I’m about to embark on a university thesis project of that could potentially be very large in scope. I’m doing research on what technologies would be best suited to my needs.

I’m considering Flash because I’d like to take advantage of its asset management system and garbage collection, but I’m not even sure if Flash will be able to computationally handle what sort of things I need to do. I was considering a combination of C++ (for native speed) and C# (for timely iterating), but I’d prefer to avoid a low-level approach (because the scope of my project shouldn’t be consumed with optimizations). As a middle ground, I was considering the jMonkeyEngine (Java game engine using LWJGL- java bindings of openGL/CL/AL et al)- Flash just may not be computationally capable of doing what I need it to do and a C implementation is a PITA. Despite this, I’d like to use Flash and Away3D is the only legitimate option.

To give a clearer idea of the sort of stuff I’ll be doing- The premise of the project is to procedurally generate assets, environment, and ‘rules’ through which players must interact with one another in a multiplayer (not more than 32 players) environment guided by procedurally generated objectives to incentivize interesting social interactions.

The rules generation component of this project deals heavily with the use of color and textures- It will be necessary to generate a wide variety of materials on the fly. Including spec/diff/norm/bump maps, some of which I would like to manipulate/noise up/apply decals- all of it subject to being algorithmically generated/manipulated. What sort of capabilities does Away3D/Flash have in regards to generating materials? I’d like the capability of using animated materials as well. I know flash has a few blending options, but I know work has been done independently to bring those sorts of options more in line with Photoshop-esque techniques, does away3D have anything of the sort?

I’ll need to be able to perform weighted morphs between sub-geomtries of objects during runtime (this will NOT be interpolated). Player meshes will be composed of mesh parts that depend upon decisions made by the players. Part of the rules generation of the game involves creating content patterns that use irregular shapes. IE- suppose I have a mesh of a pig and a dog- say these meshes are made up of nearly the same amount of vertices. I’d like to do something like this- Pig*.66+Dog*.33, and get a PigDog model (obviously, the asset development requirements will need to be very particular). It’d be achievable in straight up openGL, but I want to know if there are any classes in away3d that I could expect to be helpful with this sort of thing.

Does away3d have animation blending? Is there a socketing system? Ragdoll physics is completely unnecessary, but it’s something I’d consider further down the line- any sort of work expected to be done in this regards?

I’ll also be procedurally generating simple environments- according to the API, it seems like I could dynamically create sub-geometries to algorithmically produce a geodesic dome, buckyball, or winding caverns—would these sort of actions seem realistic in away3d? If I wanted to generate multiple glowing static meshes, what kind of hit on performance would I see? I’m just not sure what the VM Flash uses is capable of- or if I can expect to see near-native performance on light-calculations.

I know stage3d was just released and away3d is expectedly still very young, but I like the looks of the project and would be interested in getting involved if it’s suited to my needs. Please share your thoughts- thanks.