Release 2.7

MirageXR release notes 2.7
1 July 2024

Featured:

• Animations. We added basic support for animations, so if a 3D model from Sketchfab has an animation, it will be automatically played.
• Shadows. We added support for AR shadows so that all augmentations now cast a shadow onto the spatial map. This helps with getting a better feel for where objects are in 3D space.
• Diamond visibility. The task station diamond can now be switched off, making it easier to work with image targets like for ‘magic book’ overlays.

Added:

• We added support for the audio augmentation’s file picker for MP3, so that MP3-compressed recordings can now be added and played. This helps keep the file size small for the learning experience model archive.
• We added a vertical scrollbar to the Home screen on mobile UI.

Changed:

• Tutorials: We changed how we highlight UI elements in the built-in tutorials.
• We updated the editor panels for the action augmentation to the new design on mobile.

Fixed:

• We fixed a number parsing error preventing colors from loading correctly if the user is in a different country/locale that uses commas as decimal separators instead of floating point.
• We updated the library user interface on mobile and fixed a bug that would make the library dialogue disappear.

Enterprise:

• We created an AI service to refactor the character model augmentation and prepare for the virtual instructor augmentation, as a wrapper for our own back-end LLM RAG server.

Developer:

• We migrated to Universal Rendering Pipeline (URP), which required a cascade of changes, including updating of materials, adding gltfast shader materials to the list of included materials, and cleaning up some of the graphic settings to reduce the build size of the project (which had grown, post update, to over 400MB). The way URP handles camera cascades required a temporary camera rendering fix.
• We cleaned up dead code, and prepared for more. This includes cleaning out the old mobile dialogue prefabs, and moving all code for models to the correspondingaccording ContentType/3Dmodel folder.
• We refactored the TutorialManager, saving tutorial data now as serialized JSON to make it easier to update and/ create new tutorials.
• We updated the Android manifest to require the internet access capability.
• We updated iOS builds to SDK 17.
• We cleared out old Navmesh components and instead added AI navigation package support.
• We excluded MonoSupport.dll excluded in assembly and removed mcs.rsp remnants that should not be in use anymore.
• We added dependencies hunter to help with identification of unused assets, and started to hunt down and remove dead code.
• We added base64 encoding for the AIManager to support, e.g., Greek characters.
• We removed DotNetWinRT from nuget packages config.

Release 2.6

Change Log

Featured:

• Improved calibration. We improved exception handling for merged planes in the calibration process
• New action editor. We redesigned the user interface for the action augmentation editor
• chatGPT support for characters. Character models now have chatGPT support and can store a prompt

Added:

• We added file upload via file picker for audio augmentations, creating an alternative to recordings
• We now support editing and displaying of captions for the image augmentation to improve accessibility, adding a new field to the data model accordin

Changed:

• We redesigned the user interface for the audio editor
• We remove the second (turquoise) task-station diamond indicating the position of the next step
• We removed the lines leading from the aura to the task station
• We improved the dialogue messages when publishing to Moodle
• We added rotation handles to the action augmentation
• Character augmentations can now also be placed onto 3D objects, and no longer have to be placed directly onto the spatial map, remediating a "GetRemainingDistance" error that would be flagged every frame otherwise
• We renamed the ‘apply’ button to ‘done’ in the calibration dialogue to be more precise

Fixed:

• We fixed bugs in the video augmentation that would overwrite the currently selected video when cancelling the camera dialogue
• We fixed both create and view tutorials, including fixing the way the tutorial activity for the viewing tutorial is downloaded from a resource folder
• We fixed the help tutorial for publishing activities
• We fixed an error with deregistering event handlers during application shutdown
• We fixed a bug where broken 3D models from Sketchfab would trigger an error
• We fixed a bug where a Moodle upload of an existing resource would result in an error when asking the user whether to overwrite or clone, now issuing only a warning
• We fixed a bug in the Sketchfab search, which would produce errors when you search a second time
• We fixed a bug in the label augmentation where default values for ‘turn to user’ would not be set, thus only storing the value correctly when the toggle switch would be turned on/off twice
• We fixed the materials for some character models, which seemed not to be set to the right rendering pipeline
• We fixed a bug in character models complaining about a missing FollowMeToggle

