Selected Games
Researches & Other Works
Links
Heart Keys | 心·键
A 2.5D Puzzle Adventure
2025 · Ludum Dare 58
Production Lead / Team Lead
Credited as Director
Heart Keys is a puzzle adventure about collecting abilities and using them to find a way forward. Our team made it for Ludum Dare 58, where it placed 19th overall.
I led production remotely. My work covered helping newer teammates get started, coordinating decisions across disciplines, and keeping the game within the time we had. I also contributed code near the deadline and helped with the cover illustration and the game's name. The core mechanics came from the team's designers.
Leading remotely
I was away from the rest of the team, so I could not spend the jam working alongside everyone in the same room. That shaped where I put my effort. I followed the project across disciplines and stepped in where a conversation, a decision, or some practical help could get work moving again.
Several teammates were new to this kind of project. I helped them get started and talked through how their work would fit into the game. My familiarity with design, art, and programming helped here: I could discuss an idea with a designer, understand what it asked of an artist, and explain the implementation cost to the team.
Those conversations also helped me judge when we were asking too much of someone for the time remaining.
Fixing the camera, reducing the workload
The early direction was a fully 3D game with a camera the player could move around. Monument Valley was a visual reference. I liked the sense of space, but a freely moving camera would make much more of that space our responsibility. The artists would have to account for surfaces and angles a fixed view could leave unseen.
I suggested moving to a fixed-camera, 2.5D presentation. We could keep three-dimensional environments while concentrating the modelling work on what the player would actually see. Areas outside the view would not need the same treatment as the playable scene.
This was a decision I could make because I understood enough of the art workload to connect a camera choice to the schedule. It gave the team a more manageable version of the game to build, while preserving the spatial character we wanted.
EARLY DIRECTION
Freely moving camera
More surfaces to model and finish
MY RECOMMENDATION
Fixed-camera 2.5D
Focus asset work on the visible scene
Stopping repeated rework
The hardest production problem was disagreement between designers. Sometimes we had already discussed an asset with the artists and settled on an approach, only for a later design argument to reopen it. Requests to start over kept pulling the art work back into decisions we thought we had made.
I had to reject some of those requests and give the team a decision they could continue working from. Reconsidering the design had a cost: someone would have to discard work and make it again, with less time left to finish everything else.
This was the least comfortable part of leading the jam. I wanted people to contribute ideas, including teammates with less experience. But I also had to protect the work everyone had already agreed to do. Once discussion began delaying production, I took responsibility for the final call.
Finding an art style we could finish
I kept talking with the artists as their work developed. We needed a style they could produce consistently within the jam, so I proposed a simple, unlit treatment for the 3D art, with Monument Valley as a reference for stylised forms and colour. That was the production direction I advocated; the finished scenes developed their own lighting treatment.
My part in these discussions was to keep the visual ambition tied to the time available. I also helped draw the cover illustration and choose the name, Heart Keys. These were smaller contributions alongside the artists' work on the game itself.
Supporting the designers
I encouraged a puzzle direction in which players make progress by learning something about the game. From my own experience with jams, I felt that kind of discovery could work well for the audience.
The designers developed the core mechanics. I helped discuss their proposals, revise ideas with the team, and judge what we could finish. A proposal could make sense as a puzzle and still ask for more art or programming than we had time to produce.
Helping the final build come together
Selected Unity contributionsAs the deadline approached, I took on programming work directly. My contributions included level exits and scene transitions, configurable teleport triggers, and an ability-selection interface connected to the existing player controller.
The interface let players choose between Delete and Create, or Teleport and Recall. I implemented the selection and connected it to abilities already present in the project. I also added a sandstorm particle prefab and an ambience controller with audio fading.
Later work included the scrolling credits, the ending scene, and build integration. These tasks helped bring the team's separate pieces into a game players could move through and finish.
05 OCT 2025
Level exits & scene transitions
Configurable teleport triggers
Ability-selection UI & triggers
06 OCT 2025
Sandstorm ambience
Particles and audio fading
What we finished
We submitted the game on time. Heart Keys placed 19th overall in Ludum Dare 58, with an overall score of 4.283 out of 5 from 32 ratings. It also placed 60th for both Innovation and Graphics. I was proud of what the team had managed to finish together.
19th
Overall · Ludum Dare 58 Jam
4.283 / 5 · 32 ratings
Player feedback and our response
Players noticed the details the team had worked to bring together. Aentro singled out collecting the movement keys and gaining abilities. FrogO2 liked the consistency between the 2D and 3D art, the relaxed atmosphere, and the Recall puzzles.
The same comments exposed problems that interrupted the experience. Players reported getting trapped, being unable to close the help menu, and reaching the final door without being able to open it. Wles pointed out that both the orange and blue ability-selection states looked active.
I responded to the bug reports and posted our version 1.1.2 announcement, crediting the team's work and especially our chief programmer, Yumo. The update fixed an unobtainable progression key and cases where players became stuck in stages. I invited players to return and try that version.
I would revisit the selection colours and text readability in a further pass, so players can tell which ability is active before using it.
What I would establish earlier
For the next project, I would establish who makes the final call before disagreements start consuming production time. People should know when a decision is still open and when they can safely build on it. I learned that I can contribute a great deal through those decisions, even when I am working remotely and someone else is designing the core game.