3D Marker-Enabled Augmented Reality for Anamorphic Illusions

An undergraduate thesis presented to the Faculty of the Department of Software Technology from the College of Computer Studies at De La Salle University - Manila by Alexandra Cyrielle Mangune, Sean Paulo Nerie, Patrick Jaspher Joya, and Mr. Neil Patrick Del Gallego as our thesis adviser.


Abstract

This study proposes an augmented reality (AR) 3D-marker-based Android mobile application for a puzzle game that makes use of 3D items such as cereal boxes, chip cans, and the like. The puzzle pieces makes use of 3D models that are pre-sliced and spaced out anamorphically: A clear image will only be formed at a specific perspective or vantage point. Additionally, the puzzle pieces are dynamically distributed onto the 3D markers given a certain number of markers available in the play area. Usability testing results suggest that the application is better suited for a VR head-mounted display rather than an AR mobile device. System testing results suggest that the further development is needed to withstand insufficient lighting in the play area and to accommodate more than five (5) markers.


Research Overview

Scanning the markers

The physical markers used were selected based on common household items, so that it would be easily accessible to the user. In scanning the markers different methods were used, box markers were scanned by taking a picture of the 6 sides and uploaded them to the vuforia database online. Scanning cylindrical objects needs a picture of the whole side. For the car marker it was scanned using Vuforia's mobile scanning application.

Dynamic Distribution

Dynamically distributing the puzzle pieces is done once the the player has selected the number of markers to be used. Once selected, the number of pieces are then divided by the selected number and randomly attached to a parent container. These containers are then attached to the scanned markers in the order of which they are scanned.

auto-distribution
Solving the puzzle

In determening whether the puzzle is solved, there are 5 conditions that must be met. First is the distance between the Physical marker and the camera, secondly the distance of the marker to the anchor, third the placement of the marker around the anchor, fourth the orientation of marker based on its x y z coordinate, and fifth the height of the camera with respect to the anchor. Once a marker is has satisfied all 5 conditions, it's puzzle pieces in the virtual space will attach to the anchor's puzzle pieces. Once all puzzle pieces from the different markers are attched to the anchor the puzzle will be solved.

The values for these conditions were determined based on how we set up the markers. The markers are set to make sure they are easily visible to the camera to minimize miscalibrations. This is done by making the correct orientation and position of the marker would be the one that is easily detected, such as the side of the marker with the most features. Once we have the markers in position, we take the values from Vuforia's marker detection and set them as the correct values. To make it easier to solve the puzzle and account for some miscalibration, we have added a margin of error where the condition will still complete if it is within this value from the correct one.

debug-mode

System Overview

The application is made to be used on an android mobile device using Unity and a Vuforia API plug-in. The application currently makes use of a set of physical objects to be used as markers that is stored inside the application. There are 5 puzzle sets available and can be solved using 2 to 5 of any of the markers, the stones marker is necessary to use however as it acts as anchor.

Sample Set up with Physical Markers
Play Area


Puzzle Sets

Puzzle Set 1

puzzle1

Puzzle Set 2

puzzle2

Puzzle Set 3

puzzle3

Puzzle Set 4

puzzle4

Puzzle Set 5

puzzle5

Results

Average Time to Solve Each Puzzle Set (Mins.)

table1 chart

Post-Usability Testing Results (Technical and User Setup)

technical1 technical2 technical3


Post-Usability Testing Results (Awareness of Anamorphic Art)

awareness


Post-Usability Testing Results (Comparison Between the Prototype and AMON AR)

comparison


Wilcoxon Signed Rank Test

μ0 = There are no differences between the sentiments on the 3 criteria of the experi- enced users and the naive users.

μ1 = There is a difference between the sentiments on the 3 criteria of the experienced users and the naive users.

wilcoxon

Acknowledgements

We would like to express our sincerest gratitude to our adviser Mr. Neil Patrick Del Gallego, and our esteemed panel, Dr. Florante Salvador and Mr. Neil Romblon, who guided us throughout our whole journey and for giving us the opportunity to always improve. We would also like to extend our thanks to our friends and family who helped and supported us through the whole process.