Game Development: Key Knarr Combat

[VDZ]

• there should be no dodge mechanic that lets you bypass an attack too op and very ping based.

6. Quick Dodge. Broken, can be spammed with no window of opportunity for the attacker to hit. I don't see how this ability can be fixed without introducing individual ability cooldowns.

In fighting games, the concepts of 'startup lag' and 'endlag' balance the otherwise OPness of dodges/parries (and attacks). Whenever you perform an action, there's a certain part of the animation (startup) where you've committed to the action (you can't cancel it anymore) but it does not produce its result yet (attacks generate no hitbox and dodges/parries do not protect against attacks). This means you have to respond earlier to your opponent's actions and particularly laggy actions can be countered by quicker actions. After that, there's a timespan during which the action is active (attacks hit the opponent if their hurtbox touches the hitbox, you become invulnerable to the attacks and/or cause your opponent to enter hitlag if they hit you). The effect ends before the animation is complete, however, and the character is then stuck in endlag for a while - while their action no longer has its effect, they are still unable to act for a while until the whole thing is finished. This means that a missed attack or a dodge/parry while the opponent was not attacking causes the character to have a moment of vulnerability where the opponent has a sufficient amount of time to perform an attack while their enemy is helpless (called a 'punish'). Notably, there are cases where player A performs an attack, player B tries to performing a parrying action (or counterattack), but player A misses, and in that case player A's endlag may be short enough to punish player B while player B is still in endlag. (If the attack had connected, player A's endlag would have been extended ('hitlag') and player A would not have recovered in time to punish player B.)

Back to Haven, adding endlag to dodge mechanics would make them risky to use, particularly if an intentionally missed attack could punish a dodge. I think fighting games in general can serve as an inspiration for PvP combat mechanics.

'Move staling', another concept from fighting games, could also help reduce the effectiveness of dodges and reduce spamminess in general. The concept is simple: Every time you use a move, future uses of that move become less effective. Hitboxes (attacks)/invulnerability (dodges/parries) become shorter in duration, damage is reduced for attacks, and startup lag and endlag become longer (particularly for dodges). Depending on the system, the power of the staled moves will either gradually recover over time or recover when you use different attacks.

Weapons should change the telegraph of some attack to be wider, and/or longer.

I think in general equipment affecting variables like startup time, endlag and range could be interesting, particularly in making weapon type a choice based on player preference.

Skill canceling, if the animation is 1.0 sec long. and the user uses a dodge skill from anywhere at 0.0 to 0.5; the curren skill cancels and you roll.

The flipside to this is, of course, that having too much freedom in your actions can make predicting the opponent's behavior too difficult to have meaningful interaction instead of both sides just spamming their own strategy regardless of what their opponent does.

with fluent i mean no stopping of movement. imagine being followed 2v6 and every attack you do you stops you for 1/3 seconds thats a quick death.

Well again, if the attack is instantaneous, then given auto-aiming, there's literally no point to the attacks being aim-based, no? The wind-up delays are what makes it meaningful.

having an actual arrow or line showing where the attack is headed in the wind up would be useful to help with that.

Not sure why that would be useful during the wind-up, when you've already chosen heading at that point. Or are you thinking that you could still change direction during the wind-up? Because in that case, it seems you could keep auto-aiming during the wind-up, which again would make aim useless.

youre facing real players with their own style of movement they could zigzag change direction stop moving the possibilities are endless.

I guess I don't really see how it matters if they zig-zag or anything else as long as you can just keep auto-aimimg directly towards them.

Just to clarify, when we were testing attacks that didn't stop movement, the restriction they did have was that they prevented any change to the set movement during the wind-up.

What if initiating attack locked your rotation until the attack finishes? Then you'd have to be in the right position relative to your target when the hit frame happens, so simple following wouldn't quite work. Makes some room for skill and juking. Especially if there's a slight speed penalty when not moving in general direction of your initiated attack. Then we could have attacks with windup that don't take away your movement, which is a frustrating mechanic and makes combat feel unresponsive.

Precisely what Potjeh says. You all keep talking as if locking rotation and locking movement are intrinsically related, but I don't see why you couldn't just lock rotation while still allowing free movement, making you have to position yourself correctly to hit under the constraint of the chosen direction (e.g. if you're west of your opponent and initiated your attack to the north you have to quickly move southeast to hit your opponent; likewise if your opponent moves after you started your attack you have to adjust appropriately). And especially in that situation attack direction visualization would be useful to get a good feel of how to position yourself to hit the opponent.

I mean at the end of the day its aimbot which is rampant in almost every fps game with out some crazy restrictions and punishments it will happen regardless of work arounds.

