker twice with different IDs
  • The BackgroundTextureAccess and OcclusionManagement apps support only landscape-left orientation
  • On ICS devices a black screen has been observed after unlocking the phone with the “face unlock” feature

Vuforia SDK v1.0.10

 
Bugfixes / Features

All SDKs

  • Vuforia branding introduced to the Unity extension
  • Added OcclusionManagement sample app showing how to create an occlusion effect using the video texture feature
  • Improvements to VideoTexture access and updates to sample
  • Added support for building applications in Auto rotation mode
  • Removed dataset rename dialog – extension now requires 1.5 compatible datasets.
  • The android:minSdkVersion has been updated to 8 – this means QCAR can only be installed on devices running Android 2.2 Froyo or above. Support for Android 2.1 Éclair has been deprecated
  • Added new CONTINUOUS_AUTOFOCUS mode to QCAR and exposed it in samples via GUI
  • Note: Continuous auto focus is only available on select Android devices, e.g. Galaxy Nexus S, Pantech Vega Racer
  • Fixed several bugs and memory leaks
  • Added support for ICS and Honeycomb devices