Bleeding Tree
A dark-forest roguelike with a twist: every run leaves something behind for the next one - your dead character becomes a ghost, and a victorious character becomes the next run’s final boss.
Role: Game Director
Context: DADIU 2020 Graduation Game
Tools: Unity, C#, Shader Graph, Visual Effect Graph
Timeframe: 8 weeks
Status: Released prototype
What I owned / delivered (as Game Director)
Unified team brainstorming into a single direction: consolidated top ideas into a shared vision document and aligned the team around it.
Ran daily leadership syncs with discipline leads to keep production moving and reduce miscommunication.
Worked across departments to surface strong ideas early and bring them into the project where they fit the vision.
In final sprint, went hands-on in-engine:
Assisted tech with an enemy AI state machine implementation.
Created the No Input Interactive logo (the team’s faux studio).
Built rain and lightning shaders/VFX for later levels using Shader Graph and VFX Graph.
How I worked
Consolidated team brainstorming into a shared vision, with clear priorities around the “losing is good / winning is bad” meta loop.
Kept alignment through daily lead syncs and cross-discipline check-ins, closing loops and preventing drift.
Supported the final sprint hands-on (AI state machine help, VFX/shaders) to remove blockers and polish the shipped slice.
What to Notice in video/build
The run-to-run persistence twist (ghost helper / player-as-boss) and how it reframes replayability.
Mood and escalation as the forest becomes more corrupted closer to the Bleeding Tree.
End-of-project polish: atmosphere work (rain/lightning) and cohesion of presentation.
Context
Player Goal: Fight through a corrupted forest toward the Bleeding Tree, collecting powers along the way, knowing your run will leave behind a persistent legacy.
Core Loop: Choose a weapon → explore/fight through levels → gain powers → die and leave a ghost, or win and become the next final boss → repeat with the world subtly changed by previous runs.
What makes it distinct: The replayability twist is systemic: success and failure both create content for future runs, and the ghost/boss inherits the exact powers you had, making each “legacy” feel personal rather than generic.
Design goals
Flip roguelike meta-progression on its head by creating a negative feedback loop where losing is good (you get help next run) and winning is bad (you become a stronger final boss next time).
Meaningful choices wherever possible, such as optional high difficulty for greater rewards, or greater powers at the cost of high corruption.
Make the legacy system personal and readable. Players should immediately understand how their last run affects the next one (ghost helper, player-as-boss), and feel ownership over the consequence.
Constraints
Fixed production window: 8 weeks with a multidisciplinary team of 18.
Distributed ownership: ideas and implementation spread across many disciplines, requiring strong communication and fast alignment.
Late-sprint reality: polish and cohesion had to land under deadline pressure, which is why I went hands-on for AI support and environment VFX.