My game takes places in a whole big house (is a graphic adventure). Since I don´t mind having a transition every time the players enter a room I can break the scenes however I want. What would be the best way to avoid making the player wait? Should I break it per floor? Should I break it per room? Should I have one big scene? (…or two or three…?)
I´d also like to know how many polys per scene for a desktop app would be ideal (I guess it has to do a lot with the texture too). I´ve encounter lots of threads talking about polys, but most of them adress just the characeters and no the whole scene.
Thanks,
Poly count is a hard art to pin down, that’s why you’ll never see anything too definitive on it.
I’d say aim for a duel core with 2 gigs of ram and a five year old mid range GPU. That’s a very large chunk of computers out there still in use. I’d say you should be safe ~100,000, but I could be wrong about the number there.
As for being a “graphic adventure”, I’m guessing that means something like Mist?
There’s lots and lots of ways you can go about making a game like that. You can use 2d pictures of rooms and video for transitions. You can have it fully 3d and let the player move freely. You can load the entire thing in memery and cull out the unused models. You can load the rooms dynamically (in fact there are some nice, cheap assets that can help you there).
Personally, I would lean on doing the culling thing.
Just do up the whole house as one big scene, then make them a child to a room object, then enable the room with a separate trigger.
If you start to choke on memory, you can have the items instantiate upon entering the trigger, but then you’ll get chugs and frame rate drop when they are loaded up and the garbage collector eventually kicks in. (It’s bad to have the garbage collector kick in. It’s basically defragging your RAM.)
Mmm sounds interesting. Just to make things clear, the game is full 3D, the camera will most of the time follow the character but sometimes it will be fixed in some place. Not Myst, more like Grim Fandango, Point n´Click (although in regard to this question is probably not much of a difference).
Ideally I would prefer to avoid transitions of any kind, specially those that have to load something and have the player waiting. The most I know about culling is back face culling, so I´ll have to check into that. If you have any documentation you can point me to on ways to handle inteligently a big scene avoiding transitions as much as possible, I could surely use them.
Sure, here’s the manual section on it:
But if you’re doing fixed cameras I’d most likely just enable and disable content along with the camera. Don’t really need anything special for that.
Just make a collider and make it a trigger. When the player moves into the trigger enable the camera and everything it sees. Then just disable it when the player exits the collider.
1 Like