Air collisions

If air units are not allowed to bunch up then it can't happen and there isn't a problem.

Avoiding the question again, it's amazing how you can keep dancing and moving the goalposts of the discussion around so much, it's like you aren't actually interested in discussing anything at all!

Mike

How about "those are problems inherent to 2 large groups of bombers attacking one target", and fudging the system with magic noclip planes just to avoid those problems is undesirable.

You don't hear people complaining that it's inefficient to attack a single target with 200 t1 scout tanks (they collide, get bunched up, can't all fire at once, can be picked apart by defenses), it's just understood to be the case, and I don't see why aircraft get to be exempt from those problems.

Because ground units can go to a full stop. Whereas planes usually don't do that.

Its really as simple as that. You can't just apply ground movement to air plane movement.

TA (and to make a justifiable leap, PA) planes were/are all VTOL and at the speeds they were going at, slowing to a stop is totally fine.

Applying ground movement to airplane movement is exactly the problem here, because ground units have collisions to mediate the pacing of engagements, while air units move like ground units but ignoring terrain and other air units, which makes lolblobs an issue.

What, then, is the difference between a gunship and a plane? I highly doubt you'll see VTOL bombers that pass over buildings with a large turn radius and all, because that'd make zero sense.

It's a problem with a single group, let alone two.

It's not a question of efficiency; if bombers can't clip, then the only way to attack a target is single file, from a single direction, as they need to be directly over their target. That's easy to defend against, and adding more bombers to an attack would not increase the chance of successfully destroying a target. You've just made air useless.

First off, don't put words into my mouth, nowhere did I propose a fix, or say that to allow clipping is a 'solution', I am merely talking about deficiencies with bmb's proposed system. My prior comments regarding clipping are merely in reference to why it works, you can read this post to understand.

Secondly the idea that one can compare Land and Air units as you do is completely flawed. It's true that at the most fundamental level they are very similar, they can move from point A to point B, but there is so much more at work, so lets look at some of that;

Weapons: Land units primarily have weapons mounted in Turrets, often with a full 360 rotation, and in most cases attack from a fair distance(especially considering speed*) whereas Air units have primarily Hull mounted weapons that require the unit to orient towards the target. Often should an air unit have a turret weapon it will have a much smaller fire arc when compared to land units. Another exception is Homing weapons, which do not strictly require the unit to face it's target. Air units are also unique in that they are required to make 'passes' in order to repeatedly attack a [static] target, often this has the effect of imposing an arbitrary rate of fire based on how long it takes the air unit to turn around.

Movement Profile: Air units are significantly faster than land units and are significantly less maneuverable, Tanks can turn on the spot and wheeled vehicles can make much tighter turns than an aircraft could. Land units move at a much slower speed and have no requirement to keep moving, it is much easier to them to speed up, slow down and stop and can even move in reverse. While some faster units may share some principles that apply to an Air unit, the air unit in terms of actual stats is still in a league of it's own.

It leads to each unit type having unique needs from a pathfinding perspective to the point where you can't really make blanket statements apply to both aside from the extreme fundamental level I mentioned before.

Believe me, I would like a system that doesn't allow for clipping, as I said before, based on the SupCom Implementation, it's more or less the last remaining 'problem' to be dealt with, but the suggestions I'm seeing in this thread show a certain lack of consideration for the full potential scale the proposed systems will need to work in. As I said, regardless how 'improper' or 'bad' using 200 bombers to attack a singular target is, as long asn people can amass 200 bombers it will happen because someone out there thinks it's cool.

*As an Example, lets compare the Titan the Wasp from SupCom:FA'

The Titan has a Move speed of 4 and a Range of 20.
The Wasp has a Min. Move speed of 8, and a Max of 25, but only range 30 on it's weapon.

Technically the Wasp has a larger range than the Titan, but because it moves 2-6 times as fast, it's range is effectively/comparatively shorter despite the raw stats.

I think we need to think of the unit's VTOL capabilities akin to the Harrier's, after all, its Vertical Take Off and Landing, not Vertical Flight.

And Spreading them out doesn't really solve the issue either, if anything it means that more AA could fire upon the bombers and they still to pass the same general area directly over the target.

Mike

True, I know that, but then I fail to see what the difference is between planes as they are, and these VTOL planes...?

Someone was claiming stopping in midair(but not landing) was okay because the planes are VTOL capable.

Mike

I personally think of clipping as a backup plan. I would personally go for something like a flocking algorithm (like boids: http://www.youtube.com/watch?v=EZEK1lXxFfA) In one of the older topics someone already posted this vid.

