Your game sounds fun already. But I don't think the model format is going to limit your ability to include those features. I have seperated wearable skinmesh clothing and armor, static props like helmets, static weapons like swords, and dynamic weapons like bow & arrows. It's all currently running with the x format. But the mesh or format had nothing to do with how I had to program the features.

It's just all about linking abjects together in your code.

hi there, I would suggest going with any model format you feel comfortable with. If you can reformat it to build a mesh in DirectX all the better.

With regards to the body parts... I would suggest have a different number of body parts so that when you pick up an item that is viewable on the model itself that you swap out that body part with another with that particular item.

Good luck.

X files or the way you use x files is not changing. Only the API helper methods used to load the format changes. Which would usually mean a small change in your code. Not exporting all of your meshes again.

Most changes in API's will require changing code. So I don't see where this is a disadvantage with the x format at all.

There are probably a bunch of ways to achieve it, but here's one that makes sense to me. I'll use weapons as the 'swap piece' example.

You have your weapon models - a sword, a shield, and a bow say. You also have a separate character model, and it has a bunch of different poses - one for holding a bow, one for a sword and one for a sword & a shield.

Now, you can invent and place special 'tag' polygons in the model at key locations - such as the character's hands, and you can put the same on the weapon models - you can then match up these points of reference when positioning/orienting the weapon they are holding. It's up to you how the tag polygons are designated and identified etc.

That's the basic outline of how something like that might work, and I think Quake3 did something similar - I believe it seperated the torso, head and legs as well and used tags to orient them all, so as you can see it can be extended as much as you want. Although Q3 didn't use skeletal animation, but hopefully that gives you some idea :)

Good luck,

-Mezz

As superpig said for weapons, etc you really want to use a tagpoint type of system.

For swapping armor, etc what I do is have the modellers create two seperate torsos on the model and I hold a list of the renderable parts of the current model.

There are easier, less clumbsy ways to position weapons and objects onto characters. In most cases, you're going to need special animations for different weapons, so you could easily use a prop in your animation program/editor. This allows you to animate the weapon independently from character bones. Much more important for swords and shields than it is for guns.

Even if you don't have special animations for every weapon, you can have partial bone setups which animate only the bones required to hold the weapon. In my engine, two handed rifle postures use an animation which only targets the arms and weapon prop. So I use normal animations for standing, walking, and running, but I override the arm movement with that rifle-hold animation. Things like knives and pistols would really only need one hand. You're most likely going to find yourself splitting up the animations anyway. For example, a character firing a bow as he walks. He needs to animate his arms, and perhaps head and chest, independently from his lower body. Providing this kind of flexibility allows any weapon type to target and isolate specific bones, allowing other bones to default animating with the unarmed posture.

I have this stuff working, so I can give you info along the way if you need it.