snes-team

Inspiration

Recently I came across an amazing A.R experience from NFL, and got immediately inspired, and decided to turn it into an AR filter. Then I asked myself, "who did I always wanted to have a picture with ?" Why not my childhood heroes ?

What it does

With the snes-team filter you can shot a team video with the coolest super nintendo characters.

Just open the filter on instagram, and activate your device's back camera, then position you or your friends inside the target and start recording to call your virtual heroes for a photo.

You can also adjust the target position and rotation using touch gestures before shooting, to get the best shot possible.

How we built

Using the spark ar's plane tracker to positioning the whole scene and use its height as ground reference for the animations.

Plus combining Animation Sequence with patches we manipulate and synced the animations frames and characters positions

Challenges we ran into

For the first prototype we use segmentation, this way some of the characters were occluded by the users, increasing the filter realism.

However aiming to reach a large range of devices and increase the filter performance, we decided to remove this feature

Also we deal with multiple animations using in sprite sheets, chaining this animations, making a single long animation. And we needed to sync this animations with the characters movement through the space.

Accomplishments that we're proud of

The patches designed to sync and chain the animations became pretty clear to visualize and easy to manipulate. Making the animations parameters and keyframes adjustments fast and reliable during the development.

The result using segmentation was very satisfactory as well, and can be used for video case purposes

What we learned

To increase the performance and keep the filter file size small, we learned how to pack multiple animations into one spritesheet and how to manipulate this animations individually inside spark, combining basic patches to result in complex behaviors and smooth animations transitions.

What's next for snes-team

Find a way to implement the segmentation keeping good performance. And launch new versions with different characters from different contexts, using 3D characters as well

Built With

• ar
• augmented-reality
• particle
• sparkar