Aztez – A Game of Conquest and Brutality
24May/1215

The Amalur Problem

While Kingdom Of Amalur's combat wasn't compelling enough in either direction (good or bad) to dedicate an entire combat analysis post to it, I wanted to address one major mistake the game made that prevented it from feeling a whole lot better than it did. It's frustrating to me personally because it's so easy to correct! The issue I'm about to elaborate on is a perfect example of how one simple nuance can make or break the feel of entire systems.

When you swing a weapon in Amalur and you do NOT cancel it with the next attack in the combo sequence (or it's simply the last attack in the sequence), the attack animation finishes. Now in many situations you want animations to finish playing because interrupted animations can look really bad. But the game's attack animations have a significant amount of follow-through at the end of them. While this is good application of animation principles (most character animations, especially attack animations, should have this follow-through to add weight to the movement), the problem here is that the player's movement input does not cancel this follow-through and it feels super sluggish as a result.

I've illustrated what I'm referring to in the above image. The forehand slash in Aztez is a pretty straightforward attack which is why I used it in this example. The entire animation from start to finish is 60 frames long at 60 frames per second which means it takes exactly one second to play all the way through. There is about 6 or 7 frames of anticipation (animation principle; anticipation is the movement that precedes the movement), about 10 frames where the weapon is being swung through the air in front of the character (there is an active hit box inside this window that lasts about 5 frames), and then you can see that the rest of the animation is spent on follow through and recovery. The majority of it! And Amalur makes you sit through the entire duration of this. What is the solution to this problem? A simple input flag that tells the game to cancel the animation completely if the player is providing movement input at a certain point or beyond in the timeline of the animation.

What you see here in this illustration is a flag at 30 frames that allows this to occur. If the player is providing the input at that exact moment of the attack animation (or past it) the game cancels it into the character's movement state so the player may continue to move. "But Ben! This finished, weighty animation looks so good and the player will miss half of it if they are trying to move!" I know and I don't care because you don't want the player sitting through a single frame of input blackout that they don't absolutely have to. Trust me on this. All the greats do it; pick up Devil May Cry, God Of War, or Bayonetta and perform an attack with no input. Full and beautiful animation with satisfying follow through and recovery. And long. Now do the exact same attack with your thumb on the movement stick. The attack stops entirely at x frames so you can continue to run and it feels a billion times more responsive. What I ended up doing in Amalur was cancelling every individual/finishing attack with a dodge just to have that feeling of input fidelity. It was super annoying.

You're going to have that classic feeling of disappointment knowing your full animations won't even be seen most of the time but when your player is sitting there with the their thumb on the movement input it's because they want to continue moving as fast as possible and it's your job to make these concessions so it feels good to kick ass in your game. Now I'm not necessarily saying these concessions should always be made no matter what, but for as something as fundamental as this, there's little to no risk of it breaking larger systems or making them feel bad.

  • Demon’s/Dark Souls makes you sit through the follow-through in a way that feels very intentional and, as a result, combat, especially with heavier weapons, becomes very strategic, with a lot of planning required even for basic attacks.

    And it feels like it fits, without feeling tedious. In general I can agree with this post, but I’m really interested why in some cases it feels like good design and in others like poor design. Any thoughts?

    • Good point! In Demon’s Souls it absolutely IS intentional. The core tenet of games like Demon’s Souls (whose design philosophy actually stems primarily from Monster Hunter) is for the player to be successful by correctly making attacks that are highly dangerous commitments.

      But I should have more clearly stated that this isn’t a mistake by default! It’s just a mistake for beat ’em ups because of the game play expectations; players understand that individual attacks are not particularly meaningful so in this case, the concession should be made. Since Amalur was most certainly built to resemble a beat ’em up, this issue is a mistake in its context.

      I do believe you could easily take a combat game of this nature in either direction (beat ’em up or Monster Hunter) but the expectation must be set and the gameplay consistent with that expectation.

      • Hmm, facinating that you think of it as a genre thing. So to be clear, you’re saying that un-cancel-able attack cooldowns are the difference between beat-em-up and (I’m guessing?) hack ‘n slash?

        I think it’s great to look at the genres this way. It’s nuanced as hell – I usually thought of the difference as being about whether you use a weapon in the game. There may be some counter examples out there though!

        I wonder whether you might end up giving hack-‘n-slash developers too much creit though. I’m sure in monster hunter and souls, the cooldown animation is a thought-out feature. But a lot of the time I suspect animators just want to masturbate in front of you, the problem you’re discussing in this (excellent) article.

        • It’s hard to clarify for you since I don’t necessarily know what you mean when you say “hack ‘n slash”, but all I really mean is the Amalur problem is a problem in beat ’em ups, where there are distinct input expectations.

