Howdy, Stranger!
It looks like you're new here. If you want to get involved, click one of these buttons!
Quick Links
Collision Detection - Melee Combat
Eronakis
Member UncommonPosts: 2,249
This has been on my mind as of recent.
First let me express that I typically play caster DPS and don't play Melee classes as often. I find Melee to bland and boring because the tactical gameplay isn't there. I am sure melee players will say the same for a caster dps : )
I have always wanted to implement a Melee Combat System that revolved around European Swordsmanship that is roughly translated into tab targetting combat.
The system I have "developed" is quite complex as there are a ton of variables that could ruin such a system, even if the gameplay of this is fun. Typically, for collision detection there is an invisible hit box surrounding NPC/Player.
The backbone of said system is based around Guard play. A Guard, is the ready position in posture in which you hold your weapon to strike or parry. There are roughly 9-15 different Guard types just for 2H swords between German and Italian Swordsmanship. There are typically 3 primary Guard Positions. High Guard, Mid Guard and Low Guard. Each Guard position comes with a set of specific Guards.
For an example: High Guard Position would have Ochs Guard, where the blade is positioned above your head horizontally. And Roof Guard where the hilt is above your right shoulder with the blade pointing about 45 degrees backward.
Each Guard Position would provide a specific type of function. Striking (offensive hit), Countering (parry then a hit for half damage) and a Parry (block). And these functions would be dispersed in the appropriate Guard Position.
High Guards = Striking and Countering abilities.
Mid Guards = Parrying and Countering abilities.
Low Guards = Parrying and Striking abilities.
To translate this into melee combat mechanics while keeping tab targeting in mind the player would have to engage in a very rough simulation of swordplay.
The gameplay for this system for the melee player would determine what NPC would be in what position, then discern how to appropriately react or adapt in swordplay.
A system such of this would excite me to play a Melee Class because it would ad depth and tactical gameplay in a fight. However, in order for a Guard system like this to work with Tab Targetting would be very challenging to implement. You would have to have accurate collision detection for weapons to clash so it can register as a strike, parry or counter. Three off the cusp issues I see with this is, server latency which could ruin true collision detection, lack of fluid animation to somewhat feel like you're in swordplay and if you have different races with different sizes you would have to account for that as well which could wipe away polish from this system. Not to mention the resources to pull off such a feat.
I want to also NOTE that this system I have envisioned isn't twitch gameplay like For Honor but still in someways uses the philosophy of each Gaurd Position from For Honor. I have been conjuring up this mechanic for quite sometime before For Honor has been released.
*QUESTIONS FOR THE THREAD*
Would you consider giving such a system like this a chance?
Do you feel like the problems I have described above would impeded the fluidity of gameplay for melee combat mechanics?
Do you feel like pure collision detection for melee combat is needed and can it be achieved?
And lastly, what do you think is the next step for melee combat evolution?
*Note* I do understand that this mechanic might be confusing, I didn't want to type a novel about it : ) If you have any questions, I can answer.
First let me express that I typically play caster DPS and don't play Melee classes as often. I find Melee to bland and boring because the tactical gameplay isn't there. I am sure melee players will say the same for a caster dps : )
I have always wanted to implement a Melee Combat System that revolved around European Swordsmanship that is roughly translated into tab targetting combat.
The system I have "developed" is quite complex as there are a ton of variables that could ruin such a system, even if the gameplay of this is fun. Typically, for collision detection there is an invisible hit box surrounding NPC/Player.
The backbone of said system is based around Guard play. A Guard, is the ready position in posture in which you hold your weapon to strike or parry. There are roughly 9-15 different Guard types just for 2H swords between German and Italian Swordsmanship. There are typically 3 primary Guard Positions. High Guard, Mid Guard and Low Guard. Each Guard position comes with a set of specific Guards.
For an example: High Guard Position would have Ochs Guard, where the blade is positioned above your head horizontally. And Roof Guard where the hilt is above your right shoulder with the blade pointing about 45 degrees backward.
Each Guard Position would provide a specific type of function. Striking (offensive hit), Countering (parry then a hit for half damage) and a Parry (block). And these functions would be dispersed in the appropriate Guard Position.
High Guards = Striking and Countering abilities.
Mid Guards = Parrying and Countering abilities.
Low Guards = Parrying and Striking abilities.
To translate this into melee combat mechanics while keeping tab targeting in mind the player would have to engage in a very rough simulation of swordplay.
The gameplay for this system for the melee player would determine what NPC would be in what position, then discern how to appropriately react or adapt in swordplay.
