nodthenbow

Way too many words on hitboxes

Jun 23rd, 2020
134
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 13.62 KB | None | 0 0
  1. Circles compared to boxes:
  2. In the data, a circle is usually represented by a point, a radius/diameter, and some transformation like stretch to another point (optional). A box is represented by a point, a length, and a width.
  3. From a memory usage/disk space standpoint, both circles and boxes are going to be negligible compared to sprites and other stuff, but technically circles use up less space and are slightly harder on the CPU/GPU (notable for arcade era CPUs/GPUs). This is why you don't often see anything like a trapezoid or rotations in standard boxes too, that it notably harder on the CPU/GPU to do (and also annoying to program sometimes).
  4. The difference between circles and boxes in a modern game has mostly to do with what you are trying to represent. As an example, look at how a diagonal line was made in MVC with boxes ( https://dammit.typepad.com/.a/6a0120a8bc3caf970b0133f4ed0ffd970b-800wi ), then compare it to how diagonal line in MVC3 was done ( https://media.eventhubs.com/images/2011/12/30_drdoomhitbox.jpg ) with circles. This is the main reason to use circles or boxes, one is often better at representing the animations. There also isn't much of a reason to mix the styles because that is a lot of extra work when the impact is actually pretty minimal.
  5. In my personal opinion it is a bit harder/annoying to fine tune circles compared to boxes, but that doesn't translate to a change in how the game plays or feels. That's probably the reason boxes see a lot more use.
  6. High accuracy to animations compared to not that:
  7. High accuracy surprisingly feels less accurate while playing the game. One of the main reasons for this is because for the most part people don't think of a move as hitting the area it is animated as hitting, they tend to generalize the area. It probably has to do with that being some fundamental heuristic people have or something, but the reason doesn't much matter. Another main reason is that any fast movements, like a foot's placement while walking compared to where it is at the start of any moves done out of walking, are not going to be noticed on the fly as the reason why something hits or misses. The same goes for subtle movements like in this clip ( https://youtu.be/hUtn1U6SNKE ), they tend to be missed while the game is being played at full speed. This can make the game feel unpredictable in ways that make sense when they are played back in slow motion, but rarely make sense in the moment.
  8. The other issue with high accuracy is that it directly ties animation to balancing. The common solution, as you could see in that Skullgirls clip, is to just make things less accurate when needed. In general the game will look good, because most of the other animations will still match the hitboxes and because it can use a wider range of animations without needing to balance around them. The main point to take away from this is that most high accuracy boxes might not feel good to play with, but it will often visually make sense.
  9. The general boxes that most games use have a opposite feel and look than the high accuracy ones do. This means that while playing games moves will likely perform how they are expected to, but when looking at the game in a slower setting it can look janky. It is more common that something that looks janky will be noticed in a game, especially if something has a freeze frame or there are limbs sticking out that aren't supposed to be hit (like walking). I'm sure if you've played a fighting game with this style you can probably think of an example. The plus side to this is that it is very easier to have a hitbox interaction behave in the intended way in all or nearly all cases. Take this hitbox for example ( https://sfvsim.com/diff/CNL/004/scripts/673_4HP.html#7 ), this is Chun-li's 4HP from SFV season 1. This move was intended to cleanly beat any pokes that hit above the waist, so it just has a huge hitbox there that hardly matches the animation. As you can see in this video ( https://youtu.be/akAzWpqJ3tY?t=259 ) the hitbox does not match the animation (the video is post nerfs, https://sfvsim.com/diff/CNL/005/scripts/673_4HP.html#7 this shows the changes). If that hitbox did match the animations then there might have been a poke that misses the hands. Unless you want to test every poke against this move to make changes, making it a large unrelated box is how you ensure it is consistent. The main reason to not use high accuracy boxes is because it makes interactions much easier to predict.
  10. Obviously, the animation should at least imply what the boxes looks like or the game is going to look and play like pure jank, so there is a minimum requirement of accuracy (but games like MVC or Melty Blood can and have gotten away with some wild things).
  11. Hitbox/hurtbox ratios and what they do:
  12. This is my favourite part by the way, it's super interesting. Also I'll mainly be talking about SFV because it put a ton of care into interactions and the best community tools exist for it ( shoutouts to https://sfvsim.com/ ). You can play around in SFV sim to see how different hitbox ratios and placements interact with each other, it is very interesting if you are a loser like me.
