View Full Version : Crossplatform... .how much agro?

01 January 2002, 09:02 PM
We're going into our first parallel crossplatform dev effort (we've done games on the individual platforms already), and I'm getting handle on how bad this whole crossplatform thing is for us (artists). Anyone wanna share their experiences?


01 January 2002, 01:19 AM
your milage may vary, but here's our experience:

Our first game was developed for XBox. It's a toon shaded game, with a fair bit of geometry/textures/physics being thrown at the system. Developed with 17 in-house staffers and multiple contractors for cutscenes and audio.

Late late late in development, we decided to port the game to GC. GC is decidedly less powerful than XBox in a lot of it's abilities (less memory for textures, lack of pixel shaders, etc.). This meant a lot of hacks (for lack of a better phrase) had to be implemented to get it to look great and run at a decent clip.

Now, the fun part:

Our GC port was done with almost NO artists dedicated to it. I stuck around to answer any big questions, but content changes were handled automatically by our code dept. In fact, the end product is strikingly similar to the XBox version! So that was a _good_ experience in the sense that our content was built well enough to survive being shoved into a less powerful system.

We're now on our second project, and it involves simultaneous cross-platform development. A lot of our new Art pipeline involves finding the common denomenators of each system and creating content that can transfer easily to the other platforms. Examples:

We've started generating source texure content at 16 colors to allow it to trickle down to PS2 without wasting a ton of memory. This buys us a lot on the platforms that have more memory (we used 12MB of 24bit textures on the XBox for an average level imagine using 4MB of 4bit textures and using that other 8MB for more geometry). Plus, since we're toon shaded, we don't really "need" color gradients.

Geometry-wise, we're experimenting with procedural LOD generation...but the results are only so-so right now (looks like we're gonna do mesh-opts the old fashioned way!). Nothing beats an artist's eye when it comes to LOD generation. That's an interesting one. Smooth skinning, which platforms can handle it and at what cost? We're still figuring that one out, but I think we've now got a method that will give cartoony, smoothly bending figures on all platforms.

Overall, it's not as big of a headache as I thought it would be. A lot of the hard work is being done by the coders, but we attempt to build our content intelligently and automate as many tedious tasks as we can. We learned a lot of lessons on how to do that with our first project. I can discuss the finer points in detail, but I'll cut this post short (?) for now :)

CGTalk Moderation
01 January 2006, 01:39 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.