My crosshair is a 2D GUI element. The enemy is a 3D guy. Should I somehow try to do the calculations in the vertex shader somehow, or make some complicated collision shape for the player and somehow check if the bullet ray intersects with the collision box. ( I have one math book for 3d games, I'm gonna figure it out how if I know that this is the best way ).

Guys, thanks for the clear answers. Knowing that I won't have to touch the shaders gives me hope, I'm really bad on getting feedback from them, so it's kind of hard to debug.

On the other hand, my game is pretty simple. I have neither bullets, nor a proper player model, I just have a moving camera that can jump and rotate properly, a 2d crosshair and an enemy that has few anims.

But the problem is that the enemy's shape is quite complicated because I exported the human from world of warcraft and sounds hard to me to actually generate a proper collision shape for the guy. I know that the easiest collision shape is a bounding sphere, but it doesn't really work for a human model.

Can you give me something that is good for a start and that gives not so bad results? ( I have no xp in 3d collisions )

First thing tomorrow will be to try a box for the body and maybe a separate sphere for the head in order to make headshots tricky. And then I'm gonna see if it works fine.

I really like the idea with the stacked spheres, will try that too.

Thanks.

If you have a physics engine, the physics engine can do a raycast for you from the center of the camera to the target.

If not, then you can actually break up the target into tight fitting hitboxes and cylinders. With a bit of Raycasting math, you can easily implement this feature.

Ok guys, I managed to generate an AABB for every enemy based on its dimensions, now how do I proceed?

I can't find any material on collision detection between a ray and a box in any of my books. This means that it should be easy, but nothing comes to mind, any ideas? ( Bear in mind that when my enemy rotates, the bounding box rotates with him, so I don't know if it still counts as axially aligned, and I don't really know if that helps me or not )

I can't find any material on collision detection between a ray and a box in any of my books. This means that it should be easy, but nothing comes to mind, any ideas?

Figure out how to compute the intersection between a ray and a plane.

I think I figured how this works mathematically, can you see if my calculations are right.

What I did so far:

First, I found a representation for the ray.

Q(t) = S + tV seems to be the best one for this purpose. ( Where S is the starting point of the line and V is the direction )

2. I found how to represent a plane.

I need only two things: a normal ( let's call it N ) and one point that belongs to the plane( let's call it P ).

3. I found how to check if a point is on the plane ( let's name it Q ). Basically, to check if Q belongs to the plane, I just find Q-P ( which is the vector pointing from P to Q ) and then I check the dot product with the normal and if the dot product is 0, that means that the signed length of the projection of Q-P onto N is 0, which means that the two vectors are perpendicular, therefore Q belongs to the plane. Cool.

( this is the formula: N (dot) ( Q - P ) = 0 ).

So, ultimately, what I need to find is the 't' variable in the line equation, which is Q(t) = S + Vt.

If we continue the above formula, and apply the distributive law, we get this: N (dot) Q - N (dot) P = 0.

and when we substitute Q with S + tV and we solve the equation for 't', we get this: t = - ( N(dot)S - N(dot)P ) / N(dot)V .

So if I can somehow tell this to my pc, I will see where the intersection happened.

It's late. I'm going to sleep. If someone bothered to read all of this, please tell me if I'm on the right track.

EDIT : I'm on the right track.