  13. To start off with, most modern games (as in games developed after developers were able to learn from the mistakes of "early" fighting games (but honestly even modern games still sometimes forget the lessons of the past)) have a mix of what the ratios of hitbox to hurtbox of a move should be to dictate how a move typically should be interacting with other moves (while they are active obviously). I'll quickly show a bunch of differing ratios from the same games just to show that they really do mix.
  14. GGXrd:
  15. All hitbox https://www.dustloop.com/wiki/images/0/0b/GGXRD-R2_Johnny_Mist_Finer_Mid_Level_1.png
  16. Covered by hurtbox https://www.dustloop.com/wiki/images/1/17/GGXRD-R2_Johnny_j.K-Hitbox.png
  17. Mixed https://www.dustloop.com/wiki/images/c/c2/GGXRD-R2_Johnny_f.S-Hitbox.png
  18. GG tends to lean towards less having less of a hurtbox than say SF, but it isn't a rule by any means. If people were hosting images of other character's moves there would be examples of grounded covered normals, Johnny is just privileged.
  19. SFV:
  20. (almost) all hitbox https://sfvsim.com/diff/KRN/042/scripts/613_2MP.html#5
  21. Covered by hurtbox https://sfvsim.com/diff/KRN/042/scripts/604_5MK.html#8
  22. Mixed https://sfvsim.com/diff/KRN/042/scripts/601_5MP.html#5
  23. KOF02 (because I had it open anyways):
  24. All hitbox http://wiki.shoryuken.com/images/7/7f/Leona02_duA3.png
  25. Covered by hurtbox http://wiki.shoryuken.com/images/7/73/Leona02_jC.png
  26. Mixed http://wiki.shoryuken.com/images/2/2d/Leona02_stC.png
  27. ST (because it's Vega's gimmick):
  28. All hitbox (other than moving the whole character) http://wiki.shoryuken.com/images/f/f4/Vega_stfarshrt3.png
  29. Covered by hurtbox http://wiki.shoryuken.com/images/a/a5/Vega_stfarstrng3_noclaw.png
  30. Mixed http://wiki.shoryuken.com/images/5/55/Vega_stfarstrng3.png (same move as ^)
  31. Skullgirls:
  32. All hitbox http://wiki.shoryuken.com/images/0/0a/SG_par_cmp_hb.png
  33. Covered (right side is the target combo) http://wiki.shoryuken.com/images/2/2a/SG_par_cmk_hb.png
  34. Mixed http://wiki.shoryuken.com/images/4/42/SG_par_shk_hb.png
  35. With that out of the way, normally the point of how the ratio works out is directly related to how often you want the move to win or lose trades. This is why moves like good counterpokes often would have a huge hitbox that is detatched from the hurtbox (and also usually a lot of active frames, but that's besides the point). SFv has a really great example in how Chun-li's 4HP was changed. There have been links to both pre and post nerf hitboxes, but here they are again, pre ( https://sfvsim.com/diff/CNL/004/scripts/673_4HP.html#7 ) post ( https://sfvsim.com/diff/CNL/005/scripts/673_4HP.html#7 ). As you can see, they decided to nerf only the ratio of hitbox to hurtbox (other than lowering the top because it AA'd when it shouldn't have). The reasons they did this was because a common standing poke looks like this ( https://sfvsim.com/diff/CNL/042/scripts/601_6MP.html#7 ), and if you were to try to use even the most perfectly spaced poke against the 4HP before the patch it was likely that the 4HP would win anyways. That was too strong because it was so good at stuffing things that it removed the ability for other poking characters to challenge Chun-li. So they nerfed it, and they did that in a way so that it could still cleanly beat some of the moves that were covered in a hurtbox, but against stronger pokes it would only trade at best (ignoring the priority system), and against well spaces pokes it would lose outright.