Enterprise:

• The MoodleUrl can now be set from the brand configuration file unless there already is a user preference set overriding which URL to use
• We excluded Sentry settings and sentry client settings via .gitignore so that auth tokens and DSN values are not stored on GitHub

Developer:

• We migrated the project to Unity 2022 LTS (2022.3.7f1)
• We decommissioned the IBM Watson assistant service and replaced it with openAI's chatGPT
• We upgraded online error handling to the full Sentry package, improving breadcrumbs and the stack trace, as well as adding session and user info for authenticated users
• We implemented a slider for verbosity levels for in-app bug tracking
• We reduced StyleCop to documentation checks in the MirageXR assembly, and disabled a large portion of warnings on external projects and packages
• We removed the dummy video in the root directory from the video augmentation, which was used to simulate video recording in-editor on Hololens before we introduced the file picker upload
• We fixed the calibration PlayMode test and several others that were suffering from multiple audio listeners being created in the scene
• We improved the test and build pipelines for Android and Windows
• We renamed the editor folders of all content types (to ‘editors’), to avoid confusion with the reserved folder name of Unity
• We created an openAI manager, primarily as interface for character models, but with further reaching plans
• We fixed the build for Hololens 2

Release 2.5

Change Log

Featured:

• Markerless calibration. We redesigned calibration, adding markerless manual placement as the new alternative to precision calibration with the calibration marker. The ‘world origin’ will now also be remembered across activities, reducing the need to calibrate overall. As part of the overall improvement, we have also added a tracking tool to alert the user when tracking is lost.
• 3D model collections. We redesigned the editor panel for 3D model augmentations, adding quick access to already downloaded models and implementing a model library section (with food and construction tools as the first categories available).
• Online error reporting. We track errors now online, so we can diagnose problems better and react more quickly.

Added:

• The Ghost augmentation now additionally records finger movement on HoloLens 2 into the data model, but does not yet display it.
• We improved accessibility by implementing an experimental feature for image captions.

Changed:

• Character models can now react to control commands parsed from any dialogue text returned from the AI (%%trigger%% and %%trigger:step=5%%).
• The editor 3D grid and the snap-to-grid placement support are now active by default.
• We redesigned the user interface of the label editor.
• We redesigned the user interface for the audio augmentation editor panel.
• We refactored the autosave functionality, which stores data locally immediately, which had shown some erratic behavior, sometimes missing saving changes.

Fixed:

• We improved handling of 3D models to prevent them from slipping under the floor level. On rare occasions, as a result of calibration or tracking errors, models may still end up under the floor level, but this can now be fixed using the markerless calibration.
• We fixed a bug with re-editing image augmentations, where confirming without changing the image would produce an exception.
• The handle for rotating an action glyph (left and right) is now switched on again.
• Activity title and description can no longer be edited in view mode.
• We removed the zombie ‘home’ button on the aura from the mobile UI.
• Values for gaze triggers are now properly saved for Action augmentations.
• We refactored the Sketchfab communication, reducing debug log output and phasing out reporting of hard exceptions from expired credentials.
• We fixed a few bugs for the label augmentation: Minimizing the edit panel no longer makes the whole menu disappear, re-editing of the step target now recovers the stored value correctly, and pre-selects it in the list; labels of old activities are imported correctly again, and, additionally, label text is now displayed in the list of augmentations, making it easier to identify the right item.

Enterprise:

• We set up sponsorship support for our GitHub project – please donate large and plenty :)
• We fixed the xAPI implementation, also modifying the underlying i5 service, which prevented LRS communication from running, also improving the overall error handling (e.g. LRS selection under profile triggered an error).
• We fixed the sensor communication framework (VestService.cs and vestconfig.json), which failed to find its config file, and set http://test.mosquitto.org:1883 as default service, which can be used to test communication with sensors from within activities.
• We improved the update and clone dialogue messages for the Moodle cloud uploads.
• We fixed a bug, where login with the Moodle username (instead of the registered email) would allow logging in, but prevent publishing. Users now need to log in with the registered email address.
• Activities that exist only locally (and are not uploaded to the cloud) now no longer try to increase the Moodle hit counter, previously yielding an error.
• We added the possibility to configure the calibration marker to the BrandManager.

