definitive proof that animations are a result of computer calculations.

[Musashi]

For homeworld 1 though, the collision detection was only necessary for a max maybe 200 ships on screen while in a large battle on m2tw, there would easily be 300-400 arrows flying around every few seconds.

To be fair, most of those ships had multiple turrets, and their projectiles were all being tracked for collision.

[Carl]

Thanks, I was going to say the same Musashi. Plus the game had full collission detection on every ship, (you could even accidently kill a ship if it ran into another one), and spacing between ships and other stuff seems to be controlled by it and the formation/stance selected. The stuff was also often moving at high speed.

hell in HW2 a large custom battle could have 30 bomber squadrons running round along with 30 gunship corvettes, and 30 frigates and a similar number of BC.

That totals well over 2000 pojectiles and a hundred ships on EACH side out of a possible 6 sides, and I don't think anyone ever expirianced slowdown on a modern spec system.

[Musashi]

Oh and Drone Frigates had something on the order of 50 gun drones each... and the drones themselves were checked for collision as well as their projectiles.

[Carl]

Yeah, 24 drones actually, but they each had a gun and both the guns, the drones and the frigate where all checked for collission.

I'd say collission detection isn't THAT big of a drain based on that myself.

[stev666]

The fact that MTW2 on occasions due to cpu capabilities wont alow all participants in a battle to be on screen at the same time, i.e reinfocements being "delayed". would suggest to me that to add collision detection on a mass scale also would futher reduce the games already stretched resources. Of course thats only in my experience as I have an intel pent 940D.

Wtf is up with this typing intereface! I cant even go back and retype stuff without typing over the words ive just typed.

[antisocialmunky]

I remember shogun getting funky in missle battles when alot of things would land all at the same time. This was on a 600 Celeron BTW so... yeah.


Keep in mind however that all games have different calculation loads to run through. Homeworld might have had alot of power dedicated to collision but it didn't have or need a great pathfinding AI. Infact, I still remember losing frigates in the Garden of Kadesh and the asteroid belt. They weren't very smart. Plus, you didn't run that many shading operations and other stuff to boot as well as calculate combat between 1000+ individual units.

TW on the other hand, has to calculate that...

@stey - Hit your insert key, it toggles between insert and overwrite.

[stev666]

I tried the insert key but it hasnt made a difference, cheers anyway.

[Carl]

@antisocialmunky: true enough about HW1 doing less, but this is less so with HW2, and they both had to run on much worse systems than TW does. So even though the other requierments have gone up with regards M2TW vs. HW1. So too has the processing power. i think they where still on P3's back then wern't they? (1999 was HW1's first release according to the re-release box cover).

[antisocialmunky]

So too is the processing power that is needed.

Weren't we talking about hit detection and not efficiency a page ago?

[Foz]

Well I think the point everyone is trying to make is that given the immense power of current processors, you cannot simply assume hit detection doesn't exist due to the processing power you assume it would require. The analogy is drawn to an older game that is known to employ hit detection on a very large scale, in order to demonstrate the feasibility of applying hit detection in M2TW after advances have been made in computing. So essentially it's in debate to prove that hit detection would indeed be possible, and so the discussion of hit detection versus simpler calculations is in fact non-trivial, as both are feasible options.

[Zenicetus]

Well I think the point everyone is trying to make is that given the immense power of current processors, you cannot simply assume hit detection doesn't exist due to the processing power you assume it would require.

This is true, however if you double the size of the armies (easy to do, with this game) and don't see a significant slowdown in frame rate, then it means one of two things:

1) Either the game is doing full 3D collision detection for all models, including projectiles, and it's just a super-efficient algorithm where you can't see a slowdown by doubling the army size. Or...

2) The game engine is cheating. And by this, I mean "cheating" like a WWII combat flight sim (or maybe Homeworld) that calculates bullet streams instead of individual projectiles.

I think it's #2... but it would be nice to hear from the CA devs about this. I'm no game dev, but I've been involved in enough coding projects (as a front-end GUI designer) to know that efficiency of CPU cycles is something that coders care about.

[Neoncat]

1) Hits seem to be checked against bouncing boxes not the meshes. This speeds it up considerably. So its like lines and boxes only.
2) You don't need to check collision when arrow is higher in the air than tallest unit in battle.

Don't know about 2) ,but at least they have done 1). Don't want to test a thing with this game. Actually I deleted it completely and I am going to wait for atleast half a year to get rid of those blood boiling bugs.

I could make a demo about 'single-point' collision detection in work, if I just get enough time from my studies.

[Sardukar]

They most likely use bezier splines (curves) for archers. And every archer has the same arc (sometimes two different types). You could do something like have a "central" imaginery arrow that has its arc pre-determined from the squad to the current target. Then work out the position of all the arrows determined by that imaginery arc (offsets).

You have like a couple of hundred arrows (MAX at once) flying in the air; it sounds alot but im sure they could optimize it. Then checking hitboxes isnt a big deal at all.

[antisocialmunky]

I was just saying that Option A was better than Option B because A is more efficient.

[Foz]

I was just saying that Option A was better than Option B because A is more efficient.

Better from a programming standpoint, of course. And while I'll heartily agree that efficiency is a primary concern of developers, gaming also requires that developers do everything possible to make the game seamless and natural to the players. In this case, it is obvious that hit detection is a far more realistic and natural representation of combat than would be arbitrary kill animations generated strictly from a mathematical engine. As CA has gone to such great lengths to model things so realistically in the various other aspects of the battlefield (Jerome recently commented that in RTW already they had implemented situational flanking bonuses to account for the field of vision of a character, and armor being weaker in side and rear areas than in the front), I am inclined to conclude that if collision detection was at all feasible (which it seems it is), then CA would choose to implement it in their quest to achieve a more true-to-life experience of battle.

[Aedile Rich]

The arrows, mid air, changed direction by about forty degrees and shot straight towards the raiders. I repeat, the arrows changed direction mid-air.

The only conclusion that I can draw from this is that...

... it was windy?


Thank you. I'll be here all week.