CHI 2014 – One of a CHInd

This was the CHI, Ed’s first conference and ours as well!

We spent 7 exciting days in Toronto, Canada, showing off our game to hundreds of people in both a 5-minute pitch and a live demo.

We went home proud finalists of the CHI 2014 student game competition!

Staggering to CHI with Drunken Ed

It’s been almost 2 years, since we presented our Drunken Ed project at the end of the MPI lecture, when suddenly our supervisor approached us with the idea to submit Drunken Ed to the CHI 2014 Student Games Competition. The requirement was an extended abstract and an executable of Ed, which seemed feasible. So we set to work and put our development process and the challenges we faced down on four pages. Luckily Jörg and David assisted us by proofreading our work.
On development side we also tweaked some features. Andreas created new graphics and we replaced all the old ones. It looks much prettier now!

The deadline arrived quite surprisingly (I don’t know why this keeps happening, deadlines just seem to sneak up on you), and we hurriedly created a build and a final pdf. Uploading it was another lengthy ordeal.. first the pdf did not have the correct settings, then fonts were not embedded, etc. We finished our submission just in time.

A couple of months later, when we had already quite forgotten about the submission, we received great news:
“We are pleased to tell you that your paper submission, 103, Drunken Ed – A Balance Game for Public Large Screen Displays, has been conditionally accepted for inclusion in the CHI 2014 conference Student Game Competition. Your team submitted one of the top 3 papers in your category.”  Yay! \o/

The review we got was pretty good, so only a few changes had to be made to the paper before submitting the final version. Here it is! It is also available in the ACM digital library.

After that, registering for the conference and booking our flights and accomodation took us about 3 days..unbelievable huh?
Our complete Drunken Ed-team met up again last week to tweak some more features of the game. It was fun to be back together like in old times ;) Our entire TLabs team was helpful playtesting and making videos for the presentation.

Now our bags are packed, and tomorrow Andreas and I are flying to Toronto where we will be demonstrating our game and giving a 5 minute pitch. The presentation slides are still not complete, but we’ve got a long flight ahead of us, and anyways presentation day won’t be until Wednesday. Wish us luck!

And if you happen to be there, come check out our game at the CHI!

The End

Finally here it is, the final presentation day!

We got up really early, even watched the sun rise – which is a very rare thing to happen. And we are all excited to show off our project. We put a lot of work into it and we think that we have created a neat little game.

For all of you who are not able to attend our presentation and live demonstration – you’re really missing out! But here are some screenshots demonstrating the final state of Drunken Ed. (And our slides can be found here.)

Main Menu:

Displaying Hints:

The Game:

Falling:

Game Over:

9 Screenshots

Because there have been so many visible changes this past week, we’d like to make one more post. Not wanting to make you read any more, here is a demonstration of our progress in 9 screenshots.

Menu:

Game:

Game Over:

Evaluation-guided Changes II

This week we also concentrated once again on following Nielsen’s usability heuristics. These are the most important changes we have made, to better fulfill the following principles.

Visibility of system status: Ed can fall for two reasons, losing balance because he leans too far, or stumbling because he runs too fast. In the first scenario we signalize this with flailing arms, and now we also have an animation to warn the user in the second case: spinning arms.

Error prevention: If the user leaves the trackable area, he is now urged to move back to a better position.

Recognition rather than recall: All possible actions should be visible to the player. We have now implemented a visual hint for the player to perform the bending movement.

Help and documentation: All states, required actions and hints are made clear by a short line of text and additional graphics.

The Think Aloud Study

The Think Aloud study was performed in the apartment of a team member, on a quiet evening, when both flat mates were at home. These two flat mates, whose identities will not be revealed, volunteered to participate in our study. Their comments and thoughts were recorded in brief notes.

The following sheet was haded out to them, briefly explaining our study, and the tasks they needed to perform: Think Aloud Study.pdf

User Information:

User A
Age: 21
Gender: Female
Study: History of Art

User A is rather inexperienced with computer games and has never before interacted with any type of optical tracking system. She was quite clueless about how to move the character, and managed more by coincidence, while walking and hopping around. This however led to her leaving the tracking area a couple of times and the notification to step back into the area left her somewhat confused.
The drinking gesture on the other hand, was understood immediately due to the textual and graphical hints. So selecting the difficulty of the levels was no problem at all – except for the moving about of course. But once she figured out the bending part, this was no longer an obstacle.
The task of NOT taking a picture, when reaching the highscore screen took some thinking, but she patiently waited until the countdown was over. The overall impression was that the game was fun and the highscores gave motivation to try again.
At some point she got overexcited and dared to drink Vodka – the really hard stuff. The result was an almost immediate crash of our poor Ed. This level might be slightly too difficult.