Developer:

• We implemented a new Exception Manager that captures exceptions globally and logs them live to Sentry.io for improved online error reporting.
• We reorganized the folder structure and location of project files to be more logical.
• We improved debug logging, now supporting the setting of different levels of verbosity.
• We changed the glTF plugin implementation to gltfast, which also resolves issues with animations.
• xAPI calls are now suppressed when running in the editor, to avoid spamming the record store, while still maintaining their execution for testing purposes.
• To ease in-editor testing, we now assume a random world-origin position during play-mode. This makes it easier to spot problems in restoring positions following calibration.
• We introduced one-button dialogues.
• We switched the CI pipeline to scheduled builds, to prevent disk space on the container from overflowing quickly in periods of heightened commits.
• We added several new wiki pages and updated existing ones (Exception Manager, Experience Manager, Events, Character Model, Compiling MirageXR, File Organisation, UI 2.0).

Release 2.4

Change Log

Featured:

• Improved label. Label augmentations can now be customized with changing font sizes and different background and font color.
• Augmentation lock. Augmentations can now be locked, fixing them in place while in edit mode, avoiding accidentally moving them.
• Grid improvement. The placement layout grid can now be switched to invisible, hiding the grid, while still snapping to grid positions.

Added:

• We added a logout button for Sketchfab for the worldspace UI.

Changed:

• Character model triggers can now trigger any step by using the “%%trigger:step=X%%” (where X is the step number to jump to) keyword in a Watson output.
• We cleaned up code for glyph, image, and audio augmentations.

Fixed:

• We fixed a bug where Sketchfab models containing the “/” character would cause the app to crash.
• We removed world space UI remnants from mobile builds.
• We fixed a bug for the custom 3D models in the Pick and place augmentation.
• We fixed a bug with triggers, causing the jump to action step to interpret the wrong step index.

Enterprise:

• No changes

Developer:

• The default unity Debug.Log() is now replaced by a new custom logging system which filters messages based on their importance level. The changes to the UI for setting the desired filter level will be added in a later version. For now, it is possible to change the filter level using Debug.MinimumLogLevel.

Release 2.3

Change Log

Featured:

• Grid. We created a new visual grid system and a snap-to-grid feature which significantly improves the usability of augmentation placement. The drop position is indicated with a 'ghost preview' of the currently selected augmentation. Grid spacing, default rotation angles, and default resizing steps can now be configured.
• Better editors. We redesigned the editors of the image, video, audio, label, and model augmentations.
• HoloLens is back. We updated the world space implementation to bring the HoloLens 2 application back in line with current developments.

Added:

• We rewrote the calibration routines for both edit mode and view mode, adding a floor manager to allow setting the floor plane.
• We added gallery upload functionality on mobiles for the step marker feature of task stations and for the image/video augmentations.
• We added a player tutorial for mobiles and smart glasses which uses a tutorial activity that is written out internally every time the tutorial is started to ensure the viewer tutorial is not modified by the user.
• The marker feature of task stations now shows up with a preview thumbnail button in view mode.
• We added auto-saving of the activity title.

Changed:

• We simplified the functionality of the pick and place augmentation, added a trigger for incorrect placement, and improved its manipulation when using a 3D model instead of the default arrow.
• We added auto-detection of orientation for images.
• We removed the 'Search' button from the bottom panel on all screens except the 'Home' screen.
• The ARETE learning record store has now been taken offline, so we removed this LRS option. Please switch to WEKIT learning record store.

Fixed:

• We fixed the bug that prevented re-editing of effect augmentations.
• The bottom panel now always shows the current step, and we fixed a bug here throwing an error when minimizing the curtain.
• We made SketchFab authentication a tad more robust.
• The calibration marker is now no longer orientation sensitive (hanging upright vs laying flat).
• Deleting an action step now removes it from the workplace model.

