I participated in the Ludum Dare game jam for the first time last weekend, putting my engine to the test, and came out with a not-too-shabby game. It was a ton of fun, I've lurked around Ludum Dare for a while, playing games from it occasionally, but it's a very different (much more interesting) experience when you're participating. Many coffees were drunk, many a cool person were met, a good break from the eternal lockdown tedium... Overall just a good time.

So what did I end up making, you ask? The theme for this jam was "Stuck in a Loop." I decided on a space-age FPS set in a "loop" ship, since I already had a first-person camera controller implemented and everything. The Ludum Dare page for it is here, and the webgl build of the game is live here, if you're inclined to try it out. Objectively speaking, it's not very good; things are broken all over the place, assets are all sorts of jank, but: it does represent a real, playable game with animations, sound, and UI, and so it's a big step for my little engine. Where it's at now is basically the minimum viable engine.

There are a few things I wanted to put in that didn't make the deadline, particularly midway through I decided that it should be a third person shooter to show off some skinned and animated models. I had made a full, rigged low-poly model, made some animations, and then realized I probably wouldn't be able to finish that, the 3rd-person view logic, and get everything built for webgl in the ~6-ish hours before the end of the jam. Alas... Something to plan for better next time I suppose, would be a good thing to have some more pre-made view controllers around.

In a similar vein, I had started off by making my own textures, but ended up using a few things I had made before the jam, which disqualified me from being judged on graphics... In retrospect, maybe I should have leaned into that and used more premade textures and assets. I'm basically making a demo for my engine, after all, and I'm not a very good artist ;)

Another thing to note is from the feedback I got on the game, it seems a lot of people are actually using lower spec'd machines than my own. This is something I didn't expect honestly, my aging potato was low-end when I built it in 2015, but I suppose many people are playing on laptops... Definitely something to keep in mind in the future. While the webgl pushed the limits of playability for me, it's basically unplayable on anything less, which means I'll need to consider having different graphics settings and providing native builds.

Regarding the engine, I spent the month or two leading up to the jam cramming things in. New features the engine has:

• Node-based object trees, json-based map format
• Basic framework for asset streaming, everything is now reference counted
• Animations, skinned meshes
• Sound mixing, HRTF implementation merged in
• Vector graphics UI with nanovg
• Structured like a proper library now, allowing for out-of-tree builds
• Proper input handling layer, allowing for reusable controllers, simpler state management
• Configure script for setting install locations, openGL target, etc.

Much more capable than it was in June, I'd say! There are, of course, still a few very limiting factors, notably no sort of light clustering, and no geometry culling. I started to run out of time leading up to the jam, and so had to prioritize user-facing features, but these two things really limited the complexity of the scenes I could make.

So, naturally that's what I'll be working on next. The roadmap from there is to integrate bullet physics, work on asset streaming, and work more lighting and postprocessing stuff (light/reflection probes, SSR, antialiasing, maybe path tracing if I'm feeling particularly ambitious). Honestly not sure how far I want to go with the engine, but I'm enjoying working on it, so no reason to stop now.