          Ugh. It gets tricky here because no one is really on the same page about these genre definitions. Most people refer to beat ’em ups where the character is holding a weapon(s) as hack ‘n slash. I think that’s silly because that’s just a trivial aesthetic difference with no need for a split in genre definition. To me, Otogi (swords/magic) is just as much a beat ’em up as Buffy (fist/feet).

          Personally, I boil down game genres (especially in this weird hardcore arcade born-niche) into this; “What is the player’s intent on a second-to-second level?”

          In games like Strider/Kung Fu/Actraiser, the intent is very careful positioning of your character while striking many enemies precisely with individual melee attacks while X additional enemies zip about arbitrarily. I call these games “kill ’em ups” because your character dies when hit once or twice and most enemies do the same. The distinct absence of repeat aggression against any one target (with the exception of bosses) is what sets them apart from beat ’em ups.

          In games like Devil May Cry/God Of War/God Hand, the intent is simply to get your character within striking distance and perform repeated strikes of varying meaning against against 1 to X enemies, while monitoring the movements and attacks of 1 to X additional enemies all focused on you.

          And etc. I might actually make a post about this because genre splits are fascinating.

          So yeah I really like to look at genres this way too, obviously, because it’s totally all about the nuance.

          And I completely agree about animation masturbation. I fucking hate the idea of all these character animations I’ve made (133 and counting) not being seen and explored but luckily for Aztez, the designer part of my brain is dominant.

          • Hmmm hmmm hmmm! I agree with you so much about the weapon distinction being trivial. Here’s a tragic thing though: a lot of these games are designed around a story, and from the point of view of storytellers, weapons/magic/nonsense makes a lot of difference.

            I once played a shoot-em-up with a fantasy aesthetic. The characters, of course, spent a while “establishing” the “narrative context”, involving cultures and barbarians etc. It’s just so stupid to see when you know that at heart it’s the same as (say) a Star Wars licensed shmup. These games are about a medium-sized 2D object firing small-sized 2D objects at medium-and-sometimes-large 2D objects.

            So it would, of course, be fucking hard for you to draw rigourous genre distinctions within beat ’em ups. It’s hard enough distinguishing beat ’em ups from RPGs; a lot of places view the Souls games as RPGs. The only easy thing that comes to mind is saying “there are 2D beat ’em ups and 3D beat ’em ups” – but even that’s somewhat fucked by things like Alien vs Predator and the fact that Aztec, a 2D game, is clearly influenced a lot by Bayonetta/DMC/etc!

            Here’s what I’d like to see you do! Make a ginormous self-indulgent graph. On one axis put “percentage of moves that can be cancelled”, and on another axis put “average number of non-boss enemies killed by one move”. Plot some games on there – they’ll be scattered, but you might find some clumps of games that you want to call “kill ’em up” or “hack ‘n slash” or “action RPG” :P Maybe there are some more axes you’d like?

  • Shay Pierce

    Good analysis. The WoW team web through something similar. If your toon is sitting down, and the player hits the spacebar… You have to show an animation to transition between sitting and being in midair, right? They made the animations, played with the transitions, tried to figure out the right number of frames of animation to show before after jumping.

    Turns out the right number is: zero. Hit space, and you’re jumping instantly, no matter what. No matter how weird something may look, responsiveness of controls is 20x as important and fundamental.

    • Haha! It still makes me laugh to hear people say “toon”.

      So yeah! I totally know what you mean. A lot of animations most certainly need to be chopped off at the head in real time if the player wishes to do something and you don’t want to make them feel like their character exists in a viscous volume.

      Bayonetta is actually unbelievably rife with these animation decimations, and not just in and around attacks. The animators went fucking APESHIT on secondary animation, follow through, extravagant recovery, and even effects; and a huge variety of it is goes completely unseen if you’re providing ANY input. I think God Of War established a really positive tenet in that everything should be cancelable. Cancellable? Either one, really. ;)

      • This makes me think how one of the labors of love for game designers is willingly putting in the effort to make content they know a lot of players will never see, to make the stuff players DO see feel richer. One of the biggest examples I can think of is Mass Effect 2 and 3. There are entire branches of the story people will never see: Renegade/Paragon, romantic relationships with characters you didn’t seduce/allow to live, alternate story choices. Each choice you DO make then feels like it was tailor-made just for you, and you know you could go play it again and see even more stuff.

        It made me think of an analogy of hiding flowers in every room of your girlfriend’s house… no, mansion. You know she won’t find all of them before they wilt, but every time she finds another one she wasn’t expecting she will feel loved again at just how effort you went to show her your love. And in her imagination you hid even more flowers than you really did, she’ll just assume she didn’t them all. (In this analogy, un-found flowers just magically vaporize when they wilt instead of turning into rotten, disgusting messes behind the bookshelf or radiator.)

        Works at the millisecond level just like the hour level.

