Tears of the Kingdom's Bridge Physics Have Game Developers Wowed 80
Nicole Carpenter, reporting for Polygon: There's a bridge to cross the lava pit in The Legend of Zelda: Tears of the Kingdom's Marakuguc Shrine, but it's broken. More than half of the bridge is piled on top of itself on one side of the pit, with one clipped-off segment on the other. The bridge is the obvious choice for crossing the lava, but how to fix it? A clip showing one potential solution went viral on Twitter shortly after Tears of the Kingdom's release: The player uses Link's Ultrahand ability to unfurl the stacked bridge by attaching it to a wheeled platform in the lava. When the wheeled platform -- now attached to the edge of the bridge -- activates and moves forward, it pulls the bridge taut, splashing lava as it goes, until the suspension bridge is actually suspended and can be crossed.
But it wasn't the solution itself that resonated with players; instead, the clip had game developers' jaws on the ground, in awe of how Nintendo's team wrangled the game's physics system to do that. To players, it's simply a bridge, but to game developers, it's a miracle. "The most complicated part of game development is when different systems and features start touching each other," said Shayna Moon, a technical producer who's worked on games like the 2018 God of War reboot and its sequel, God of War: Ragnarok, to Polygon. "It's really impressive. The amount of dynamic objects is why there are so many different kinds of solutions to this puzzle in particular. There are so many ways this could break."
Moon pointed toward the individual segments of the bridge that operate independently. Then there's the lava, the cart, and the fact you can use Link's Ultrahand ability to tie any of these things together -- even the bridge back onto itself. [...] Tears of the Kingdom was seemingly built on top of Breath of the Wild, reportedly with a large portion of the same team working on it. "There is a problem within the games industry where we don't value institutional knowledge," Moon said. "Companies will prioritize bringing someone from outside rather than keeping their junior or mid-level developers and training them up. We are shooting ourselves in the foot by not valuing that institutional knowledge. You can really see it in Tears of the Kingdom. It's an advancement of what made Breath of the Wild special."
But it wasn't the solution itself that resonated with players; instead, the clip had game developers' jaws on the ground, in awe of how Nintendo's team wrangled the game's physics system to do that. To players, it's simply a bridge, but to game developers, it's a miracle. "The most complicated part of game development is when different systems and features start touching each other," said Shayna Moon, a technical producer who's worked on games like the 2018 God of War reboot and its sequel, God of War: Ragnarok, to Polygon. "It's really impressive. The amount of dynamic objects is why there are so many different kinds of solutions to this puzzle in particular. There are so many ways this could break."
Moon pointed toward the individual segments of the bridge that operate independently. Then there's the lava, the cart, and the fact you can use Link's Ultrahand ability to tie any of these things together -- even the bridge back onto itself. [...] Tears of the Kingdom was seemingly built on top of Breath of the Wild, reportedly with a large portion of the same team working on it. "There is a problem within the games industry where we don't value institutional knowledge," Moon said. "Companies will prioritize bringing someone from outside rather than keeping their junior or mid-level developers and training them up. We are shooting ourselves in the foot by not valuing that institutional knowledge. You can really see it in Tears of the Kingdom. It's an advancement of what made Breath of the Wild special."
Re: (Score:2)
Wtf are you talking about? NIntendo has always released shit exclusively on their own platforms.
Re: (Score:2)
I haven't been following Nintendo much lately, but haven't they been releasing smartphone games? I don't recall them making any smartphones.
Re: (Score:2)
Wtf are you talking about? NIntendo has always released shit exclusively on their own platforms.
Nintendo recently sent a C&D to Valve to stop them from distributing the Dolphin emulator. Normally I would expect Valve to defend developers on their platform, but they actually were the ones who notified Nintendo that Dolphin is getting set up to distribute on Steam.
At this point, whenever someone mentions TOTK, all I think about is the legal flailing of Nintendo to halt legal distribution of software that followed the TOTK release. I suspect I'm not alone in making this association. There's someth
Devil's Advocate (Score:5, Interesting)
With all screwing around aside, that is fairly impressive for such a small portion of gameplay.
Power Not Push (Score:1)
The arrow didn't exert nearly enough force to be able to push the cart across a pool of lava
I don't have the game so do not know if this is correct, but from watching the video I thought the arrow basically "powered up" the wheels, as it had a lot more the look of a machine that had been switched on...
The other points I agree with, however overall it looks fun. I'm pretty forgiving of clipping as long as it's not affecting gameplay.
Re: (Score:2, Informative)
Yes, the in-game rule is that hitting a "Zonai device" with a weapon switches it on.
Re: (Score:2)
The bit that caught my eye mainly is how I've seen most physics engines, Havoc in particular, make things flop around wildly for no particular reason, with some things that were at rest suddenly flying in the air at times despite nothing ever touching them. But we don't see any of that here, the movement seems to be pretty smooth.
Re: (Score:2)
The arrow didn't exert nearly enough force to be able to push the cart across a pool of lava
No it didn't, that's not was was happening in the video. Link placed a charged Zonial Cartin the lava and activated it by hitting it with an arrow.
The bridge clips badly through the ledge on the other side
I don't see the clipping. Are you confusing it with the shadow being cast? Looking closes the bridge segment is actually there but in the shade so at a glance it looks like it clipped, but on close inspection it doesn't appear to be the case.
Re: (Score:3)
How come the tires didn't burn in the lava?
Re: (Score:2)
Because Dante's Peak (1997 movie)
Re: Devil's Advocate (Score:2)
Because Zonai devices are magic.