A system such of this would excite me to play a Melee Class because it would ad depth and tactical gameplay in a fight. However, in order for a Guard system like this to work with Tab Targetting would be very challenging to implement. You would have to have accurate collision detection for weapons to clash so it can register as a strike, parry or counter. Three off the cusp issues I see with this is, server latency which could ruin true collision detection, lack of fluid animation to somewhat feel like you're in swordplay and if you have different races with different sizes you would have to account for that as well which could wipe away polish from this system. Not to mention the resources to pull off such a feat.
I want to also NOTE that this system I have envisioned isn't twitch gameplay like For Honor but still in someways uses the philosophy of each Gaurd Position from For Honor. I have been conjuring up this mechanic for quite sometime before For Honor has been released.
*QUESTIONS FOR THE THREAD*
Would you consider giving such a system like this a chance?
Do you feel like the problems I have described above would impeded the fluidity of gameplay for melee combat mechanics?
Do you feel like pure collision detection for melee combat is needed and can it be achieved?
And lastly, what do you think is the next step for melee combat evolution?
*Note* I do understand that this mechanic might be confusing, I didn't want to type a novel about it : ) If you have any questions, I can answer.
Comments
Ochs Guard (High)
http://www.thearma.org/essays/newstancepics/ORrs.jpg
Roof Guard (High)
http://www.thearma.org/essays/newstancepics/VDPDrs.jpg
You wouldn't be artificially forced to attack the tank, but ignoring the tank is also not a realistic option, as he only has to reach engagement distance and, if you move away from the tank without CCing or killing him, you take heavy damage via a free retaliation attack from the tank. Talents (or whatever you wanna call them) can be used to further accentuate the need to address the tank, such as slows on retaliatory hits, large bonus damage (for the melee DPS diving squishies), etc.. Tanks could play a very large tactical role with positioning that forces teams to work together.
Balabce would have to take into account that casters are placed into a heavily disadvantaged role as soon as a melee enters engagement range, opening up defensives, CCs, and other counters that would be as important for keeping folks alive as straight heals.
I am interested for your more in-depth reply.
I would like to mention that in order for this system to work, the approach to the trinity combat model would have to change. To immerse the player with swordplay, melee battles would have to be 1v1.
Here is a small write up to explain this...
Section 2: Reinvention of the Trinity Gameplay Model
After many hours of attempting to make Melee Combat more meaningful with more strategy, the Guard System can only work if the approach to the Trinity Game Play Model is changed. To have proper Guard Play there can no longer be a group of players fighting one NPC. Thus, the ‘tank & spank’ model must be omitted to fully capture the Guard System.
The Tank role is the core that molds the traditional trinity gameplay model together. The approach to tanking must change. Instead of being distinct defensive fighters that mitigate damage, Tanks will no longer be considered a specific class role, but rather jump to the archetype level to become a universal role for any melee class. In this new system of the trinity model, tank will fight 1v1 battles against other melee NPC’s to keep ‘aggro’ off other magic casters and support classes. Naturally melee will fight melee classes 1v1 first. There will be moments and situations were a melee player may have to fight a caster 1v1 as well.
The dynamic of groups changes with the reinvention which creates another layer of roles. Attackers and Support. Melee classes are considered Attackers as they will engage in combat, whereas Support classes will help the group and aid melee classes. Magic Casters can also be considered Attackers as they will naturally fight other magic caster NPC’s 1v1.
Instead of fighting one NPC at a time, the group of players will engage in an encounter of a group of NPCs. Think of it like the game of American Football. It’s a team sport but with individual responsibilities. Your Quarter Back distributes the ball (attacks), while your Offensive Line blocks and protects (support) your Quarter Back. There are other different types of attackers that compliment your QB, such as your Running Back and Wide Receivers. Each layer on the team performs individually but when executed correctly its considered a team effort. They act as one cohesive unit.
As this game will have a heavy focus on group play, let me give an on the surface example of how a group will approach another group of NPCs. Player Group = 3 Melee Attackers, 2 Caster Attackers and 1 Healer. NPC Group = 4 Melee Attackers and 2 Healers (support). 3/4 Melee NPC’s are accounted for by the Player Group. The other Melee NPC may be crowd controlled, or focused by the 2 other Caster Attackers. What I am trying to convey is that every Melee class can only fight 1 Melee NPC 1v1 to ensure the system of the Guard play stays in tact.
I'd also see is other important combat elements included -- knockdown, grappling and pushing. Knocking the opponent off their feet gives a significant advantage. Grappling is all forms of holding the enemy from acting. Pushing attempts to move the foe backwards.