Enterprise:

• We added new settings to BrandManager and BrandConfiguration, now allowing for the activation and deactivation of augmentation types, as well as for setting the calibration marker.

Developer:

• We retired Vuforia on mobile platforms and replaced it with ARfoundation.
• We updated OpenXR and Vuforia in the HoloLens implementation.
• We fixed the action linux and windows workflows to support LFS.
• We added a debug log window to the smart glasses implementation (click person icon, click "{}").
• We added tutorial manager facilities for 'world space' hints on mobile.
• We now have a new bottom multiline dialog with toggles.
• Long tap on mobiles now brings up the context menu (see new model augmentation editor).

Release 2.2

Change Log

Featured:

• App optimization. We have optimized the application. Now it runs faster and heats up the phone less.
• Gallery images and videos. The image and video augmentation as well as the step marker now allow adding files directly from existing photo and video galleries on the device.
• New Search. The search functionality has been improved, integrating search directly into the home screen.

Added:

• New search screen has been added.
• Horizontal scrolling has been added to the StepList View.

Changed:

• The operation of the bottom navigation bar has been changed.
• Activity settings have been updated.

Fixed:

• We fixed bugs related to Image Marker Augmentation.
• Content in records no longer appears upside down.
• We fixed bugs related to Pick & Place.
• The spawn position has been corrected.

Developer:

• We fixed the handling of screen press events.
• Bottom multiline dialogue with toggles has been added.
• Vuforia has been removed from the mobile version of the app.
• iPad: the display of the debug console has been fixed.
• Application crashes due to Vuforia initialization have been fixed.
• The work with 3D models has been optimized.
• Playmode tests have been updated.
• Debug log messages have been improved.

Release 2.1

Change Log

Featured:

• Automatic saving. We improved auto-saving of activities.
• Calibration marker position. The calibration marker can now be moved and registered by the app in a new location, while editing an activity.
• Debug console. We added a debug console for mobile devices to help diagnose complex, device-specific content or functionality issues.

Added:

• New warning message is displayed when selecting public upload from the publishing menu.
• We brought back the Sketchfab web-based single-sign on.
• “Get Calibration Image” in the calibration tab now opens a sharing dialogue.
• We added a feature to capture the new position of the calibration image which is useful when the user moves the calibration marker but wants to keep the position of AR content.

Changed:

• We made minor changes to improve the uniformity of the mobile UI.
• Changing an activities name or description will now auto-save the activity.
• Changing an action step’s name or description will now auto-save the activity.
• Changing an augmentation’s ‘keep alive’ settings will now auto-save the activity.
• The VFX augmentation is now referred to as ‘Effects’.
• We renamed “Activity settings” to “Publish”.
• Locate now targets the Pick and place arrow rather than the sphere.

Fixed:

• Logging out is now persisted on application re-launch.
• Recalibration button is no longer visible when minimizing the calibration menu.
• Publish toggle is now inactive when neither local nor cloud upload is selected.
• Activity settings are remembered when re-launching the app.
• Illegal characters of filenames (\ / : * ? " < > |) of 3D models no longer prevent their use.
• We fixed a bug changing the content type of augmentations if their name contained the name of another augmentation type.
• The pick and place ‘reset on action step’ option now works correctly.
• Delete augmentation now works.
• Recalibration now deactivates the active action step before starting the first action step.

Enterprise:

• The privacy policy is now linked from the profile page.

Developer:

• We changed the predicate for the visual effects augmentation from ‘vfx’ to ‘effects’.
• We added a verbose logging system to improve debug outputs.
• We added a debug console to mobile builds for listing log messages at run-time to improve device specific debugging.

Release 2.0.1

Change Log

Featured

• New and improved mobile user interface. We completely redesigned the concept of user interaction and dialogues on the mobile platforms. The app now has a significantly improved usability and user experience, and many new features.
• Context help. We introduced a new dynamic context help system, with a comprehensive set of help dialogues for anything we could think of.
• Interactive tutorial. We added a facility for step-by-step interactive tutorials and added a tutorial introducing the user to the editing functionality.

