It's a lot of time that I'm looking for a very good way to implement a 2.5D game with static cameras, pre-rendered backgrounds and 3d player, and that's the reason because I'm here
, but before a bit of story...5 years ago I developed a very little prototype with OGRE (but not a good prototype), and than some month ago I started to create a protoype of this system using JMonkeyEngine (I'm mainly a java programmer, but I've also some knowledge of c++), that is very similar to old alone in the dark episodes.
Result is good, in pratical I load inside the scene just:
- the minimum required for collision (AABB boxes and walking areas);
- a stack of 2D images for layers (backgrounds, midgrounds and foregrounds).
- all scene cameras settings.
Walking areas are divided into sub areas, each sub-areas will be mapped into an XML file, where for each geometry ID there is an information of the depth to assign to the 3d object standing over that area, for the current camera view. At this point the render queue will be filled with all 3d elements and layer sorted by depth value.
In this way I can simulate in a good way the 2.5D feature.
(For interested people I can give some more detailed information)
BUT
There are still some problems. Infact I don't know whether this is the right standard for manage 2.5D, because probably I can do it easily with a pre-rendered depth map and one image only for view (Isn't it?).
THE QUESTION
I will not leave the work with JMonkey, but I'm interested to create a more performance and light version of this engine with Irrlicth engine, so for this I've some question before starting.
1. Is it possbile? (I think yes)
2. (MOST IMPORTANT) In your opinion which will be the best solution for performances, low memory usage and size? (Layers + pre-rendered Depth Map? Or not?)
3. Is there any documentation or examples about 2.5d? Because I'd like to follow a standard, but is difficult to find some specific information.
Thank you in advance for help
Ray