Inspiration
I wanted to know what it would feel like to hold a memory the way someone handles an image in Minority Report. A small test made the idea click for me. I recorded a cap, brought my hand into the frame, and waved my fingers. In the headset I could bring that gesture close enough to see each finger, then set it back in my room and watch it again.
What it does
AfteRimage plays short spatial recordings in mixed reality. Pinch a memory to pick it up, spread your hands to enlarge it, and release it wherever you want it to play. You can pause on a gesture or scrub back to a particular expression. Three memories come with the Quest app, so someone trying it for the first time can start with their hands and a headset.
I also use it to make videos. I can park the playback menu off to one side, choose a green or black background, then enter Clean view to hide the controls and lock the shot. The Quest recorder captures the result for editing.
How we built it
The iPhone recorder captures color and depth together. The rear camera captures a scene, and the front TrueDepth camera captures a selfie. On the Mac, I turn the measured depth into a sequence of 3D point frames and prepare the camera images that supply the fine visual detail. A Unity and OpenXR viewer plays the sequence on Quest 3, alongside recorded audio when the capture includes it.
One camera sees one side of a moment. A small lean reveals depth around a face or a hand. Turn much farther and gaps appear where the phone never saw the surface.
Challenges we ran into
Before September 24, my prototype could replay a rear-camera recording on Quest. During the competition I added ten-second front-camera selfies, a shutter timer and optional microphone capture. I built hands-first playback and a movable menu so someone could place, resize and scrub a memory while seated. Green and black backdrops, framing guides and Clean view grew out of trying to record an actual shot through the headset.
Two problems shaped the implementation. At arm’s length, gaps appeared between points on surfaces that looked solid up close. I adjusted how much of the image each point covers and tested the same recordings at different distances in the headset. Switching between dense memories also took several seconds. Preparing the bundled memories ahead of time made repeat switches quicker, and repeated sleep/wake tests helped me improve recovery when someone takes the headset off mid-session.
Accomplishments that we're proud of
This is my second hackathon. Something I didn't have the confidence to participate in before. That to me alone is an achievement.
What we learned
I've learned how to calibrate a VR environment
What's next for AfteRimage
I’m targeting January 28, 2027 for the Meta Store launch. My first priority is making the path from a personal capture to Quest easier for a new user. After that I want to explore capturing a moving subject from several angles so you can see more of the moment as you walk around it.
Built With
- arkit
- avfoundation
- c#
- meta-quest-3
- openxr
- python
- swift
- swiftui
- truedepth
- unity