User B
Age: 26
Gender: Male
Study: Computer Science

User B has quite a lot of experience with gesture tracking systems himself, and quickly caught on to the controls. He performed the drinking gesture rather sloppily however, which resulted in a somewhat strange animation. Other than that the performance went smoothly. He felt the Easy level was boring, since it was no challenge for him. The Vodka level however seemed too difficult once again. So oddly Wine seems to be his drink of choice all of a sudden. Furthermore he commented on the countdown on the Game Over screen, which he found too long.

Discovered Issues

From the issues mentioned by our candidates, we have singled out the following, ordered by severity and starting with the most urgent. Most of these have already been implemented.

• Bending Hint: If the player seems incapable of moving the character around, a hint is displayed.
• Leaving the tracking area: If the player leaves the tracking area, he is now told approximately in which direction he needs to move to get back.
• Difficulty: The difficulty of the levels will be tweaked and tested repeatedly in the following week.
• Animations: The drinking animation still requires refinement and will be worked on.
• Waiting time: Countdowns and waiting times will be reviewed, tested and adjusted, to allow a smooth flow and comfortable gaming experience.

Task Breakdown

This is a list of who did which subtasks each week.

Week 6 – Implementation

• Alexander Biskupski: Track when player leaves mark, improve drinking gesture detection, track arm and head angles
• Andreas Fender: Tweak mechanics, fading screen transitions, new fonts, stumbling animation, physics improvement, character texture
• Tiare Feuchtner: Manage blog, update Menu graphics, hint implementation, perform study
• Marcel Karsten: Create game and menu graphics, display distance in game
• Jonas Willaredt: Redesign game over screen, fixed highscore

Week 5 – Heuristic Evaluation

• Alexander Biskupski: Heuristic evaluation of Drunken Ed, track hand position, ignore multiple users when tracking
• Andreas Fender: Heuristic evaluation of Drunken Ed, character animation
• Tiare Feuchtner: Heuristic evaluation of Drunken Ed, manage blog, highscore list
• Marcel Karsten: Heuristic evaluation of Drunken Ed, endless street, graphics
• Jonas Willaredt: Heuristic evaluation of Drunken Ed, image capturing, highscore saving

Week 4 – Implementation

• Alexander Biskupski: Implement body tracking with Kinect
• Andreas Fender: Set up framework, create player model
• Tiare Feuchtner: Create main menu, game over screen, blog
• Marcel Karsten: Implement game play logic
• Jonas Willaredt: Manage highscores, image segmentation

Week 3 – Paper Prototype / Initial Tests

• Alexander Biskupski: Ideas for paper prototype, p.p. construction, testing
• Andreas Fender: Ideas for paper prototype, p.p. concept, material acquisition, p.p. construction, paper character
• Tiare Feuchtner: Ideas for paper prototype, p.p. concept, p.p. construction, gui elements, testing, blog about Observations
• Marcel Karsten: Ideas for paper prototype, p.p. construction, highscore screen, testing, blog about the Paper Prototype
• Jonas Willaredt: Ideas for paper prototype, p.p. construction, highscore screen, testing, blog about the Making Of

Week 2 – Users / Storyboarding

• Alexander Biskupski: Discuss interview questions, scenario ideas, conduct interview, discuss design principles
• Andreas Fender: Discuss interview questions, scenario ideas, create storyboard, conduct interview, discuss design principles
• Tiare Feuchtner: Write interview questions, scenario ideas, organize blog posts, description of interviewees, report interview findings, compose design principles, create poll
• Marcel Karsten: Discuss interview questions, scenario ideas, conduct interview, discuss design principles
• Jonas Willaredt: Discuss interview questions, scenario ideas, conduct interview, discuss design principles

Week 1 – Project Initialization

• Alexander Biskupski: Project description
• Andreas Fender: Brainstorm short descriptions, sketches, Project description
• Tiare Feuchtner: Blog management, Brainstorm mind map
• Marcel Karsten: Brainstorm short descriptions, sketches
• Jonas Willaredt: Brainstorm short descriptions, persona

Evaluation-guided Changes I

According to the results of our evaluation, we have implemented some features to fulfill some of Nielsen’s design principles. The following points illustrate our changes to comply with the corresponding principle.

Visibility of system status: To inform the user about when he needs to interact, we now display a short countdown when the level starts. Furthermore we will let the figure flail about, when a critical leaning angle is reached – this way the player will sense that he is about to fall and fail the game. Both scenarios can be seen below (the second image also shows the figure’s skeleton for debugging purposes).

Motivation (Additional Principle): To motivate the player to try again an get better at the game, we display the highscore boards above the level selection.