I think the problem is that your 'guard' idea is much more like a simulation of fencing, rather than fighting. Fencing is clean. Fighting isn't. I don't envision the fighters fencing with enemies while the specialists do their work.
I would probably disagree with a 1v1 melee fight as the pinnacle of a combat situation. I could only see that as an intermediate step towards something else. If I can't indulge my fantasy of fighting off an army of goblins, what's the purpose? The example also assumes equal numbers on each side. What happens with a 12 on 5? Do the extras on the 12 side wait around to jump in like a bunch of guys waiting to rush Bruce Lee?
I could easily envision a melee only system without casters of any nature. Everyone is a combatant. CC is done by people who knock the other guy to the ground and simply holds him there. Healers drag the injured back to safety and administer first aid. Archery is something done without someone in the archer's face.
Would I play? Probably. Even the fencing model version breaks the 6-on-1 fights of other MMORPGs. So that's a good thing. The fighting model version could also be used to allow larger size 'groups', even up to 72 men 'companies' or even larger 'armies'.
The various guard positions you propose are probably a bit extreme for my tastes. Even 3 positions (high, low, medium) requires 3 attack types to those locations, making 9 total. The model itself quickly becomes too complex with 9 positions (HL, HC, HR, ML, MC, MR, LL, LC, LR). What happens when A attacks HR, but B defends LC? The combinations tend to make both offense and defense more 'action' than not.
My advice: think ancients rather than Italian Renaissance.
Logic, my dear, merely enables one to be wrong with great authority.
We tried to build a similar system with guarding that required you to guard incoming swings on the basis of left, middle, right instead of high, middle, low. What we found is that you couldn't react fast enough, so it was really just random blocking. Even when we slowed down the swing to a painfully slow speed, the reaction time window was still too small. So we scrapped the system. The system you are proposing sounds more like stances than reacting by pressing a key at the right time, so I think the system you are proposing would work better than what we tried to do. I especially like the idea of having access to limited abilities based on each stance/guard position.
I have already achieved the pure melee detection that you would need. It uses PhysX collision on the GPU instead of hitboxes and provides near pixel accurate collision. So if you see it collide on the screen it is a hit, if it doesn't collide on screen it isn't (it is still server authoritative though). I thought latency would be a big issue, but in play testing it doesn't really seem to be. We are using client prediction to simulate every character on each player's PC, so that does help.
I think the next step is to get rid of tab targeting and combine real-time combat with deep strategic tactics. I also think it is time to get rid of the Trinity. One of the ways we do this is by having players collide with each other instead of being able to walk through them. The other is by getting rid of the concept of threat and threat tables. This completely changes combat. There is more reason to fight multiple mobs like you mentioned, when only 3 or 4 melee in your group can even attack one mob (due to player collision). With this you now need proper fighting formations for your group of players. You might need a formation that protects the healer or other squishy players. Also, getting rid of the threat system helps with this too. If the AI is not forced to attack certain players and is free to attack the weaker players, then you have to adapt to this to survive.
You can have collision detection so that melee players can physically block the opponent from running to caster players, but for hit detection it's irrelevant.
So it's like having WoW warriors having Defensive Stance, Battle Stance and Berserker Stance, except your names are less intuitive.
Guards have varying functions. These are the proper terms used. It's nothing like a WoW Warrior. The Guard system actually represents a physical posture of how you hold your weapon. There are no such function or visual postures for different Guards in WoW. So you're comment is irrelevant.
You're right, for the initial structure for the Guard system I am proposing is a cleaner version, like fencing in a way. I did have intentions of grappling, knockdown, knockback and pushing. But I felt like the Guard system as is, was already too complex for that and those features could be expanded upon later. They would integrated into the system or spread across certain classes.
The idea of reinventing the tank role and approaching the trinity gameplay model differently actually sprang from this Guard system. The vision of the Guard system could not work efficiently and effectively if melee classes didn't have 1v1 battles. I personally find the strategy of tackling NPC Encounters like this to be fun and allow some deliberation between Encounters for the player group. You know, to actually help bring back a community where players actually converse with each other in a group.
Unfortunately, I am changing that fantasy to have a more tactical one where a player would have to think on how they would approach their NPC Encounter and individual opponent.
I have simplified the Guard system to 3 Guard Positions, High, Mid and Low. You're right, it would be overly complex to also account for the lines of strikes and cuts for the Guards. That's why I wanted to simplify it.
Example: NPC is in a High Guard Position. More than likely that NPC is looking to strike or counter your attack. So the player would have to discern what Guard position should be appropriate. If it were me, depending on class ect, I would go Low Guard to "Offset" the High Guard.