Added

• Onboarding: We added onboarding swipe-thru slides on first app launch, explaining the key concepts of mirageXR with text and animations.
• Bottom bar: We added a new bottom tab bar with icons for quickly switching between activity stream, profile, search, and the new dynamic context help.
• Quick edit toggle: We introduced a new edit toggle button in the top right corner of the mobile user interface for quickly switching from viewing to editing.
• Collapsable main panel: Users can now minimize the main menu, providing ‘prev’/’next’ quick navigation buttons in the collapsed view for moving forwards and backwards between action steps. Especially on smaller screen phones, this frees valuable screen real estate up for a less cluttered view of the activity.
• Model augmentation: We added boundary box handles for the model augmentation as alternative to the direct manipulation with pinch and rotate.
• Pick & place augmentation: We added trigger functionality to allow jumping to the specified step if the pick object is placed in the correct target location, and we added reset options for placement.
• Pick & place augmentation: We added sound effects for the pick & place augmentation (for correct and incorrect placement).
• Character models: We added trigger functionality to character models, moving on to the next action step, when audio or animation are finished playing (whichever takes longer).
• Character models: The AI mode of character models now supports the use of “%%trigger%%” control commands in the text string of their dialogue responses, triggering to move on to the next action step.
• Audio augmentation: We added a 'jump to' option for audio triggers (not just 'next step').
• Action augmentation: We added gaze trigger functionality for action augmentations.
• Preview: We added a preview button from the publish settings to remind content authors to test their activities before uploading to the cloud.
• Marker augmentation: We improved the marker augmentation to display the target image with a 'find this’ instruction and we worked on the anchor stability of the task station during tracking.
• Locate: We reintroduced the locate functionality to activate a red arrow viewfinder pointing to the augmentation it is activated for.
• Acknowledgements: We added logos of new collaborators to the acknowledgements.

Changed

• Login at the start: The app starts now with the login, also adding buttons for registering (opens browser for web registration) and anonymous guest login.
• Activity stream: We upgraded the activity list to a swipeable activity stream on mobiles and improved the sorting and search functionality.
• Publish dialogue: We created a new ‘Publish...' dialogue to simplify data handling for content authors, with quick access for saving locally and saving to the cloud in public (or private).
• Image and video augmentations: We added boundary box handles for easier manipulation to the image and video augmentations.
• Step order: We improved the process of adding action steps during editing.
• Best augmentations first: We reordered the list of augmentations in the editor by popularity.
• Content selection: We improved support for content selection, adding a short description and additional context help to the augmentation list on mobiles.
• Calibration dialogue: We created a new calibration guide and dialogue, which now starts automatically when a user opens an activity to view.
• Highlight current step: We are now visually highlighting the current step in the step list also on mobiles.
• Keep alive: We created a new dialogue with from/to dials on mobiles for the ‘keep alive’ functionality of augmentations, simplifying setting from which action step to which action step an augmentation shall remain visible.
• Updated views: We implemented new views for activity settings, steps, list of step contents, content selection, step settings, and profile.
• Activity: We improved the process of deleting activities.
• Screen layout: We improved the basic screen layout for tablets and large screen phones (e.g. Motorola Edge Pro). We increased icon resolutions.

Fixed

• We updated the reference resolution for the new mobile user interface, which was causing crashes on some devices.
• We extended the audio trigger functionality to work also with the GhostTrack augmentation.
• Fixed portrait mode for video augmentation player.
• We fixed bugs with the AI mode for character models.
• We fixed bugs causing the image augmentation to crash (iOS), reset (Android), or not display (all).
• We suppressed the calibration video instructions from displaying during the editing tutorial, so that the optional dialogue does not occlude the tutorial.
• We fixed issues with the appearance of character models ('zombie mode') and improved the appearance for some.
• We fixed issues with calibration.
• We fixed a bug causing label augmentations to crash when using the trigger.
• We fixed a bug with pick & place objects forgetting their orientation.

