Prototyping for AR

Session 808 WWDC 2018

Designing for AR can be intimidating and discovering design flaws late in the process can be costly. See how low-tech traditional prototyping techniques can help you validate and refine your AR app and game design ideas.

[ Music ]

[ Applause ]

Thanks, Mike.

Hey everyone.

How many of you're interested in AR?

All right.

Cool. Now how many of you have an AR app in the App Store already?

Okay. A few less.

I get it. AR can be pretty intimidating.

Well, today we're going to look at a few techniques that we use which will hopefully help you in two ways.

First, they're going to hopefully help make AR more approachable and second, we're going to show you how you can save time and money by making sure you build the right things before you write a single line of code.

So let's get started.

If you've seen previous talks by my team here at WWDC, you may already be familiar with our revolutionary app Toast Modern, the best app for finding and rating the toast around you.

I know it's amazing.

Well, we've been building up our database of toast, and now we've been tasked with adding AR to this app.

I'm sure you're all aware that the artisanal toast scene in the Bay Area it's pretty big right now.

Well, today it's going to get a whole lot bigger.

Introducing ToastAR, the world's first AR toast recognition and rating app.

I know. Thank you.

Thank you.

Through the use of computer vision and AR, ToastAR lets you check the ratings on a piece of toast and get this, where it's from.

Revolutionary I know.

I mean, you're all developers for Apple platforms so you expect the best APIs and the best hardware.

Why not also expect only the best toast too.

So how did we prototype this beautiful, revolutionary app?

Well, you may think we used a programming language like Swift or a framework like ARKit or maybe even a sophisticated application such as Unity.

These are all valid ways to prototype, but they're not the only solution.

Sometimes you can use low-tech physical prototyping techniques to work through your design problems before you write a single line of code.

So today I'm going to show you how you can use just the camera app and the world around you to prototype for AR just like we did for ToastAR.

So before we get started, there's two terms you should be familiar with.

First world space.

When we talk about world space, we're talking about design elements in the world around you and your device.

And when we talk about screen space, we're talking about these elements on the screen of your device.

So let's take a look at the original design we got for ToastAR and work through our process.

In this design, we have a label in world space which shows us where the toast is from and its current rating.

Now, you may want to jump in and start coding this immediately.

It looks pretty simple, but instead we're going to save time and money by making sure we are building only the right things before we write a single line of code.

AR is about the world around you.

So testing out your designs is difficult through a single still or even video.

Physical prototyping, however, allows us to do this in an approachable, low-tech and highly-iterative manner.

So how did we get started here?

Well, we printed out our labels and stuck them to some toothpicks.

I know. I know.

It may seem odd, but this low-tech process is going to hopefully help us in three ways.

First, we're going to be testing our designs in context.

Remember, AR is about the world around you.

Second, we're going to hopefully figure out any issues with our design before we write a single line of code.

And third, through this process, hopefully we'll get new ideas that will help make ToastAR even better.

But who even has a printer these days, right?

You can even just draw something and cut it out.

So now that we have our labels attached to our toothpicks, well, we stick them in a piece of toast.

Then, just using the camera app, we can test this design out in world space.

I know. Revolutionary here, right?

[ Applause ]

When we tried this, we learned one thing really quickly.

These labels, at an angle, are pretty difficult to read.

So the label is hard to read at an angle.

Maybe we should have it always face the camera.

So let's try this out.

So with some help from a co-worker and some extremely sophisticated hardware, okay, it's a clear ruler but beautiful.

We tested this out.

And we found that yes, the label is easier to read when it's always facing the camera.

Well, this behavior is commonly referred to as billboarding.

And it seems like it'd be pretty easy to implement in SceneKit using an SCNBillboardConstraint.

But before we start coding this, let's continue to test this design and make sure we're only building the right things.

So how does the label work when there's a lot more text or when it's at a distance?

Well, we can try this out pretty easily too.

Again, with the camera app, we can test this out and we find that yes, it's a little difficult to read when these labels are at a distance.

Maybe we should bring these labels on the screen.

So how do we prototype this?

Well, we're talking about screen space, and we want to do this quickly.

So we're going to go ahead and use Keynote.

First, we're going to record a video of our toast.

We'll record a slow pan.

And notice that we're going to pause on the toast to give our upcoming animation some time.

Next, we're going to bring this into Keynote.

So you'll create a custom document size which matches the device you're eventually going to put this on and then simply import your video and bring it into Keynote.

Third, we're going to animate our labeling, import our design, add a move action, give it an appropriate duration, and move it to the desired destination.

Now here's the important part.

We're going to check the build order and make sure our video and our animation begin at the same time but we're giving the animation a delay so it's on screen when the toast is in the field of view.

Finally, we're going to put this on a device and test it out in context.

So we'll put it on a device using Keynote for iOS.

And we'll test this out in context.

Great. Looks like we could fit all the text we need on this label.

And putting this on screen space will hopefully make it easier to expand the scope of this label at a future date.

Now let's see how this scales, though.

What happens when there's multiple pieces of toast in the field of view?

So using the same techniques, we'll try this out too.

Now the problem here is when there's multiple pieces of toast in the field of view, I have no idea which piece of toast I'm currently interacting with.

Maybe I should try putting an indicator in world space so that I know which piece of toast I'm currently interacting with.

Well, we can try this out too, quickly.

So, again, with some help from our co-worker and our trusty ruler, we can try this out too.

Print out our design, attach it to the ruler, and try this out.

It's the camera app world space.

This seems to help us.

We know which piece of toast we're currently interacting with.

So let's put it all together.

Using the techniques we just reviewed, we can put together a rough demo of our prototype.

And when we know we're building the right things, we now know we have some high-fidelity prototypes to base our code off of.

In just a few hours, we worked through problems that could have taken us days to code through.

So what did we learn?

Well, we learned these labels in world space are difficult to read at an angle and at a distance, so we brought them on screen.

But we also learned that we needed indicators in world space to let the user know what they're currently interacting with.

Physical prototyping allowed us to test these AR experiences out quickly.

We used the camera app to test our world space interactions.

And we used Keynote to test out our screen space interactions.

So how did prototyping help us?

Well, we tested our ideas, and now we're going to be saving time and money by making sure we build the right things before we write a single line of code.

We also got some new ideas along the process.

These ideas helped make ToastAR even better.

And take it from me, it's pretty amazing.

Hopefully when you start working on your next project, you now have a few new tools when you tackle your next AR app.

Thank you so much.

[ Applause ]

Apple, Inc. AAPL
1 Infinite Loop Cupertino CA 95014 US