So, it's less about lines of strikes or cut for the Guard play but rather through the Guard Type. The purpose for the Guard position is to give the player visual clues of how to react to their opponent and to determine functionality. (Striking, Countering, Parrying).
Thanks for your response.
What intrigues me about Guards is that it gives the player visual clues and different imagery when in combat. It would help define distinction to classes. You're kind of right that each Guard position "acts" like a stance in some ways. The Guard Positions won't determine direction lines of swings or cuts. That would be too challenging to implement and quite possibly overwhelm the player.
I am a big fan of structured design that is somewhat simple but yet gives players options within that structure. Have the system to be easily learned as you go and once you've hit maxed level you should be proficient at the system.
Even though the Guard Positions "act" like a stance, players would still "react" to a strike, counter or parry. There was a reaction bar in place to show the player the time they had to react to a Strike, Parry or Counter. While it's still tab targeting, you're "locked" on that target by still reacting via button press. You're not swinging or aiming like Twitch Combat.
The very elements of why you scrapped the system are also the same reason why I am also considering a revamp on mine.
In regards to the collision detection, I have heard that it was potentially possible to get it accurate by pixels like you've described but I haven't seen it implemented yet. I could see players being first person to get a close up on how to see visuals when reacting to what Guard position. To help players who want to stick to 3rd person, if the weapons seem too hard to visually capture a hit, parry or a counter, a simple UI alert above the character expressing damage, "parry" or "countered".
I am still a major fan of Tab Targeting combat because it still ensures the traditional MMORPG Gameplay. This Guard system was an enhanced version for Tab Targeting. I personally don't want to get rid of the Trinity or Quadturnity (includes CC) model. I feel like that method of gameplay still focuses on Class Roles which to me is the crux of what an RPG is about. I just feel like the approach to the trinity gameplay model can be changed, like I have proposed.
The strategy I was going for is what type of Melee fighter your class is and what is best to fight that other Melee NPC. During deliberation periods, players would determine how to best tackle that NPC encounter to survive.
I still have an Aggro & Taunt mechanic in place to ensure Melee and Caster Attacking classes can target 1 NPC.
The system was meant to change how players approach an NPC group (encounter). At the start, Attacker Classes would 1v1 their respective target. Once their target is dead, they would then focus fire. -- You can't really get rid of the focus fire element unless all NPC's die at the same time. The frequency of that happening is very nil.
Thanks for your post!
Also, how would that system work with an enemy that has no sword?
If the player does not aim his sword, if he just pushes attack button, after that the player doesn't really know whether you've got complex AI on both sides trying to choose sword positions and a hitbox detection, or a RNG choosing the result of attack and then an appropriate animation to play.
Let me answer you both here.
To Vrika, yes it would be like a "dance" persay. I had to revamp this system a handful of times to make it as simplistic as I could.
The sequence wouldn't be "Chose Guard > Ability". The Guard would be placed on your Combat Bar giving access to specific abilities. When you button press that ability, it will perform a strike, counter or a parry or even a combo of 2 of the 3. Think of each Guard like a Skill Set you have access too.
Example:
Wrath Guard (High Striking Guard)
Ability 1 = Wraith Strike: Deals massive damage
Ability 2 = Guard Break: Breaks an opponents Parry to cause damage.
Front Guard (Mid Parrying Guard)
Ability 1 = Parry: mitigate strike damage or countering damage.
Ability 2 = Pommel: a quick parry to mitigate 25% damage and then a 2 second stun. (uses the pommel on the hilt to bash the opponent)
The idea is that the player would have enough Guards to determine what Guards they want active on a limited Combat Bar. (Combat Bar has max 15 abilities for both melee and caster).
So the player can determine what Guards may be "best" or "most efficient" for fighting a particular NPC.
---
To Answer 4507....
*Note* that each class has a specific weapon handle. I'll give you 3 examples.
Knight = Countering Fighter, traditionally 2H Weapon Handle.
Warrior = Balanced Fighter, 1H + 1H and 1H + Shield Weapon Handle.
Assassin = Deceptive Offensive Fighter, 1H + 1H Weapon Handle.
All 3 classes have different Weapon Handles, and those are the only 3 Weapon Handles that would be in game.
If a Knight (Player) were to encounter a Warrior (NPC), they both have different weapon handles. So one would automatically consider different Guard Positions, because the posture of weapon would be different from a 2H weapon. To negate that, if a Class has either 1H + 1H or 1H + Shield, they would still follow the same format of the High, Mid and Low Guard positions. What would determine that is where the Primary Weapon is being held.