Enterprise

• We updated the base URI for xAPI statements, retiring ‘wekit-community.org’.
• We removed the Android advertising SDK package from project (it was never used).
• We added a new profile setting for selecting repository servers from a dropdown list of preconfigured endpoints.
• We added new profile settings for selecting the learning record store from a dropdown list, automatically configuring their xAPI endpoint URL, replacing the free text entry.
• We fixed issues with the Sketchfab API direct login and authentication.
• We reinstated the broken deep link launch from QR codes in Moodle (allowing MirageXR to launch from any QR-code enabled mobile camera app).
• We added new app icons.

Developer

• We migrated to Unity 2021 LTS and updated the CI pipelines to use the corresponding images.
• We updated the ARfoundation versions.
• We created a new UI kit to unify the presentation layer.
• We introduced a new dialogue manager for presenting interactive dialogues.
• We implemented a new drag & drop controller for ordering of UI elements.
• We added CONTRIBUTING.md instructions, replacing the agile development wiki page.
• We updated the CI pipeline badges in the README.md.
• We excluded workspace layout settings from git index and added to gitignore.
• We removed the UserSettings folder from git index and added it to gitignore.
• We removed some ghost meta files that were still tracked by git.
• We updated the cache action on the Android build pipeline.
• We fixed the Android CI pipeline signing error and left a note in the CONTRIBUTING.md about not ticking the developer key option in the build settings.
• We fixed the problem of the Android build pipeline running out of space before concluding the build.
• We added missing standard Windows fonts on the Android build pipeline, which were causing many dialogues to not display type in automated preview builds.
• We added StyleCop support and adjusted the rules to our needs, also reformatting large parts of the code to fix some of the warnings.

iOS

Version 11.0+ (ARKit required)

Android

Version 7.0+

UWP (Windows Holographic)

10.0.17134.0+

Release 1.9.2

Change Log

Featured

• We refactored the core, breaking up ActivityManager and WorkplaceManager into more modular classes, rebuilding a better logic and flow with higher performance. This affects in particular: resuming activities, timing, event handling, bootstrapping, more clean separation of model-view-controller, persistence, addressables, with improved information hiding.
• We added an in-app tutorial for the mobile user interface.
• We migrated the project from GitLab provided by XR4ALL to a new code hosting platform on GitHub, set up new Github runners for Windows and Linux, and refactored our CI pipelines.

Added

• We added new animations for the character models.
• Pick & place now allows setting whether location resets or is remembered when repeating the activity. We also added contextual help to explain configuration better.
• The model augmentation now has a log-out button, which allows forcing to reauthenticate with Sketchfab.
• The Learning Analytics xAPI endpoint (learning record store) can now be configured in the settings menu

Changed

• We fixed a bug in the activity list, which prevented displaying and searching the title.
• We improved the built-in tutorial for the world space smart glasses user interface, changing wording, adding steps, and fixing a bug.
• Several improvements to the character augmentation ensure that: the position marker does not slip so easily underneath the spatial map; handling of triggers; and a bug causing the edit panel to stay visible in the viewer.
• The app version is now passed through via the IEEE P1589-2020 activity model to the repository.

Fixed

• We fixed cross-platform issues with the image and video augmentations.
• The pick & place augmentation on iOS was missing the sphere indicating the target zone.
• In the settings screen on mobile platforms, the labels are now displayed correctly.
• We fixed an issue causing glyph, character model, and drawing sometimes to forget size or position.
• We fixed an issue with the activity thumbnail not resetting, when creating a new activity.
• The activity list now correctly updates when returning to it after a new activity was created.
• “https://” is now added automatically to the Moodle URL.
• On the smart glasses UI, the grid augmentation menu now stays open when browsing to an empty action step and back.
• The pointless ‘relocate activity list’ icon is no longer displayed on mobile platforms, only on smart glasses.
• Some QR links were broken due to a bug in the Moodle repository plugin. This is now fixed.
• We added support for non-Latin characters to TextMeshPro, fixing issues with special characters in passwords and usernames.
• We did some improvement to lighting of holograms.