Basically your aircraft follow a couple of rules:

-user input/flow field (polar coordinate 2d)
-cohesion (polar coordinate 3d)
-seperation (polar coordinate 3d)
-allignment (polar coordinate 3d)

cohesion holds units together, when multiple craft (of the same type and/or team) are near each other they try to go in formation, after that, using alignment which steers all the craft that are near into the same direction. When craft get too close they separate and steer away from each other.

Finally you have user input which gives 4 possible positions in total for a single craft to want to go to for the next frame. The only thing you need to do when you have all these positions is assign importance to each of these positions depending on the distances between craft.

For example:

user input/flowfield = 0.5
cohesion = 0.2
separation = 0.1
alignment = 0.2

Then you can calculate the final desired position to steer towards by computing these positions with their importance to one single position.

This way you do not have to check actual air collision, but still have a way to avoid clipping. Note that craft may still clip in extreme conditions, but there is almost no way to design against those.

Then again, if Uber would go for the traditional approach, I would not really mind, but I would be interested in how they want innovate even further.

Flocking algorithms are good for movement, but once the blobs of aircraft engage each other it becomes much harder to get good behaviour during actual combat.

Also, it solves a problem that's already solved, while not solving the one that isn't; formation flying can already prevent clipping during normal aircraft movement. The problem it doesn't solve is the bombers (which have to converge on a single point).

For this very specific case only a subset of units should be able to approach the target at the requested fire distance, having remaining units flying around this subset of units. It all depends on the type of weapon the units have.

It's basically the same for ground units except that you only have to deal with 2 dimensions.

With Air units that would mean having to deal with 3 dimensions.

Now if you want to be able to attack with 200 units, then these units should be equiped with long range weapons.

They do?.. They have to?.. They Must?..

Should they?

(devil's advocate)

Lots of clipping construction planes in todays livestream, but their movement in general was obviously not finished.

If you look at the way that bombing worked in WW2 the planes would be in a formation about as wide as their target but they would also be in a very long line. Which wasn't so great for the guys at the back of the line who got to fly over the target last, after the AAA gunners were all alert and at the ready.

So if bombing was more of an area thing ala WW2 style this would help a bit with stacking I think. It would also let you have big slow area bombers with massive payloads for strategic strikes and smaller faster bombers with small precision payloads that do far less damage for tactical strikes.

Why the focus on pathfinding?
Planes in TA and SupCom had no pathfinding. They just flew straight against their destination. I wouldn't really call the formations in SupCom pathfinding either.
Making colliding aircraft just push away each other on collision is a simple way to avoid clipping. Although it might brake the immersion if 2 planes in a head on collision just slides ontop of each other.
Knight mentions that bombers converging at one point might drop bombs ontop of eachother. Well that is only a problem if bombs collide with friendly units. If the bombs just pass through friendly units it is not a problem.
Maximizing firepower is then a matter of using formations so that bombers doesn't prevent eachother from aligning with their targets by being pushed by other bombers.

Making proper pathfinding for airplanes is a lot more complex than for ground units as several people have already pointed out. The relative high speed to acceleration and turn speed means that airunits have high turning radius and they have to be able to determine, not just where they will be but also where other units will be.

I don't see the need for pathfinding for air units that take other airunits flight paths into account when determining their own flight path unless air units take damage on collision.

If you have so many bombers so thickly packed that they can't avoid dropping bombs on each other... why refuse them the opportunity of blowing themselves up? It's just one of those things that
A) looks cool and
B) Provides an almost necessary element of diminishing returns in a natural way.

Sometimes there is such a thing as overkill.

Pathing isn't really as important as smartening up bombers is. TA bombers couldn't do shiat without user input, which is why the Hurricane ended up more useful for its laser than its bombs. Supcom bombers similarly had a single mind when it came to bombing, which led to huge amounts of overkill and dumb choices that demanded micro to overcome.

Smarter bombers take less user input, and get their work done more consistently as a heavy weapon's platform. Consistency is good for understanding and tweaking balance. Also, if a thousand bombers spread over a base instead of converging on a single point, then pathing becomes less of an issue.

they sorta moved like the construction planes in TA which made me swell in reminiscences.
I'd say they should keep it like that.

Yeah a proper formation can usually avoid this as well. But usually you want to spread out to avoid splash damage as well which also gives diminishing returns and gives the AA more time to fire.
Gameplay wise I don't think that friendly fire collision for bombers have that big of an effect if there are already other incentives to spread out your planes.