Game: Sogen

Genere: VR, Multiplayer, Puzzle, Open world

Team size: 20 - 4 developer

My Tasks:

• Developed a custom AI knowledge system integrated with OpenAI tools (Whisper, ChatGPT, Wit)

• Built multiplayer networking solutions using C#, RPC, Google Cloud, and Photon API

• Designed an in-game progress dashboard and HUD UI

• Created an 8K-360 video streaming and decoding system

• Leveraged Git, Visual Studio, Teams, and Asana for collaborative development and project management

• Supported the QA team and maintained well-documented code for consistency and ease of updates

Developed a multiplayer dashboard that tracks game progress by displaying gathered objects and visited locations. Key features include:

• Location Markers: Each location is represented by a sphere, indicating areas explored by players or teammates.

• Object Collection: Tracks objects collected throughout gameplay by players or team members.

• Interactive Gaze Scanning: Allows players to interact with the dashboard using gaze-based scanning.

• Mode Activation: Enables activation of different dashboard modes for enhanced functionality and display customization.

In VR, traditional flat UIs are not effective, so I implemented a curved UI system by rendering the interface onto a mesh. This approach required a unique interaction model, as conventional clicking does not function as with standard UIs. Key features include:

• UI Customization: Allows players to move, rotate, scale, and apply different skins to the UI, with settings saved for the duration of the play session.

• Interaction Mechanisms: Initially developed with VR controllers, later upgraded to enable fingertip interactions for enhanced precision.

• This system enhances VR immersion by adapting the UI’s shape and interaction style to fit VR environments naturally.

In pursuit of VR sculpting with natural hand interaction, I tackled significant memory challenges by diving deeply into mesh deformation. I developed a custom deformation system in pure C#, applying movement to each vertex of an object upon interaction. This involved simulating spring-like motion at the point of interaction, with each vertex’s response based on its distance from the action point over time.

Initially, this approach was memory-intensive. To optimize, I experimented with Entity Component System (ECS), using 10,000 small objects to create a responsive environment, later exploring fluid-like simulation for more realistic sculpting behavior.