Hi all.

I've been in the design process of a JRPG style game for a long time, but we're a little lost on the tools to use for an effect that is relatively obsolete nowadays. My friends and I have some experience working in 3D engines (we had originally planned to use UDK when we started out, the initial idea was to use full 3D).

The idea is to create pre-rendered environments as seen in the PS1 Final Fantasy games. I've searched this forum and found a couple of threads on the subject, so I have a basic idea of the principle involved. My question is whether anybody knows any engines that support this kind of thing? Or at least a point in the right direction. Giving fake depth to a 2D image doesn't seem like the sort of thing that would be particularly easy in UDK and it certainly doesn't have much coverage on the forums!

The other problem is that we do want our battles to be in 3D environments, true traditional JRPG style. So a purely 2D engine wouldn't do the trick either.

Thanks for any help :)

Fox

I'm not an expert, however, I believe you can achieve the same effect by drawing your prerendered image as a quad, and then overwriting the depth buffer with the image's prerendered depth buffer. This might be achievable through the use of shaders, and most 3D engines have shader support.

Hope this helps. [smile]

I don't know of any engines which would straight up support this feature but it shouldn't be too hard to implement. Build the initial environment in your 3D editor, and also build the 3D collision layer to fit it. Fix your camera angle and there you have the starting point.

In the engine you set the camera up exactly as it is in the editor. Load the collision and there you have the 3D world ready to go. Render the background as a full screen quad then render the 3D objects.

If you want simple parallax (some objects in front of the player, some behind) then simply pre-render the background into two images. One you render before the 3D models, the other you render after. If you want more complicated interactions (scenery you can walk all the way around for instance) then you would add that object as a billboard into the scene.

Thanks for he responses :)

So, the principle behind this is that my characters will effectively walk on the complete 3D environment as made in 3Ds Max. But no actual model is loaded, only the collision information. The pre-rendered image of the environment is then placed behind this 'invisible environment' which, from the correct camera angle, gives the illusion that the characters are walking on it?

Have I understood the concept there?

Yep bang on. You will probably have other things in 3D as well though - pickups, audio emitters, particle emitters, exits, dynamic lights, etc.

Probably be a good idea to be able to toggle drawing of the collision mesh on and off so you can check if the background lines up.

Quote:Original post by Noggs
Yep bang on. You will probably have other things in 3D as well though - pickups, audio emitters, particle emitters, exits, dynamic lights, etc.

Probably be a good idea to be able to toggle drawing of the collision mesh on and off so you can check if the background lines up.

Thanks a million! I read something on the UDK forums about how to create a collision mesh for an environment in that engine, although not sure about the specifics of importing ONLY the collision mesh without an environment attached.

Anyway, that sounds like a much smaller hurdle to clear. I'll post results if we get a prototype working using this method :)