1. We prepared very simple introductions in the game and it was not enough for testers to find out what to do first in the game.
a) Intro was not informative enough about what this game’s goal is or what this game is.
b) Hint scrolls in the game were actually easy to find but the hint on the behind wall was relatively harder to find. More clear indications that tell players what kinds of actions (move forward, turn and cast magic) are possible in a specific scene will be useful.
c) A map on the screen is needed for more pleasant game play.
2. Magic System
a) Our intention was for players to find out naturally/smoothly how magic spell system work through puzzles in the game but test-play didn’t go so well according to our expectations. Jacob and I agreed that we had better implement more detailed introduction that help players understand various characteristics in the magic system.
b) 3 out of 4 testers had difficulties in fining out correct order in overlaying markers at first.
c) Even though all test players understood ‘spell overlaying’ mechanics after all, I think none of them was able to clearly find out ‘spell steering’ mechanic. This is maybe due to the fact that puzzles in the prototype version were not so appropriate for ‘spell steering.’ Overlaid captured screen and recognized magic spell displayed on top of that were a little bit distracting.
d) For some testers, dummy markers in the magic book were confusing.
e) The magic book as a physical interface was effective. Testers felt it more interactive.
3. AR Technique
a) It seemed testers easily understand what AR technique is. However, as they haven’t experienced how AR is sensitive to the light conditions (brightness and reflection), recognition failure happened many times during the play.
b) Most of failures were caused by ‘blocking square borders with fingers’ situation.
c) When players were showing spells to the cam, it was difficult for them to be aware of how that spell is tilted or panned.
d) We need more robust image processing algorithm but I’m worrying a slight increase in recognition ratio with a lot of algorithm processing overhead.