The 1-Pixel Collision Box

In the early 90s, a trend developed among shoot-’em-ups that was affectionately (or was it rancorously?) dubbed “bullet hell.”

An example of bullet hell. It gets worse. Much worse.

The impetus behind it was to add visual flair to 2D games in order to compete with the craze surrounding 3D games. Arcade cabinets were more powerful than ever before, so these shooters could handle many more sprites and the extra calculations required for the accompanying collision checks. The approach worked relatively well, providing plenty of “Holy Shit!” moments. However, there was one major issue: playability.

ikaruga2
Ikaruga’s dual colour scheme was the focus of its gameplay.

Shoot-’em-ups tended to be one-hit kill games, and simply saturating the screen with harmful projectiles made them incredibly difficult (if not downright impossible). Now arcade games are meant to take your money, but that wouldn’t happen if no one played ’em. In order to implement bullet hell without alienating customers, something had to give.

Part of Ikaruga‘s solution was to make all bullets and your ship one of two colours, and then simply ignore collisions between like-coloured objects. Other titles used shields and various powerups, but the original solution, and, in a way, the purest to the genre, was the (roughly) 1-pixel collision box.

Instead of surrounding the majority of the player’s ship with an area susceptible to fire, a single pixel was used to indicate its vulnerable spot. This was a rather elegant solution as it required no other changes and didn’t present an extra hit to performance. Players were also less likely to feel cheated if they came out on the positive end of some collision-fiddling. The end result looked something like this:

The visual oddity of having the player’s ship fly straight through harmful projectiles was lessened by the nature of the top-down perspective. This view had issues with representing depth/elevation, and that actually made it easier to imagine bullets just skimming over the player’s ship. The 1-pixel collision box also had the side-effect of making the player feel more skilled at the game, which — in the very least — provided the illusion of empowerment.

Saint Dragon’s eponymous lead actually looked pretty good in motion.

As a side note, something of a similar concept was used in an old Amiga shooter called Saint Dragon. The player’s ship, the eponymous Saint Dragon, consisted of a head and spiraling tail. The head used regular collision detection, but the tail was purely aesthetic. This added scale and personality, and allowed the player to control a vehicle that seemed grandiose despite being virtually identical to countless other shoot-’em-up ships. The head could be easily destroyed, but the tail would actually absorb many types of bullets and even damage most of the enemies it touched.

Supplemental:

Since I’ve posted this article, it’s gotten a lot of attention from numerous shmup enthusiasts. Many have been eager to bring up the specifics of the tiny-collision-box phenomenon, as well as variations on the theme. I think that’s great, and one of the sites that has been pointed out to me contains a lot of interesting information on the genre. Of particular note are its threads on shmup strategies, the dos and don’ts of good shmups, and the glossary of common shmup terms.

11 Comments

  • re st. dragon: “purely aesthetic” suggests to me the enemies and bullets passed straight through the tail without collision detection – they didn’t, the St. Dragon tail was collision detected, but indestructible and destructive, a bit like an R-Type orb thingy in terms of 2d shooter gameplay mechanic, only you began with it, and of course you had to learn to move to wrap the tail around the head to shield yourself to have any hope of getting very far (and the tail didn’t detach and fire at stuff like an r-type orb :-))

    As you can see in this video at times:

    http://www.youtube.com/watch?v=MeY9kOot9iI

  • Huh, for some reason I remembered it mostly passing through projectiles and enemies. Obviously it absorbs most bullets in the video, though, so you’re right, it’s not just purely aesthetic.

    My bad.

  • FYI Ikaruga did not “introduce” this concept, it evolved naturally over time. Maybe you meant “introduce” just as a figure of speech, but this is worth mentioning: If you start with mid 80s arcade games you can see where and how the hitboxes got smaller and smaller until they contracted to a tiny point much smaller than the sprites they were supposed to represent, reaching Ikaruga-level smallness I think with dodonpachi (not sure about that). But it was actually pretty gradual, games like Raiden or Battle Garegga had fairly small vertical hitboxes, and even Gradius had a pretty tight hitbox. In fact the idea has existed for so long, I’d say it’s almost more like the sprites around the hitboxes got bigger and more elaborate than the hitboxes themselves got dramatically smaller.

