The Importance of Inaccuracy

Discussion in 'Planetary Annihilation General Discussion' started by ledarsi, October 19, 2013.

  1. cola_colin

    cola_colin Moderator Alumni

    Messages:
    12,074
    Likes Received:
    16,221
    PA might use parameters that are not very friendly to whatever it is you guys exactly want, but I am quite quite sure that it does simulate the bullets by some more or less realistic formulas.
  2. TheLambaster

    TheLambaster Active Member

    Messages:
    489
    Likes Received:
    131
    No, in the end there is no difference. The difference is only internally. The result is the same. I see ledarsi's point though. But I don't really agree. Imo randomness is generally a bad thing. It is what ruined Company of Heroes and Dawn of War 2 for example. In fact the lack of randomness is the only thing I like about Star Craft and the likes.

    Edit: Just to make this clear: I am not against simulating projectiles. I am against random projectile deviation and other random factors.
  3. krakanu

    krakanu Well-Known Member

    Messages:
    540
    Likes Received:
    526
    I'm not sure what to make of your post. Are you questioning the validity of my explanation or asking for clarification?

    Regardless, I will attempt to clarify. Yes, each projectile type is essentially a set of parameters. You can increase the initial velocity to reduce the time-to-target, increase air resistance to make it drop faster, etc. Once the parameters are set, the next step is running the simulation. However, how the simulation handles these parameters could vary drastically, so its not as simple as you make it sound.

    Your reference to the AI is irrelevant to the discussion. I have no idea how Sorian is programming the AI, but I'd imagine you are grossly oversimplifying it.
  4. nanolathe

    nanolathe Post Master General

    Messages:
    3,839
    Likes Received:
    1,887
    I restate, there is a difference. The result is much more than just an "internal" calculation. You physically see a shell hit a target and it'll do damage. A shell that hits nothing does damage to nothing. Terrain interviens; no damage. Unit moves out of the projectile's arc; no damage.

    This isn't a difficult concept to grasp. There is a fundamental difference between a simulated projectile and a random number generator.
    nateious, TheAnnihilator and ace63 like this.
  5. krakanu

    krakanu Well-Known Member

    Messages:
    540
    Likes Received:
    526
    There is a vast difference between the two. Simulated projectiles can produce "emergent behavior" which is not possible with "roll-to-hit" projectiles. An airplane could accidentally fly into an en route nuclear missile, a fast tank could push a slower one out of the way of an artillery shell, a battleship's volley can overshoot its target and hit nearby structures. There is a myriad of situations that are simply impossible in "roll-to-hit" scenarios without them being explicitly coded in.
  6. krakanu

    krakanu Well-Known Member

    Messages:
    540
    Likes Received:
    526
    Let's take a specific case: An artillery shell is fired in a high-arc trajectory, causing it to take some time to reach its target, a tank far on the other side of the map. An air transport picks up the tank while the shell is still in flight.

    Roll-to-hit projectiles: When the shell is fired, the game rolls the dice, and decides this shell is destined to hit its target. As the shell travels, the air transport picks up the tank and flies off. The game has already determined this shell must hit, so it awkwardly manipulates its initial parabolic trajectory to follow the tank until it hits the target.

    Simulated projectiles: When the shell is fired, it is given a specific velocity, air resistance, and direction, which determine its exact path. You could draw a parabolic arc in the game world the instant the gun fires and the projectile would follow this exact path.

    The air transport moves in to pick up the tank, after this happens, there are a multitude of possibilities:

    1. The transport picks up the tank and moves out of the way thus saving the tank.
    2. The transport is too slow, it is hit by the projectile as it moves to grab the tank and dies.
    3. The transport grabs the tank but is hit as it lifts off, both the tank and transport die.
    4. If the tank is held externally, the projectile could hit the tank without hitting the transport, though this would be difficult if the shot is coming from above.

    I see a very clear difference between the two, and the second option seems much more interesting gameplay-wise.
    nateious and polaris173 like this.
  7. MrTBSC

    MrTBSC Post Master General

    Messages:
    4,857
    Likes Received:
    1,823
    what is inaccurisy then?
    say a projectile that hit the suposed position of the unit but missed the unit itself because it moved away in the moment that projektile got launched, is such a projectile inaccurate and therefore not simulated?

    or in case of firing against structures what makes it so intresting about an artilleryunit missing a static targed?
    the unit has a visual on its targed through a scout or radar .. again a static targed ... the artillery itself deploys aswell and targets the position of the structure say a metalextractor ...
    so why should it miss? no other obstacles like cliffs highground or intercepting units just plain open field ...
    what inaccurisy should be there?
    Last edited: March 24, 2014
  8. nanolathe

    nanolathe Post Master General

    Messages:
    3,839
    Likes Received:
    1,887
    Inaccuracy in this context would be introducing and simulating some variance in the firing solution. You can do this via simulating a fluctuating wind speed that affects a projectile in transit (as TA did for the Bertha/Intimidator) or introducing a slight variance in muzzle velocity and target resolution as projectiles are fired. You "wobble" the shot's trajectory as recoil takes effect on the muzzle of a weapon as it is fired, just like in real life.

