Triage Postmortem

Triage is a top-down patient management game, in which the player – a Triage nurse – must use the 10 minutes remaining in their shift to manage the Friday evening rush of patients. Using a controller, the player will switch between patients, moving them to the appropriate rooms.

Triage was created over a six-week period, using Unity 3D for creation, as well as having 3D assets created in 3DS Max. The project was managed in Hack’n’Plan for a group of fix people.

Project Management

Triage’s first and foremost failure-point was its project management. Triage took too long to reach critical milestones and failed to achieve its own completion goals. Tasks input into Hack’n’Plan didn’t get broken down enough, and while the task completion estimates were generous in many cases, they should be given out further for jobs to take less than one hour. More time should be allowed for testing a build before uploading it. Further using what I’ve learned about using it, I need to provide more detailed project backlogs that allow for bug-testing and smaller tasks, and encourage my team-mates to create their own to ensure they have the time they need. No proper task tracking and completion wasn’t chased by me and didn’t fix itself. As the Project Manager, I need to be more strict about deadlines and ensure that my team is prepared for every deadline. Too frequently did we lose time fixing bugs when we could have been testing, because while the backlog required that the feature is “implemented” it was never specified that it would be properly tested like the feature should have been. Since I was also the Lead Designer, the backlog of tasks should have full written excerpts from the GDD, which specify the exact behaviour of a system. Instead, members of the team were less likely to create the system as specified and make their own design decisions without my guidance. By deliberately providing snippets of the GDD in backlog notes, as well as chasing up team-members from deviating, I can avoid situations where the product does not fit the brief.

The use of Hack’n’Plan to manage the project for the first time opened up the team to failure since it wasn’t used appropriately. Hack’n’Plan’s alert system is disabled by default, so deadline notifications weren’t set up until mid-way through the project. Additionally, Hack’n’Plan is in beta and does not have a mobile app like Trello does. The only way to access Hack’n’Plan is to be using an internet-connected PC, which isn’t intuitive to use while on the go. To avoid Hack’n’Plan’s pitfalls in future, I need to properly research its uses, and the information it provides. Alternatively, Trello provides a mobile and web solution with a better notification suite. I’ve used it before, and I believe it provides a competitive and simpler solution for a small team. What application to use will be decided at the beginning of the next project, depending on what my next group prefers and is ready to install on their phones if need be. Furthermore, by having workmates setting their tasks as “completed”, I can confirm with them their completed work during a stand-up session as the project manager

Collaboration

Part of Triage’s development relied on the skills of a collaborator to supplement their abilities into our product. During our pitching phase, we failed to attract any interested Animators or Audio specialising students, due in large part to my failure to socialise and attract people through networking. Without an Animator, we reached out to a friend, who introduced us to an animator who was interested in our project. After re-pitching and explaining our expectations, I left her to work on the project, several weeks away from submission. However, after a few weeks and only one consultation with us to clarify our art style, she vanished. We were unable to contact her and lost our only collaborator very close to the end of the trimester. The people I did reach out to in response for last-minute collaboration didn’t respond to my messages, and as such, we did not have any other options but to use work from the Unity Asset Store.

To avoid losing collaborators again, I need to find and speak to them directly, networking and attracting them by appealing to their skillsets. Further, I need to continuously check up on my collaborators to ensure that don’t lose interest. If someone does leave the project, by checking in on them sooner, I will know if I need to find students sooner and can start searching immediately. If I had known earlier, I might have attracted a collaborator in time.

Project Workflow

The use of Version Control during Triage’s development was problematic. The project was setup with five distinct branches, each of which designed for a separate part of the project, with the intention of merging occasionally. So many branches meant that the project’s development was stretched thin, and workmates couldn’t work on one feature because they required another branch to be finished and merged. The merging itself was not clarified since there was no set schedule for merging builds. Asset serialisation and meta files were not setup correctly, meaning that some progress was occasionally lost due to the files being prone to conflicts through GitHub.

Mid-way through the project, the superfluous branches were deleted in favour of three – Design, Programming and Build. The same issues still prevailed, amplified by some other problematic Version Control practices. For example, incomplete features were pushed into the repository, and some features were disabled for debugging. As a Designer, losing features and/or being unable to test the game because of bugs and issues that weren’t top-priority made it impossible to test features and refine mechanics. Commit messages with elaborate changes had non-descriptive comments accompanying them. Without descriptive commit messages, I cannot know what is lost during a conflict, nor do I know how important the changes are that will be lost.

While setting up a project in future, Unity must be fully setup to avoid conflicts later on in the project. Incomplete and untested features CANNOT be pushed to the repository until finished and bug-free. Only two branches should be in use, Trunk and Stable for the two stages of developmental features. A set schedule will be enforced to ensure merges are predictable and work around testing and feature implementations of everyone’s work. Commit messages must be concise and descriptive if a conflict does occur.

Custom Asset Creation

To contribute to Triage’s art style, the team created custom assets to provide unique models for the scene. Some basic gurney and general hospital assets were created, providing a variety of Hospital-themed items. However, these assets have been set up under the pretence of using a Blender filter to provide the predetermined voxelated look. The filter didn’t get used in the end, and the objects themselves did not have a texture or UVW maps for Unity to bake lighting onto. The end result was an out of place model that artefacts under baked lighting. In future, using a Unity-specific model creator, such as Pro Builder, will enable us to create assets that are more compatible with Unity’s lighting and rendering features, as well as editable by everybody with Unity already installed.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s