Skip to main content

What’s new in ARKit 4


Here it is. The latest version of Apple’s Augmented Reality framework, ARKit 4, has just been announced at WWDC 2020.

Let’s see what’s new in ARKit 4 and for Augmented Reality app development on iOS.  If you’re more inquisitive I linked all stuff directly to Apple documentation.

Location anchors

ARKit location anchors allow you to place virtual content in relation to anchors in the real world, such as points of interest in a city, taking the AR experience into the outdoors.  

By setting geographic coordinates (latitude, longitude, and altitude) and leveraging data from Apple Maps, you can create AR experiences linked to a specific world location. Apple calls this process “visual localization”, basically locating your device in relation to its surroundings with a higher grade of accuracy.  

All iPhones & iPad with at least an A12 bionic chip and GPS are supported.

Depth API

The new ARKit depth API, coming in iOS 14, provides access to valuable environment depth data, powering a better scene understanding and overall enabling superior occlusion handling.   

It relies on the LiDAR scanner introduced in the latest iPad Pro.   

The new Depth API works together with the scene geometry API (released with ARKit 3.5), which creates a 3D matrix of readings of the environment. Each dot comes with a confidence value.  All these readings combined to provide detailed depth information, improving scene understanding and virtual object occlusion features. 

Improved Object Placement

The LiDAR scanner is bringing more improvements to the table of AR development. In ARKit 3, Apple introduced the Raycasting API, which allows placing virtual objects with ease on real-world surfaces by finding 3D positions on these. Now, thanks to the LiDAR sensor, this process is quicker and more accurate than before.  

Raycasting leverages scene depth to understand the environment and place virtual objects attached to a real-world plane, such as a couch, a table, or the floor. With the LiDAR sensor, you don’t need to wait for plane scanning to spawn virtual content, as it does it instantly.   

This added to the improvements of Object Occlusion make virtual objects appear and behave more realistically.  

My thoughts

It looks like we will get the LiDAR scanner in the iPhone this year. I hope that will bring more and more apps to the AR world.

If you read my Twitter you probably know that I am writing AR apps too and one of them is even my engineering work 😊


Popular posts from this blog

How to build FAQ Chatbot on Dialogflow?

  After Google I/O I’m inspired and excited to try new APIs and learn new stuff from Google. This time I decided to try Dialogflow and build a Flutter Chatbot app that will answer some frequently asked questions about Dialogflow. This time I want to be focused more on Dialogflow rather than Flutter. Firstly, go to  Dialogflow ES console , create a new Agent, specify the agent’s name, choose English as a language and click “Create”. As you created a new agent go to setting and enable beta features and APIs and Save. Now let’s model our Dialogflow agent When you create a new Dialogflow agent, two default intents will be created automatically. The  Default Welcome Intent  is the first flow you get to when you start a conversation with the agent. The  Default Fallback Intent  is the flow you’ll get once the agent can’t understand you or can not match intent with what you just said. Click  Intents > Default Welcome Intent Scroll down to  Responses . Clear all Text Responses. In the defau

Vertex AI – One AI platform, every ML tool you need

  This year on Google I/O (Google’s Developer conference) Google presented a new platform that unites all ML tools. Vertex AI brings together the Google Cloud services for building ML under one, unified UI and API. There are many benefits to using Vertex AI. You can train models without code, with minimal expertise required, and take advantage of AutoML to build models in less time. Also, Vertex AI’s custom model tooling supports advanced ML coding, with nearly 80% fewer lines of code required to train a model with custom libraries than competitive platforms. Google Vertex AI logo You can use Vertex AI to manage the following stages in the ML workflow: Define and upload a dataset. Train an ML model on your data: Train model Evaluate model accuracy Tune hyperparameters (custom training only) Upload and store your model in Vertex AI. Deploy your trained model and get an endpoint for serving predictions. Send prediction requests to your endpoint. Specify a prediction traffic split in your

What the Flutter? ExpansionPanel

  A common pattern in apps is to have a list of items that you can expand to show more details. Sometimes these details don’t justify an entirely separate view, and you just need them to show up inline in the list. For that, check out ExpansionPanel, a widget that when tapped on will expand a panel. Start with the  headerBuilder , which returns what the first line of this panel will be. It takes a context and a boolean, so you can change what it looks like when the panel is open vs closed and returns a widget. Next up is the body, which contains the contents of the opened panel. And finally is the boolean  isExpanded  to indicate whether or not this panel is currently open. But what to do with this flag? Well, ExpansionPanels almost exclusively appear as children of ExpansionPanelLists. Here, we can maintain a list of which panels are open and use  ExpansionPanelList’s  expansionCallback parameter to update them. This callback takes an index of the panel that’s just been tapped and whe