In this case, The Warrior's 3 Positions would be, High (primary) Mid (offhand), Double Mid (Primary and Offhand) and Low (Primary) and Mid (offhand). Even though a Duel Wielder would have access to two weapons, being sword, blunt, axe or dagger, they would still use their Primary weapon to determine a main source of, a Strike, Counter or Parry.
The High, Mid and Low Positions are still essential to all 3 Weapon Handles.
Animal NPC's would have a low AI rank and they would be attacking from certain Guard Positions. How would the player know that? That information would be best described in the size of the animal. Smaller animals attack in Low Guard Position, Medium animals attack in both Low and Mid Guard Positions and Large animals would attack in High, Mid and Low. There would be some UI notification or template that would describe that to the player when you target said animal NPC.
*See* my post where I answered you and 4507.
Guard Proficiency: Determines your success rate with that guard against all other Guards. Example: If Parrying Proficiency is 75% then it will have a 75% chance to cause a Parry which mitigates 100% damage.
You would still need Weapon to Weapon Collision for a Parry and a Counter Strike.
A parry is where you block with your weapon. A Counter Strike is where you parry then immediately strike. The Collision would detect what was registered. So, yes would still need collision detection for weapon to weapon.
Check out this video for what I am talking about. Each person is in several different Guards striking while the other parries. This is a good example of why there needs to be Collision Detection.
The music is terrible so don't mind that lol
maybe that works ina single player game who most of the combats are duels, but i'm pretty sure it would be boring pretty fast
Though most will talk about how they wanna be able to take on numerous opponents at once, that really never happens. Even in action movies, rarely does the hero dispatch more than one or two opponents simultaneously in hand-to-hand combat; the extra henchman very much wait in line to be pummeled one or two at a time. How does that work in the context of the system as you describe it above? I honestly would enjoy a system where multiple mobs would attack your player rapid fire, and your blows would knock them to the ground/backwards/into environmental hazards/whatever and the next would take his/her place as you see in show and movies. Those mobs could get back up and then jump back in, but the system would never place more than two opponents attacking simultaneously to avoid overwhelming the player with stimuli he couldn't hope to respond to simultaneously. If it were rapid fire and fluid enough, it would feel more like action movie combat than any MMORPG (or really, RPG) system out there. Think Rocksteady's Batman franchise, but without the ability to dodge roll/jump all over the place, the numerous extra gadgets, or the one-button counters. Even when Batman is fighting a dozen plus foes, rarely do more than 2-3 attack simultaneously. The rest attempt to surround Batman, but you never have all of them simultaneously bullrush him, because the suspension of disbelief of having Batman counter 10+ people simultaneously is too much, even for a video game.
Is that something you think needs to be a part of this system? A rotating attack chain that utilizes a sort of round-robin system for mobs? I could see depth being added through different mob types/weapons, where the player could also have skills or abilities that help them adapt to the mob currently attacking. Again, rapid fire and fluid, and that would be an intense combat system.
You could simply set percentages for each result of the action. For example a overhand strike against high guard could have 95% chance to get parried and 5% chance to hit, then play the animation for either a parry or a hit depending on result.
You are allowed to cheat when making computer games. You can use the simplest possible way to simulate actions, in this case percents that tell how likely each outcome is. The player will still see all the parries, hits and counterstrikes because you can choose the animation after you've calculated the result.
I like the idea of this and placing a tank between an attacker and a glass canon instead of the old "taunt" mechanic would be interesting. Shields should also have better collision detection. Then you have shields from the huge tower shields to wrist bucklers
- Al
Personally the only modern MMORPG trend that annoys me is the idea that MMOs need to be designed in a way to attract people who don't actually like MMOs. Which to me makes about as much sense as someone trying to figure out a way to get vegetarians to eat at their steakhouse.- FARGIN_WAR
The way I have envisioned the new approach to the trinity gameplay model is to focus more on group content without diminishing solo play.
There will be areas of zones where there is obvious needed for group play, and obvious for solo play.
Solo areas would have NPC's that are more spread out and sometimes weaker. Typically they may be patrol or roamers. Player awareness is something that needs to come back as an expected skill to have like in EQ. The design should have that in mind. When there are cases when more than 1 NPC aggro's you, depending on various variables, the Player can or cannot survive. A skilled player would be able to survive most 2v1 situations if that happens. It's encouraged to "pull" 1 NPC if you're soloing.
Group areas, would have NPC group packs. For the new approach to the trinity gameplay model to work as intended, AOE abilities will have to be extinguished or very limited because that lessens the merit of the system in place.
It's basically a different perspective of what we are use to in the current system of combat.
Classes would have different ways to help null an add while soloing if needed. Such as stuns, CC, snares, roots ect. Those would have to be used strategically.