  36. On top of this, this means that by modifying hitbox to hurtbox ratios you can turn some moves that would normally be horrible counterpokes into moves that have good enough of a hitbox to counterpoke only specific moves. For example, normally as a counterpoke or poke Blanka's 2MK ( https://sfvsim.com/diff/BLK/043/scripts/611_2MK.html#6 ) is unreliable because it is going to lose to anything disjointed, and if it hits on another move's active frame it will usually trade at best (assuming the active frame hits the extended hurtbox). However, against a move like Cammy's 2MK ( https://sfvsim.com/diff/CMY/042/scripts/616_2MK.html#7 ), which is a really good poke because of the reward alone, Blanka's 2MK becomes a solid counterpoke because there is a large area where Cammy's hurtbox goes much farther out, and that's all it needs to hit. Here is an example of how that plays out in SFVsim ( https://sfvsim.com/#CMY*0*0*0*0*0*0*HIT*NONE*%5B%5B%22STAND%22%2C0%2C0%2C~%5D%2C%5B%222MK%22%2C~%2C~%2C~%5D%5D!Z27*2.95*0*0*0*0*0*GUARD*BACK*%5B%5B%22STAND%22%2C1%2C0%2C~%5D%2C%5B%222MK%22%2C0%2C32%2C~%5D%2C%5B%22Rolling-H%22%2C~%2C~%2C~%5D%5D!BOTH ).
  37. In general there isn't really a point where the exact ratios need to be such and such until you get to the point that a poke is only able to be counterpoked by a longer range poke. This is one of the places where the issues show up because then there is no changes in the interactions except in three distinct places. The three places are, neither move hitting, the longer move hitting, and both moves hitting. Neither move hitting is just there because it has to be, in general this interaction is roughly the same as not pressing a button because it's either going to go back to neutral or someone is getting whiff punished. One move hitting is going to keep the risk vs reward static until we get to the third place where both hit. This can lead to boring neutral games because there can be a no man's zone where one player has no reason to press any buttons because they will at best whiff without being punished. Getting into the third place has some interest because the timing of the buttons will matter more here and the player with longer range can attempt to move and cause a whiff, which doesn't have any incentive in the no man's land. The other thing that can cause issues is if every move is surrounded by a hurtbox with a mostly universal size always, because then counterpokes can only win clean if they hit an extended hurtbox before the active frames come out (assuming that there aren't things like a priority system to handle trades dynamically). This also means that in order to have counterpokes that don't trade there would need to be extended hurtboxes before every move's active frames, and moves that you want to not lose to counterpokes as much would need to have less of a extended hurtbox. If you do that then every move would also double as a counterpoke and the only thing that would make them better or worse at it would really be active frames (and like height or priority and whatever other systems the game has). This is much more rare of an issue to see in games because if you take those steps to support counterpokes you would usually just see that it makes more sense to not drastically change your poke's hurtbox to cover the hitbox when you could instead just keep the half extended hurtbox while the active frames are out. So usually if a game has entirely covered hitboxes it's a game designed before the whole poke/counterpoke meta was well known, or it is someone trying to experiment with something, or it is someone who just doesn't know.
  38. Those reasons are why almost every single modern fighting game has a variety of ratios between hitboxes and hurtboxes, it's really just a useful tool to get moves to do the things you want them to. Trying to stick to one ratio is going to make most moves act the same, much like how if every move had the same range or startup/active/recovery would also make a lot of things act pretty similar, and there is arguably no gain at all to not having mixed ratios.
  39. This whole thing doesn't really apply strongly to all fighting games. This mainly applies to games that play around in the same reaction space as things like 2D capcom or arcsys games, where there are a ton of moves that need to punished on a guess in neutral. This doesn't really apply much to most of an NRS game or nearly every modern 3D game (although I assume that it would apply to a 3D game that did require things like counterpokes to the level of most 2D games), because having a move whiff meaning that your opponent can usually punish you hard heavily reduces the amount of trades and near trade interactions in neutral. It doesn't matter much the ratio of a hurtbox to hitbox if 90% of the time there isn't a reason to have trades at different ranges. That being said, as far as I know from playing way too much Soulcalibur is that I think there is some care taken to have a mix of hitbox to hurtbox ratios in that particular 3D game, but it also doesn't seem to be a big focus.
  40. Anyways, hitboxes are cool, they are like half the feel of a game even though they are only a background system.
Add Comment
Please, Sign In to add comment