In a first person single player game, is it ok to have the PC use a very high res model? I understand that more polygons mean slower performance and such, but is that more polygons that have to be rendered or loaded in memory? A model with a high poly count would slow performance when the player is looking at it i know, what about models out of view, like the player himself? It would matter in such cases like when the player can see his own arms and hands when performing an action, as that would be fairly close up and would benefit from high res.

Well, for the player himself, you typically only have a mesh for his arms and his gun. you usually don't see the rest of the body so you don't need a mesh (you never see the chest, so why model it)

Otherwise the max mesh resolution you can have depends on:

1) how many other polygons you have in the scene (small indoor maps = higher polycount per unit area, large outdoor maps = lower)

2) the specific engine you are using: some engines are optimized for high poly counts, others not so much

Outside of that you will typically have Level of Detail change depending on distance from the camera. as an object gets further away you snap/blend to lower resolution models (you can't tell the difference between a high and low rez model at far distances). So you're always higher resolution near the camera.

-me

Usually the player doesn't see himself. He would only see the full model when looking into a mirror. I would guess it depends on your definition of "high poly", though. You can probably get away with a few thousand polys using most engines, but even then it depends on the engine. Poly count is a very subjective thing that has lots of factors. Are you using pixel shaders? What is the highest supported resolution? Fill rate is more likely to be a bottleneck than poly counts.

Usually when wandering around in first person mode, most games have a completely separate arm and gun model they render. This way they can delete faces that can't be seen. If you have ever opened up any models from Half Life 2 you'll see that the gravity gun, for instance, is really just half a shell. You can use tricks like that since the model rendered as you are carrying it can be different that the model rendered when it's laying on the ground. This enables you to have high-detail on the gun in your hands and use a lower poly count one when it's on the ground.

Quote: Original post by JasRonq
In a first person single player game, is it ok to have the PC use a very high res model? I understand that more polygons mean slower performance and such, but is that more polygons that have to be rendered or loaded in memory?

Both actually, more polys to be rendered when the model is onscreen, but also more polygon data that must be kept in memory.

Quote: It would matter in such cases like when the player can see his own arms and hands when performing an action, as that would be fairly close up and would benefit from high res.

Yes, you would benefit for having the arms be higher polygon/texture res. Just calculate the polygons being used for the arms into your overall poly-budget (marking it as something that could be taking up polys at any time) and you'll be fine.

I will see about mixing the hi and lower res models i have in Poser together so I can keep hi res on the arms. If there are only arms and what the hands are holding in some games, how are collision detections dealt with??

What do you mean by how collisions detections are made? A simple cylindrical bounding box can be used for collision detection for the body, unless I'm misunderstanding your question.

Usually characters are collided using a bounding volume, not their render mesh. So you'd use a box, sphere(s), cylinder, etc.

ok, i get it, i rather assumed it was closer to the mesh than a bounding primitive.