As it stands
The idea for this one has been around for many years, but it was only when I saw a video by Walaber about the physics of JellyCar Worlds that I thought it'd be worth giving it a go! Within a few days, using C++/OpenGL, entt and very little else, the basic idea was there, and it was amazing how easy complex topics could be reduced down to some fairly simple code.
The first version, a few years back, was dropped mainly out of frustration with the physics continuously exploding and being too inexperienced to fix them properly. Revisiting this a little later, it turns out that just a few TINY changes relating to edge detection was the entire solution. There are still issues where characters can go "into" another body (these are also the same ones demonstrated in the JellyCar Worlds physics video!) but on the whole, it means there's a decent balance of simplicity/performance and the all-important squishiness.
That last iteration was parked when I was faced with the choices of adding competition: either multiplayer, or bots - both of which are complex topics that (at the time) were a little beyond what I had the time/patience for, around a day job and family/kids! I got some very basic bots working, with A* and some acceptable aiming, but the crunching involved in working out targets, taking dynamic platforms into account, was quite rough.
This time around, I went straight in for multiplayer, starting with a quick standalone test using enet and libjuice, a Google STUN server and a little relay on my VPS. With that working, I went about splitting my game into logical server/client pieces, mostly as it was easier to follow what goes where! The current game is entirely server-authoritive; *intent* is sent by the clients (move, jump, duck/squish, aim/fire), and the server responds with a full (for now) snapshot of all dynamic particles and details of the body's material, and the client then just build the meshes from this data. The rest is just audiovisual events, e.g. explosion at X, Y of size S, fire or splash.
Next steps
The next big steps from here are likely:
- Multiplayer: currently, there is one "dev" lobby, I want players to be able to create their own lobbies (public or private) and join them. Plus there's definitely a lot of improvements in the snapshots that could be made, as right now, there are sometimes issues with the camera following/interpolation when things get hectic...exactly the time you DONT want these issues!
- Bots: whilst multiplayer was my main goal this time around, the bots at least provide extra chaos, and also the ability to play solo. But right now, they're mostly "charge, see, spam fire". There's a *bit* of an effort in there to try and pick an appropriate weapon, or to run away/avoid bad things etc, but certainly not suitable for the finished product. Also there needs to be some degree of difficulty setting, because right now, if they have a line of sight on you, you're done for.
- Level design: currently the levels are my simple test ones, where I've just thrown on a few platforms and let it be. I've not yet sat down and designed a proper level with consideration of the layout, pathways, etc that might rein in some of the chaos and allow for more tactics than a fairly open level!
- Mobile: one of the things I intended from the start was that it could be played on an Android tablet, which my daughter was insistent on! However, the engine I was building out didn't really have any consideration for mobile. I've done a fair bit of work to separate out the main renderer which will help, but it'll be a decent effort I think to make this entirely mobile/tablet-friendly. NVRHI is probably my focus, here.
- There are still a lot of bugs. Some of these are actually part of the charm, but they still need taming so that we don't end up with game/match-breaking situations.