    It's as simple as adding or subtracting a fraction of a degree from a firing solution, or adding or subtracting a fraction of a percent of the muzzle velocity of each shot fired, generating very slight differences in where a shell will fall after being fired. This can be done as abstractly (via a distribution) or as realistically (actually calculate wind, rotation of the planet, unit weapon recoil based on caliber of the weapon) as desired.
    ace63, polaris173 and godde like this.
  9. krakanu

    krakanu Well-Known Member

    Messages:
    540
    Likes Received:
    526
    If I gave you a mortar launcher and showed you the position of a building nearby on the map, would you hit it perfectly every time? Probably not, because you have no training using a mortar.... but you'd also miss because of variables you cannot account for. Things such as wind speed, air temperature, minute differences in the amount of gunpowder in the shell, slight variations in the metal of the mortar tube, etc etc. All these things contribute to inaccuracy, and I'd imagine more complex artillery pieces have even more variables to account for.

    Of course, these things are not actually simulated (usually). Instead, the gun is given some amount of variance in the direction of its shells. Sometimes it will fire directly at the target, other times it will fire 5 meters to the left. You could argue that this is "roll-to-hit", but it is different in that the destiny of the shell is not predetermined at launch, it is instead determined upon hitting something. Whether that something is the target, a nearby mountain, or a passing airplane, is unknown until you actually simulate the projectile traveling.
    polaris173 and godde like this.
  10. TheLambaster

    TheLambaster Active Member

    Messages:
    489
    Likes Received:
    131

    Either there is a random factor to projectile path or there isn't. I didn't concern myself with the question of simulated vs. non-simulated. I only considered randomness in regards to the actual first target unit, not possible secondary targets, getting hit when the first target was missed, or even planes hitting nukes. Because this is not the point. Nobody is suggesting a non-simulated system, I guess? At least this is not what the OP was about. The question is about randomness WITHIN the simulated system. Not about simulation vs. non-simulation. And in regards to this very specific question (randomness or not) my assertion stands. There s no difference. Simulated inaccuracy has the same effect has a pure chance system, within the simulated system itself. Do you see now, what I was saying? I am not as stupid as to not 'grasp the fundamental difference of the two systems'.



    Don't get me wrong is this sounds kind of belligerent. I just don't like to be suggested to be stupid...
    Last edited: March 24, 2014
  11. godde

    godde Well-Known Member

    Messages:
    1,425
    Likes Received:
    499
    Homing missiles missed a lot in TA. Not just to mentioned that the missiles would hit wrecks, trees and hills that were in the way but they would also miss planes a lot as the missiles didn't always turn fast enough to hit the planes and the missiles frequently ran out of fuel before they could catch the planes.
  12. tatsujb

    tatsujb Post Master General

    Messages:
    12,902
    Likes Received:
    5,385
    ye but it's not being exploited. what a waste!
  13. MrTBSC

    MrTBSC Post Master General

    Messages:
    4,857
    Likes Received:
    1,823
    which to me it is ... as i said it brings a part of randomness to the game i dont want ... if a shell hits something different that blocked its flight like a plane, mountain or a wall because it simply was in the way that is something i am ok with ... however it missing a structure on a clear flightpath because some arbitrary value has been put in that messes with my tactics instead of the enemy countering it with a wall or something else would piss me off ... and i dont think that fits with the WYSIWYG philosophy uber has ... i cant see that "5 to 15% spreadchance" except for the missed shot on impact when it is too late already ...
    TheLambaster likes this.
  14. lokiCML

    lokiCML Post Master General

    Messages:
    1,973
    Likes Received:
    953
    Out of curiosity have you been reading the same thread that I have?:confused:
  15. MrTBSC

    MrTBSC Post Master General

    Messages:
    4,857
    Likes Received:
    1,823
    who are you talking to and can you be a bit more specific on your issue?
  16. lokiCML

    lokiCML Post Master General

    Messages:
    1,973
    Likes Received:
    953
    I am talking to you.;) Nobody has been advocating randomness. That four letter word.:)
  17. MrTBSC

    MrTBSC Post Master General

    Messages:
    4,857
    Likes Received:
    1,823
    advocating something and causing something are 2 different things ...
    just because you think something is good in theory doesn´t mean it has to be in practice
  18. polaris173

    polaris173 Active Member

    Messages:
    165
    Likes Received:
    204
    It creates more meaningful weapon differentiation. It also provides another variable to tweak when balancing units. Want a super strong artillery piece that's somewhat cheap for roughly clearing out a horde of Dox? If it was super accurate it would probably be OP, but giving it a slightly randomized spread helps make it balanced. Gunships lasering down on everything almost perfectly? Give them a little firing spread to make up for their crazy DPS, and make it an awesome crowd control unit, and not an OP commander death machine.
  19. Dementiurge

    Dementiurge Post Master General

    Messages:
    1,094
    Likes Received:
    693
    I think TBSC's point is that this supposed "variance" is just going to be implemented as random number generation.

    I'm at a loss to think of how one could implement it in a fashion that wouldn't be some semblance of random.
  20. MrTBSC

    MrTBSC Post Master General

    Messages:
    4,857
    Likes Received:
    1,823
    if the commander stands there like a monument then it just deserves to be blown away by artillery even by just a single one ... also for stuff to hit perfectly would requier them to be hitscan or/and superquickhoming ... not the case on any unit afaik ... velocity and arc of projectiles help a lot with not always hitting targets ... and i would be damn suprised if my gunships were not able to gundown a stationary turret quickly

Share This Page