[DaBrain]

I'm exporting a placeholder ship right now and it is still very painfull to set up each turret with firing points, axes and fp banks.
The most annoying thing is, that it's actually always the same turret... and I'm copying and turing everything for the next turret again.

In the end I have to set up the weapons in the tables as well.


Wouldn't it make a lot more sense to have different turrets models and just have a standard setup of them in the ships.tbl, while we could still exchange the turrets in FRED?


A little bit like the attached missile models on fighters work. You would define the turret positions and just attach any turret you'd like to them.
That way you could have two completely different armed and slighty different looking ships of the same class in a mission.

Also, you'd only have to set up one turret per turret class, meaning preparing a capital ship for in-game use will be a lot easier, faster and errors in the setup are way more unlikely.

[chief1983]

Didn't a recommendation about this show up a few weeks back?  Something else about putting turrets in separate POFs and attaching them to the ship via tables?  Would actually be a really awesome feature and you're right, it would make setting up larger ships a whole lot easier I bet.  Of course, actually placing them seems like it would be a pain, but if it's been done with the missiles I'm sure the turrets would be about the same.

[FUBAR-BDHR]

I was talking about making them separate ships and docking them a few weeks back. 

[Commander Zane]

Wouldn't that just wreak havoc with the ship limit?

[FUBAR-BDHR]

Well it was only for use in special circumstances not for what DaBrain was talking about.  It just might have been what Chief was recalling. 

[Aardwolf]

This always annoyed me back when I used to try to model capships.

Tentative 

[DaBrain]

Well, attaching some other stuff would be nice idea as well.

I also hope this will save a little bit of memory and later some rendering time.

Every mesh in a POF takes up space in the v-ram. It depends on how many turrets are on the ship and how complex the mesh is.
By using on POF instead of having all of them in the ship POF, we could save some memory.


Later with geometry instancing, it might even reduce the rendering costs.

Wouldn't that just wreak havoc with the ship limit?

I hope they wouldn't have to be in the ships table, or in any other table at all.

Like the skybox POF, they should just be loaded when they're needed.

...I doubt that will be easily possible though...

[chief1983]

They would be a separate model altogether though, and there may be a limit there as well, as in the number of .pof files that can be loaded at one time during a mission.

[Aardwolf]

The way I see it, they'd be separate models, but wouldn't require table entries of their own (since ships.tbl already covers that).

Later with geometry instancing, it might even reduce the rendering costs.

 

[chief1983]

I'm not talking about anything related to table limits, but the number of pof files that can be loaded at one time might be limited too.  That would involve all sorts of things including separate cockpit models, etc.

[Nuke]

ive been asking for this for years.
you could also lod them based on distance to player rather than players distance to parent. you could have many more turrets on a ship and only have the closest few show at full detail. you would also get around some of the sorting issues with the ancient way freespace deals with subobjects.

the turrets themselves arent technically ships. they would probibly be an object derivative, but something new. turrets could also be modeled on plane and rotated to the proper orientation, so all turets that look the same, but mounted in different places such as stop bottom or side, and would still have the same model.

of course i think such a feature should work for all submodels, and not just turrets, since it seems to be faster way to render stuff. of course such a system would need to be scriptable and animation happy.

[Fury]

This would be all sorts of awesome.

[Nuke]

it may also speed up collision detection.

[DaBrain]

I thought it might make the collision detection more complex, as the turret/module/whatever would have to use it's own collision in addition to the ship collision.

[chief1983]

I think having two models with independent collision models might be easier than the way it's done now, with the main model somehow having to encompass all subobjects.

[Sushi]

IIRC from the collision code, ALL submodels with a bounding box that overlaps the test points will get fully checked.

I THINK this means that the more submodels you divide the ship into (in general) the faster collision detection will go. Maybe.

EDIT: I'd love to see an empirical experiment on this...

[DaBrain]

That is interesting... completely different from what I thought.


I'd really like to understand the collision system in FS2 better. Is there any kind of documentation on this. Something an artist can understand?


About the empirical experiment...
Fine, we'll do it.



Anyway, how 'possible' is my request?
I won't give in to any illusions of this being simple to do.

[Nuke]

since turrets would be about the equivalent of objects, you make a ship, which is essentially just a turret. then take some large capship pof, and remove all the turrets from it. place it in a mission and place the turret ships where the turrets used to be. for control make alpha 1 an ai and have it follow a waypoint path, having to shoot at various points along the hull with different weapons. test with various weapons and missiles, different fire rates and velocities. then write a little script to count frames. if the ship is in mission the same amount of time, the the faster meathod would have the most frames under the same duration.

[Aardwolf]

But, unlike the way you (Nuke) explained it, the turrets would not need to do collision checks between each other or with the ship they are attached to.

[Nuke]

without actually coding it to do that my experiment wouldn't be able to account for those variables, so you could think of it as "brownie points" or "gravy". if there is a major performance increase as far as collisions go, then it could only stand to improve it, by reducing the number of thins to check against

i had an old idea a long time ago, before scripting really matured. for developing a virtual turret script. these would technically be ships, in that they had a base model (the hull), as well as the turret model (which could move if it was a multipart). at the time i didnt have access to the data i needed to make it work and shelved the idea (along with the whole turret script). now the scripting system is much more capable of doing it. not to say that doing it with scripting would improve performance any. in fact it would probably make things worse and cause all kinds of collision and physics issues. it would also make using them in missions rather difficult because there would be little or no fredder control over them.