SnapToReality: Aligning Augmented Reality to the Real World
summary
SnapToReality is a Unity prototype that extracts 3D edge and planar surface constraints from a Kinect feed and an OptiTrack-tracked Lumus optical HMD, running edge detection at ~15 FPS and plane detection at ~4 FPS. The system snaps virtual objects by matching pairs of compatible constraints, visualizes planes with a fuzzy centroid and connecting line, and shows in a 40-participant study that placement and orientation are faster (4.5s vs 11s) while scaling shows no significant benefit; it surfaces issues like limited field of view, sensor noise, and dynamic constraints.
AI-written summary · about
chapters (11)
- 0:00Introduction to SnapToReality
- 1:30Concept of snapping in AR
- 3:16Prototype system overview
- 5:03Extracting real-world constraints
- 6:27Snapping algorithm explained
- 7:49User evaluation of constraints
- 8:32AR content placement study
- 9:54AR content creation demo
- 11:44Design considerations for AR
- 13:34Dynamic constraints and visualization
- 14:39Future work and conclusions
related talks
42mHow to Augment Your Reality
43mBeyond The Printed Page
16mSceneCtrl: Mixed Reality Enhancement via Efficient Scene Editing
48mRendering techniques for augmented reality and a look ahead at AR foundation
01h 10mCreating Tools for Spatial Computing
29mExtend your reality with AR
44mBetween Two Realities: Testing AR / MR
12mDo You See What I See? Bring Live Pedestrians into an Outdoor Collaborative Mixed Reality Experience