  • Interesting and hints at a ton of stuff about the difference between visual fluff and high end gameplay. Like Chris was saying the only ponderous swings I want to be unstoppable are super heavy moves that trade a slow-ass windup and follow through for unblockability or some other good tradeoff.

    It made me think about what world-class players would do to minmax if the cancel/combo option was at 20 or 15 frames.

  • Matt

    It’s funny you brought this up because I just noticed how awkward this is. I’m playing through the Devil May Cry collection and am currently on 2, the terrible one, and believe it or not it has this problem. You can stop attacking and be able to dodge fairly quickly, but you can’t cancel. These fractions of a second feel like forever when you’re trying to do something and the game isn’t responding. It’s even worse when you factor in the length of enemy attacks.

    Let’s say the enemy takes one second from telegraphing to hitting you, for example. Subtract about half for reaction time and another half until you get back in control. Most of the bosses attack just a tiny bit quicker than you making it almost impossible to react in time. It makes these situations where you stop doing anything because you know the enemy should attack again soon and you don’t want to be busy doing anything, the exact opposite feeling you should have in this genre.

  • Jorge

    Generally, removing this kind of recovery from a move should be handled with care, as some moves should indeed have some recovery in order to establish risk and reward amongst the palette of available moves. I kind of never liked the “everything is cancel-able” approach that GOW has, simply because I will cancel EVERYTHING if I can, only because I have no issue adding a command on top of a move if it makes it safer. That being said, a move that starts combos, is used for probing, establishing distance/zoning or is used often, should have the most versatility. In that case, this approach is absolutely necessary. No attack should be lead with an over-committed move as it lends itself to possible counterattack. That being said, big damage, heavy recovery moves with fast start up are perfect for counter-attacking. It makes for fun defensive initiative interplay.

  • kOOk

    Another point about sitting through the entire animation: no matter how beautiful it is, players aren’t going to want to sit through it thousands of time.

  • Owen

    Pulling my hair out regarding this same issue in Darksiders 2. In it, Death (main character) has a lunge/dodge maneuver which will quickly move the character in a direction and give some frames of invincibility. Unfortunately it might be the worst dodge mechanic I’ve ever seen. First issue is it seems to randomly not cancel some attack animations, nor are they cancelable by movement, but if done 3 times in a row (to get a speed boost while moving around or just compulsively) begins a maddeningly long un-cancelable recovery animation.

    Completely ruins the feel of the game!

  • Pingback: On Genre Distinctions « Aztez Development Blog()

  • Pingback: Combat Analysis: DMC « Aztez – A Game of Conquest and Brutality()

  • Pingback: Frame By Frame: The Sword’s Basics « Aztez – A Game of Conquest and Brutality()

  • I actually told the Amalur crew about this during development, and the horrible mismatch between the input gap the player had to leave in order to react to enemy attacks and the gap the enemies actually present. I was told by the producer that my views are “bullshit”.

    The problem with Amalur’s combat is the same problem that plagues so many western developed action games: the designers are far too in love with their own animations. As a player, I couldn’t give a damn about how long it took the designers to animate the character and how much they want to show that work off. If I correctly read an enemy attack and respond at exactly the right moment, then I win that particular exchange. In Amalur, the player has to pre-emptively make sure that the player character was not committed to animation while in anticipation of an enemy attack. It’s a dynamic which acts like an invisible wall which stops the player from delving deep into the combat mechanics and how they relate to the enemy attack signposts. You are constantly discouraged from taking risk.

    The rigid system of Amalur presented many problems, which the designers fixed with a series of band-aids. For example, the evade has no invincible frames and you could only cancel normal attacks with it. But they wanted players to be able to “just evade” like they see in the Japanese action games, so their solution was to shift the player character hitbox ahead of the character during an evade, which in certain situation would make it look like you had performed a “just evade”. My point to them was that the player has no knowledge of this. There is no way any player can possibly know where the character hitbox is positioned, so how then can a player learn and make use of this mechanic? It makes FAR more sense to trigger a small number of invincible frames on the button press; this allows the player to gain an exact sense of where the invincible window is (the player’s own input), and an exact sense of the precise moment an evade is required can then be developed.

    Another example was the Parry mechanic, you couldn’t frame cancel the Parry which made Parrying multiple overlapping attacks impossible. So their solution was to add an AoE blast effect to the Parry which would stop all other enemy attacks instantly. This removed the unfairness of not being able to Parry multiple attacks, but removed the whole dynamic of the player learning how to read and deal with multiple enemy attacks at once.

    The system in Amalur was so frustrating for me also because the solutions to the game’s problems are SO simple!

  • Tipo

    Hey! I’m finding myself more and more enthralled by action games and I’d love to know them more in depht: can you recommend an entry-level book or some other material that explains the design choices behind a game’s combat system?