3 things we’ve learned about designing for Magic Leap One

During my recent work at Threesixty Reality, I’ve had the opportunity to run a series of UX research sessions for a new experience on Magic Leap.

It’s an incredible privilege to be able to research and design immersive tech. Standards are yet to emerge and the design language is constantly evolving. For these reasons, it’s even more important that we share what we learn with the rest of the community. These are some UX considerations on designing for Magic Leap. They emerged from observing users going through the experience in lab-based sessions.

1. Limited field of view + wireless controllers can generate confusion and frustration

While improving on Microsoft Hololens 1, Magic Leap still provides a limited experience in terms of field of view. Today, this is probably the most limiting factor of UX when it comes to Augmented Reality.

The ML-1 comes with wireless controller and with a recent update it provides six degrees of freedom, i.e. can change its position and rotation.

The combination of a limited field of view and a wireless controller poses some interesting challenges for UX:

  • The controller (and its raycast, i.e. the beam used to point at things) is only visible if the controller itself is in the field of view
  • What you’re focusing on (your gaze target), and what you’re pointing at (your raycast target), are likely to be two different points in space
  • While your gaze target is always in the FOV (because it’s where you’re looking), your raycast target might not be. In other words, you might be focsuing on something while you’re pointing at something else.
Gaze target (yellow) and raycast target (blue) don’t always overlap

When using a wireless controller, it’s best to show the feedback of a selection close to the object that users are selecting.

When feedback is displayed outside of the field of view it’s basically invisible

Users shouldn’t need to move their head to see the feedback, which is not only inefficient but could also cause strain to the neck (at 0.76 pounds, the Magic Leap can still cause fatigue).

The other problem with feedback being far from the selection is that users might inadvertently move the controller from the current target, consequently losing the feedback.

When the feedback is far from the target, it’s easy to lose track of where the controller is pointing

2. Users want to get close to things

Another interesting insight about Magic Leap comes from the so-called near clipping plane. This basically means that when something gets closer than 37cm to the headset, it disappears.

In their documentation, Magic Leap say that this helps with the comfort of the experience, as seeing things too close to you might cause discomfort. A 37cm near clipping plane has two consequences:

  1. Everything is cut off, including your hands and the controller
  2. It’s impossible to get close to things

During the project we worked on, people were immersed in an environment with interesting objects that could be picked up. We found that they wanted to bring the object close to them to be able to observe the details.

Because of the near clipping plane set at 37cm, anything that got close to them, would disappear. This caused enormous frustration .

Magic Leap One is still a development kit and it’s likely that developers will be able to change the minimum clipping plane once a consumer version is released.

In the meanwhile, if what users want is getting closer to the objects, we use scale to make the objects bigger: scaling an object would a similar effect and allow them to see more in detail.

3. Voice-over and audio need to be timed right

We repeatedly see users being so in awe of VR and AR to the point that their attention is completely captivated by the visual and interactive elements of the experience.

Attention is a limited resource and it’s key to orchestrate the different stimuli so that important information is not missed by the user.

Applications that rely on voice-over to on-board users and provide critical instructions can fail because they deliver audio information at the wrong time, often too early, when the user is still engaged and exploring the environment.