Consistency and standards: To clarify the difficulty of the levels symbolized by the different types of alcohol, we print the words “Easy”, “Medium” and “Hard” above the corresponding highscore boards.

Both of these points can be seen in the following screenshot.

Heuristic Prototype Evaluation

We have analyzed our game regarding Nielsen’s 10 Usability Heuristics and added two further principles, which we deem essential. Here we have listed them ordered by severity.

1. Visibility of system status: The system status should be fairly clear in the Menu and Game Over screen, however not so in the actual game. This aspect is absolutely necessary, and we plan to achieve it by displaying some kind of loading bar or countdown when the level starts, so the player knows when he needs to interact. Furthermore we will let the figure flail about, when a critical leaning angle is reached – this way the player will sense that he is about to fall and fail the game.
2. Motivation (Additional Principle): All difficulty levels are available to play, so an advanced player does not need to stagger through all the easier stages first to reach a sufficiently challenging level. The highscores displayed above the level selection should provide the necessary motive to start another game and improve your score.
3. Error prevention: An error we expect to ocurr, is the user leaving the trackable area. Therefore the user will be urged to position himself on the x marked on the floor, if he should wander about too much.
4. Help users recognize, diagnose, and recover from errors: If the player walks away, he will be notified, that tracking has been lost. He will then be able to resume the game, if he returns in time. Otherwise the game will switch back to the Main Menu.
5. Match between system and the real world: have We use only very few words and phrases, which should be easily understandable. We will try not to use any kinect or tracking specific vocabulary (e.g. “calibration”). The figure is controlled simply by the gestures drinking and leaning, which should be well known concepts to any individual in our target group.
6. Aesthetic and minimalist design: All textual information will be kept to a minimum. Additional info will be provided through animation or symbols, if needed.
7. Recognition rather than recall: In all user interfaces, possible gestures will be hinted to the player, by displaying “Peter Pan’s shadow”. With this we mean, that a shadow of the figure will be displayed, which performs the required action and prompts the player to imitate him. If this has no effect, arrows will be displayed indicating the critical action. During the game, the selected difficulty level should be constantly visible to the player. We might visualize this by letting the figure hold the according bottle. Furthermore we will display the distance and time the player has already stumbled along. At end of the game a score will be calculated from this.
8. Learning curve (Additional Principle): A player can improve his performance and score, if he plays more often.
9. Help and documentation: Given by the setting, ther will be no documentation or menu with a ‘Help’ option. All necessary help will be given ingame by displaying the shadow, symbols, or even text if necessary.
10. Consistency and standards: Both in the menu, and in the game, the figure can always be controlled by leaning to the side. This should afford some consistency. The unusual level-selection by performing the drinking gesture will be hinted, to compensate for it’s lack of standard. Furthermore the difficulty levels are unconventionally defined by alcohol types – to comply with a standard we will also subtly denote them with the words “Easy”, “Medium” and “Hard”.
11. User control and freedom: The only scenario we could think of, is if the player chooses a wrong door. To undo this mistake, the player has various options. He can either behave in such a way, that the figure falls down quickly, ending the level and restarting again with the menu, or he can simply walk out of the tracked area. The system will then display a countdown, during which it will allow him to return and resume the game, or finally switch back to the menu.
12. Flexibility and efficiency of use: Since it is not an application, where certain tasks should be completed in the shortest time possible, we deem this aspect irrelevant. It is a game, and not a complex one at that, so we cannot think of any shortcuts we could implement. In addition, this game is meant to help the player pass the time in a waiting situation, and make it seem as short as possible. Thus making the playing process as short and efficient as possible, is not our goal.

First Functional Prototype

This week we finally got to work on the real thing: We began implementing our prototype!

After dividing up the tasks and setting up the framework for our project, we each got to work and Drunken Ed began to stagger.
Ok, we have to admit, at first he only hovered, and then he more or less summersaulted across the screen. By now he is only capable of a lopsided slither – but it gives the right impression. In addition, our swaying world is decorated with rudimentary trees to give you a sense of movement.

Instead of selecting the levels by lifting a foot to climb the stairs (as we had intended in our paper prototype), we are now using a drinking gesture. Thus, when standing in front of the respective door, you lift your hand up to your face, as if taking a long swig from a jug of cold beer.
This change was suggested upon presenting our paper prototype, and we followed the good advice. The drinking gesture is easier to track and the player is less prone to do this unintentionally. We also use this gesture again, to restart the game at the end. This way we hope to satisfy the principle of Consistency (by both Nielsen and Norman). Also the player can be moved by bending your body, both in the game and in the menu.

Below you can find a some more screenshots from our development process.