This post is migrated here from my 2012 blog

Mystic is a new personal demo game project I’m working on, it’s supposed to be a multiplayer co-op variant of a space shooter, the exact details of gameplay are still not yet defined, but I already put a bunch of hours into it.

mystic

p.s. Mystic is the project code name, the actual name of the game is going to be different.

The highlight features of Mystic are going to be:

  • Multiplayer co-op gameplay, 1-4 players
  • Facebook integration
  • Player matching/game lobbies
  • Flash, shader-driven via Stage3D and AGAL

Technically, my implementation is going to be:

  • Design: fully iterative, I start with a rough idea and build gameplay experiments around it, keep what’s fun ditch what’s not, until I have a solid mechanic to build a game around and extend.
  • Platform: Flash 11+ and Stage3D with AGAL. Frankly I would go the Unity3D route if I had access to fullscreen shader-driven effects (or at least low-level access to render-to-texture), but the free version doesn’t allow either and I can’t afford the pro version for a mere experiment.
  • Shaders: AGAL, however by the time I start writing serious shaders I’ll look for a higher-level alternative, otherwise it may be worth it to do a CG assembly to AGAL converter (if that’s possible at all) and a few utility pythons to automate the process.
  • Graphics: 2D running on GPU with advanced shader effects (motion blur, lighting, particles, etc..), it should retain the soul of retro but looks and feels quite modern at the same time. I might go for something kinda like what Ethanon does: http://www.youtube.com/watch?v=Xmn6zhDJGLE
  • At first, I considered writing my own game framework on top of Stage3D, but then I discovered Starling and the fact it already includes most of what I had in mind like batch rendering and texture atlas support. But it lacks a main feature I need: shader functionality, by display object and fullscreen, so I have to add support for that myself, would still take a lot less time than if I started from scratch. Did I mention that Starling is open source? ^_^
  • Networking: this is still not very clear at the moment since the gameplay mechanics are not defined yet, my first choice would be Player.io but if I require high-frequency low-latency updates (aka UDP) Player.io might not be an ideal solution since it only supports TCP and apparently a maximum of 10 messages per second (100ms minimum latency) (related discussion). I might investigate other solutions like Heroku (if it’s capable of something like this at all), or perhaps directly running the server on a rented host. The problem here is that changing the server might mean changing the solution entirely, so if I picked something unsuitable and then wanted to switch to something else in the middle, I’ll probably have to rewrite the server-side from scratch.

Goals:

  • Creating an awesome game of course ;)
  • This would be my first personal multiplayer game, and first time doing motion prediction and latency compensation.
  • Adding full support for Facebook and Server/Client communication to my Actionscript 3 library.

CODE NAW!



blog comments powered by Disqus

Published

10 June 2012

Category

gamedev

Tags