Enterprise

• We added a possibility to automatically build multiple apps with separate config files.
• We added an option in the BrandConfiguration class that allows now to specify which augmentation editors shall be made available (with different configurations possible for mobile / smart glasses).
• It is now possible to configure the xAPI authentication and the Moodle URL via config file.

Developer

• We migrated the project from GitLab provided by XR4ALL to a new code hosting platform on GitHub.
• We setup new Github runners for Windows and Linux, and refactored our CI pipelines, migrating to GitHub actions, finally also adding a dedicated Android CI build pipeline.
• Add abstraction layer for addressing GameObjects in the scene.
• We introduced formal reviews as part of the changed approval process: every merge request is now reviewed and upon acceptance the original developer has to merge. This improves capacity of sprint masters, shifting workload back to the originating party.
• We improved documentation, including with UML, in particular of the core and the connected augmentation system.
• We implemented automatic syntax checking with StyleCop.
• We cleaned the repository of some 3D models that are no longer used.
• On mobile platforms, we replaced Vuforia with AR Foundation.
• Notifications on slack now include the merge request details.
• Pipeline preview builds now display the correct version number.

iOS

Version 11.0+ (ARKit required)

Android

Version 7.0+

UWP (Windows Holographic)

10.0.17134.0+

Release 1.8

Change Log

Featured

• Updated Keyword Manager. MirageXR now uses voice prompts "Hey Presto", "Hey Mirage" and "Sim Sala Bim" to start voice interaction. Voice interaction has also been improved by adding visual cues for spoken words.
• IBM Watson Integration. Character augmentations can now be fitted with an IBM Watson assistant, improving user interaction.
• Tiltbrush. MirageXR now includes the popular XR painting tool Tiltbrush! The new augmentation allows adding 3D-painting elements to the training procedures.

Added

• Most visual augmentations can now be used as a trigger for the next action step.
• Gaze triggers can now be used to jump to any action step.
• Near-touch interaction added for dialogues on HoloLens2.
• 3D object manipulation is now possible through widgets at the object’s boundaries.
• Added a toggle to the mobile video editor panel that allows a user to input the orientation of a video augmentation.

Changed

• The MirageXR in-app tutorial has been updated, improving performance and maintainability.
• Special characters like ampersand (&) are now disallowed in activity names, as this could cause problems with Moodle.
• Some unnecessary voice lines were removed from the MirageXR voice assistant Maggie.
• The glyph trigger option was moved to the editor panel.
• On loading or creating a new step, the step will be labeled as the current step (in orange, not gray).
• Disabled shadows on HoloLens and improved their quality on other platforms.
• Icons and splash screens were updated.
• The rotation of the character augmentation was improved on mobile devices.
• The default placement of UI elements on HoloLens was optimized.
• The drawing augmentation button was removed on mobile platforms.
• The augmentation panel on task stations was changed from list to a grid view.

Fixed

• A bug where audio would not replay on ghost augmentations was fixed.
• The spawn point is now correctly visible on HoloLens 2 for both eyes. The custom shader was replaced with a wireframe sphere.
• Editing ghost augmentations will now also replace old audio with the new one.
• The Home button on the Aura has been moved slightly up to avoid being under the floor line.
• The Pick&Place augmentation no longer forgets the target sphere size.
• Deleting a model used by Pick&Place now works correctly.
• The GhostTrack animation and its sound are now synchronised.
• The Sketchfab thumbnail is now correctly displayed.
• The video editor now opens correctly for editing.
• A visual bug where action steps could not be deleted was fixed.
• An issue with the audio augmentation where the loop and trigger options would interact badly was fixed.
• The rotation and scale of the Image augmentation are no longer resetting on load.
• An issue where the Aura would jump up when sitting was fixed.
• The calibration guide is no longer causing errors when selecting “Don’t show again”.
• Private activities are now displayed correctly on Android.
• The bounding box on the measure tool was disabled, as it was causing errors.

Developer

• Some old folders and files have been removed or moved.
• Dropdown and guide text of character augmentations is replaced by Unity native UI elements.

Platform Compatibility

iOS

Version 11.0+ (ARKit required)

Android

Version 7.0+

UWP (Windows Holographic)

10.0.17134.0+