    I don’t think the hitboxes being too big was ever a major issue, either, except among developers that were terrible at making these games. And I would not go so far as to say which came first: the bullet patterns or the hitbox, because the two concepts work so closely in tandem with one another that they were probably conceived simultaneously in, again, a natural way. Some developers probably thought, “How can we keep players playing longer?” and the solution was “Make it harder for bullets to hit them, so it feels more ‘fair'” and then the concern would be “Wouldn’t that make the game too easy?” and the next solution would be to add more bullets.. But it also probably just as often went the way you said, where they were looking to add more bullets, and felt they had to reduce the hitbox size out of fairness.

    Also there’s a lot of nuance to this idea, too. Most of these hitboxes aren’t literally a single pixel. Also sometimes (oftentimes?) the hit detection works in different ways, like a bullet needs to completely fill the hitbox in order for it to register a hit, rather than just touch the edges or overlap (it might seem like a minute difference but it causes the game to play a lot differently, and makes the hitbox itself feel more nebulous and uncertain).

    If you can tolerate the elitism, go to the shmups forum (http://shmups.system11.org/) and try reading a scoring/strategy guide and you can see the variety and depth to these games. Most game designers probably would not want to delve to deep into these ideas, because they are very niche and don’t have very wide appeal, but I think it’s nice to know that the ideas are out there.

    I guess this is all just nitpicking but I hope you welcome this level of rigor in observations about game design.

  • Yes, of course it’s welcome.

    As for Ikaruga, I simply mentioned that it used a flipping mechanism in tandem with the colour-coded bullets, not that it was the first example of 1-pixel collision boxes. After all, the idea of bullet-hell (as loosely defined a concept as it is) is popularly attributed to Batsugun and all the subsequent Cave shooters.

    Now collision boxes themselves never wholly encompassed the player’s ship’s as that’d be incredibly frustrating. Instead, they were always catered to the ship itself, and yes, sometimes they were quite small. However, from what I noticed, they were never shrunk down to just a 1-pixel box until the arrival of bullet hell shooters.

    The concept of the collision box “filling up” or completely encompassing a projectile is new to me, though. And yes, you’re right, it would behave differently. For example, a bullet heading down-right over the player’s ship could overlap a large portion of it, including its center, without being registered as a hit if the far right side of the ship’s collision box was not touched. This would mean that unlike 1-pixel boxes (which always register a hit upon being intersected/overlapped), these “full boxes” would allow bullets to virtually pass over every part of the ship without hitting it.

    Interesting. Thanks for the info/link.

  • Please note, the collision mask in Touhou (from the embedded video) is more like 6×6 pixels. Also, the Touhou series is harder than you think, most people I know can’t get past the first level on the standard difficulty (depending on the game).

  • retrogamer is right: the collusion boxes are often 5×5-ish or more. One reason why small collusion boxes is nice is then your character/ship can be creative. For example, instead of a streamlined ship that is made to evade bullets you can have cute girls with ribbons dangling all over the place. Definitely improves sellability.

    Two very neat things from many bullet hell shooters, including Touhou games, that you didn’t mention:

    Some games give bonus points / firepower / etc. when you “skim” bullets. This means that if a bullet hits your ship but not the collusion box you get extra points, etc. Some games even actively encourages you to put yourself in harm’s way if by offering extra score multipliers that can only be earned by getting “hit” like this.

    Many, many games also offer a button that lets you slow down your ship and makes the hitbox visible. It reduces your speed but makes dodging many things easier. Some games also alter your weapon when this button is pressed; most often your spread attacks become concentrated so you can stay under a boss and shoot it with all you have. By enabling this mode you really focus only on the area next to your ship and to do well in a lot of these games you need to shift your focus seamlessly between normal, spread out, kill everything mode and focused dodging mode.

  • Interesting, especially the first point. I also liked how Galaga Legions simply increased the amount of damage you’d do if you were close to the enemy.

Leave a Reply to wing Cancel reply

Your email address will not be published. Required fields are marked *