It’s been about a year since the last studio I worked for closed, and I decided not to update my résumé. In that year, Lona and I have built something together that we hope will soon stand on its own. This is my attempt to document that first year in hindsight.
Lona and I both come from creative tech backgrounds. She’s a singer, theater director, and audio engineer. I’m a software engineer—both in and out of the games industry—and I’ve dabbled in game design and visual arts. But we wanted a clear separation of concerns for our job titles and eventual roles at the company, so we decided she would be Creative Director and I would be Technical Director.
I suppose it didn’t matter much since we’re mostly directing each other, but knowing our lanes has definitely helped keep things moving. I still contributed as a designer, and she still got her hands dirty in‑engine with level design and scripting dialog trees. But when it came time to solve specific problems or make important creative decisions, we knew who owned what.
For our first game, we decided to build a single‑player experience. This kept the testing loop simple and avoided the overhead of running servers. We knew we planned to add multiplayer and expand our shop later, so we needed to be careful not to drift too far into total isolation with a one‑off codebase. To that end, we needed a good framework—a good game engine.
I know many solo and indie devs prefer Unity or Godot lately, but I’ve been working in Unreal Engine since the Unreal 3 UDK was new, and all of my professional game‑dev experience has been in UE4 and UE5. So we picked Unreal 5 not only for its feature set, but for my skill set.
During COVID, Lona got deep into Animal Crossing while I spent a lot of time catching up with Breath of the Wild, and we both sank many, many months into Stardew Valley. By the time we committed to UE5, we had a pretty good idea of the genre we wanted to pursue. We would blend the games we loved into a singular experience—a Cozy Adventure.
I’ll save the game‑design details for another post and focus here on how we planned to build this thing. We needed tools, systems, art, sound, code—everything. Where to start?
A Version of Versioning – Why We Chose Perforce Over Git‑LFS With Just Two People
I started with Git and Git‑LFS to manage our codebase and my back catalog of Unreal Marketplace and FAB assets for prototyping. I wanted to give Lona all the creative tools I could, so we dumped every asset that looked like it might fit our vibe into a single project. This was fine at first, but we started running out of disk space. Git‑LFS was multiplying the size of our project to keep version history while we figured out our folder structure and made small edits to materials and meshes.
I tried offloading the LFS storage to a file server with a big drive, but it started eating that up too. There may have been a more optimal setup, but eventually we switched to the Perforce indie license—and it has worked great.
Visual Studio
This was a no‑brainer. I’m a coder; Blueprints are great, but we were definitely going to write some code. Both Git and Perforce have VS plugins, and they were very helpful. GitHub’s Copilot was also a nice extra set of “eyes” and a fast research assistant when you’re solo‑coding. Sometimes it got a little weird with its assumptions, but it also saved me time by auto‑completing whole functions that matched my design patterns.
Blender
I’ve been playing around with Blender for about 20 years. I’m very comfortable there, and the industry—especially indies—has really embraced it lately. For FBX edits, importing and converting assets, and authoring first‑pass content, Blender was in.
Adobe Creative Suite – Photoshop, Illustrator, Fonts, Stock Art, and the Substance 3D Suite
There are definitely open‑source options for these things, but we needed to get moving fast and had a lot of work ahead of us retooling assets to fit our aesthetic. So we decided to pay for the industry‑standard toolset.
Articy Draft 3
We knew we wanted a strong narrative and didn’t want Lona to have to wait for me to develop a toolset before she could start designing it. I had played around with Articy Draft before, so it was an easy choice.
Logic Pro
Lona already had a Mac Mini and a MacBook Pro in her audio lab, and we didn’t need anything else. We might go into specifics on her process later.
Unreal Engine 5 & Plugins
Base Setup
- World Partition streaming open‑world level
- Nanite and Lumen for rendering
Custom Systems
User interface, character scheduling, calendar, day/night cycle, inventory, interaction, farming systems, puzzles, resource gathering, and weather.
Plugins
- Water & Buoyancy – An island needs the sea, rivers, and waterfalls
- Dungeon Architect – Procedural dungeons so there’s always something new to explore
- Procedural Foliage – Filling in forests and coral reefs
- AI Systems – StateTree, World Partition Nav Mesh
- Mutable – Procedural skeletal meshes for customization and variation
- Articy Plugin – Importing and integrating narrative data
- Mass Gameplay – Data‑first gameplay model to keep the game running while assets and levels stream out; road‑based navigation; simulation for all sorts of things
So far, so good. It’s been a heck of a first year. Let me know if you want more information on anything specific.
Discover more from Secondhand Carrot
Subscribe to get the latest posts sent to your email.