Aimbots are rampant in most FPS games because of the speed ratio between attack execution time and player movement/reaction time; firing bullets is near-instant and you can't dodge them, so to kill the enemy you just have to hover your mouse cursor over them and click. World of Tanks (and this applies to World of Warships as well) provides an interesting contrast: battles take place over long distances (compared to FPS games) and use heavy ammunition, so there's some delay between firing your shell and it actually reaching the target you aimed at. As tanks are frequently in movement (not just forward/backward, but also turning, going up/down due to terrain elevation etc) you need to compensate for various factors and predict where your target will be to aim accurately. The game actually has a built-in auto-aim feature that fires straight at the enemy, but manually aiming is almost always preferable because of the factors you need to take into account. Even an intelligent auto-aim that would take into account current enemy movement would miss when the enemy slows down or changes direction, while an intelligent player would understand that of course the enemy is not going to bump straight into that obstacle or move into enemy fire. As long as there is sufficient movement in combat and attack directions must be committed to in advance, aimbots won't be very useful.

Please fix this, soon. It basically destroys the potential of the game to attract new players.

Is that really true, though? I get that high-pixelcount displays are fairly common, but are high-DPI displays really all that common? On all the 1440p/2160p monitors I've tested, the game has still been perfectly playable for me. I'm well aware that there exist true high-DPI displays on laptops, but are they really "common" by any metric?

While I still wouldn't call them 'common', their adoption is increasing extremely rapidly. At my workplace we configured a system to use 200% display scaling quite a while ago to catch display scaling bugs due to a rapidly increasing number of reports from users on high resolutions, and recently we bought a complete 4K monitor to also catch issues caused by high resolutions unrelated to display scaling because of their rapid adoption. When they noticed the relatively low price for the display (it cost 250 euros) some of my coworkers started considering buying a 4K monitor for themselves as well.

Loftar, would you be so kind to move interface scaling up on your todo list?

I'm honestly not quite sure how to properly fix it, with less than rewriting the whole UI as vector graphics. I mean, sure, I guess I could add some option to just run the rendered output through a bilinear scaler or something, and I guess you could argue it would be better than nothing, but I'd also argue it'd probably look like shit. Feel free to discuss. I mean, I really don't want to dismiss the issue, it's just that making the UI scalable in an acceptable way seems like a pretty daunting task.

Looking like shit is still better than being unplayable. And besides, on a high-DPI display you don't really notice it looks like shit like you would on a 'normal' display; cheaply upscaled stuff only looks a bit fuzzy.

The widescreen trend is why I don't use high-DPI monitors for my desktop computers. 4:3 is love, 4:3 is life, even if that means I won't get any high-DPI or high-refresh-rate monitors. The only reason I accept 16:9 for my laptop is only because there's literally no alternative.

While I still agree 4:3 is better (especially interface-wise; being able to actually see the stuff on the edges and in the corners allows for much better interfaces), at some point you need to let go; the battle has been lost. Everything is designed for 16:9 nowadays and increasingly breaks on 4:3, and if you care about how 99% of end-users experience the game you're making you'll need to play it in 16:9.

Implementing combat stamina is also the way to go, so everyone who is spamming moves would suffer from low stamina at the crucial moment.

Stamina as a required resource used in combat would be an interesting option to reduce spamminess. Care must be taken that it does not overly restrict movement, though; two guys standing still punching eachother will likely not be very interesting.

[Potjeh]

In case someone is interested in how to automate a game: https://deepmind.com/blog/article/alpha ... arcraft-ii could be an interesting read.

Not that it's not cool, but it's still much less impressive than the dota 2 openai. In RTS games where you control up to hundreds of units, computers can much more easily thwart humans purely by micromanaging much better. It's physically impossible for humans to control 100 units individually at the same time, whereas it's pretty simple for a computer. The openai team specifically set limits to their bots with reaction times etc, to make sure that it was more a question of beating top players strategically than purely by mechanical skill.
I get that they have to brand themselves but this is just pure bullshit:
"Although there have been significant successes in video games such as Atari, Mario, Quake III Arena Capture the Flag, and Dota 2, until now, AI techniques have struggled to cope with the complexity of StarCraft."

Actually Alphastar also has constraints on it to eliminate unfair advantages over humans - it has a limit on max APM, and it can't simultanously control units on different virtual screens. It's APM is well below that of grandmaster league players, and it still beats them frequently. And even then it doesn't beat them through superior micro (it's micro would probably be like low masters league), but instead by making superior strategic choices (army composition, sustainable economic growth etc.). And no, it doesn't have maphacks either, it has to scout the opponent to know what counters to build. So yeah, I'd say it's more impressive than the Dota AI.

[jordancoles]

edit: nvm

[blank]

An idea, you could make zigzag make the user become invisible for like 1 second and as a result anyone chasing him in combat stops and makes the opponent have to click. Reduces opening by a good bit and has a long cooldown. Basically what this does is gives the user a better chance in running away from a 1 v 6
It's main use is to cause a disturbance.

[jorb]

An idea, you could make zigzag make the user become invisible for like 1 second and as a result anyone chasing him in combat stops and makes the opponent have to click. Reduces opening by a good bit and has a long cooldown. Basically what this does is gives the user a better chance in running away from a 1 v 6
It's main use is to cause a disturbance.

Non-trivial, but I do not hate the suggested effect.