Re: (Score:2)
The lava completely failed to deform the cart as well as the bridge segments
It's not supposed to. The gameplay wouldn't work if it did. That's just the rules of the world. Lava will burn wood or melt ice, but won't damage other materials.
The bridge clips badly through the ledge on the other side
It's not clipping. It's moving through a shadow.
The arrow didn't exert nearly enough force to be able to push the cart across a pool of lava, let alone up the ledge on the other side
It's a mechanical device. The arrow isn't pushing it, it's turning it on. Another common mechanic of Zelda games.
Re:Devil's Advocate (Score:4, Insightful)
Nintendo games tend to favour allowing the player to experiment and find interesting solutions, over being completely realistic. It makes sense in the context of the game and how things work in that world, and the player is rewarded for coming up with a novel solution.
It really is impressive, not just because it's a great use of the physics engine, but because Nintendo managed to give the player the freedom to build machines out of component parts without breaking the game completely.
The bit about institutional knowledge is right. Many companies lay people off when times get slightly tough, but Nintendo, like many Japanese companies, retains them through thick and thin.
Games used to have physics simulations... (Score:5, Insightful)
Newer games have mostly abandoned realistic physics, and that is a just a pity, but probably caters to the mostly ignorant audience that cannot distinguish fiction from physics, anyway.
Re: (Score:2)
What you're describing is partly or entirely preprogrammed physics events, rather than the ability to string things together arbitrarily and have them not result in chaos.
Re: (Score:2)
That's probably true of most examples, but Garry's mod proves that Source (and thus Half-Life 2, Portal, etc.) could do this sort of tomfoolery in '04, on a single-core PC with a ~128MB video card and 512MB RAM. HL2 requires a 1.7GHz (single core) CPU and a DirectX 8.1 GPU. The switch has a quad-core 1 GHz CPU, 4GB RAM total, and 256 Maxwell-based CUDA cores...
Re: (Score:2)
Kinda but only kinda.
I remember a friend of mine getting some sort of sandbox and nail gun mod for half life 2 back in the day. We had a lot of fun mailing things to other things, like 20 jet packs to a burned out tank etc and watching the physics in action. It didn't generally take all that much for the solver to get unstable and for the equations to oscillate and diverge.
You can deal with many many objects (like world of goo), if you don't do collisions and, crucially have everything in a fairly narrow ra
Re: (Score:2)
I honestly don't see what the fuss was about with the bridge video. There aren't that many things in the scene, they're all rigid bodies with simple shapes, and nothing is moving unusually fast or slow. I would expect the average developer to be able to write a little toy physics engine that could handle a scene like that in a quiet afternoon or over a weekend. (If that sounds crazy to you, do a search for Verlet Integration. This kind of thing isn't nearly as difficult as you might think.)
What's amazi
Re: (Score:2)
Agree. Wake me when the lava convincingly interacts with the bridge and cart. Or when waves crashing on a beach look realistic.
Re: (Score:2)
Re: (Score:2)
or turning tanks into flying machines by attaching balloons, but that already gets us into the "unrealistic physics" section.
Yeah because THAT suspended my disbelief in a game which has a gun that causes people's heads to inflate like a balloon and float away, and where you can grapple hook into the ground to survive a fall from incredible height (because nothing makes surviving a fall more likely than hitting the ground even faster!) :-)
Re: (Score:2)
Speaking as a game developer, I wouldn't agree with that. I think you're just remembering the more memorable stuff from a time when advanced physics engines were uncommon. Physics engines were expensive to license back then, so a lot of that was probably set pieces designed to show off the fancy physics tech.
Nowadays most games are made in Unity or Unreal, both of which have quality physics engines built in. If you're building your own I think it's much more common now to have realistic physics in general,
Re: (Score:2)
Speaking as a game developer, I wouldn't agree with that. I think you're just remembering the more memorable stuff from a time when advanced physics engines were uncommon.
Then please name examples of contemporary games that support the likes of "shooting branches off trees", "collapse buildings partially or completely", "blasting holes anywhere in the ground that you can take cover in".
I would really like to play such games, but have not seen much of such, lately.
Re: (Score:2)
Games still have physics engines, but the novelty wore off and we don't see "physics puzzles" so much anymore. This is in many ways a good thing. Simulated physical interaction between the player and other game elements is fundamentally unpredictable. Making a puzzle out of it means you either make it super contrived so only the intended interaction can occur (like this example), or it becomes a frustrating fiddly mess that the player can screw up in an infinite number of unexpected ways. Even if a puzzle o
Re: (Score:2)
Believe it or not, it's not that hard to get predictable behavior out of a physics engine. Simple things like not tying your time step to your frame rate go a long way to making things behave more consistently.
Making a puzzle out of it means you either make it super contrived so only the intended interaction can occur (like this example)
I don't think that's an accurate. In the Twitler thread, there are quite a few people talking about alternative solutions to the same puzzle. Also, despite what the summary would have you believe, the physics on display in the video aren't at all impressive. I'm sure the game shows much more inter
Re: (Score:2)
What you can't get is predictable behavior out of a player. When you have too much freedom to mess with physically simulated things, someone will always find a way to "do it wrong". Predictability in the physics engine only comes into play if the only possible physics interaction is the desired one (see "contrived").
Re: (Score:2)
What you can't get is predictable behavior out of a player. When you have too much freedom to mess with physically simulated things, someone will always find a way to "do it wrong".
We speak of games here. There is no "wrong" way to do things in games, as long as the player is entertained.
If a game gets into a weird/broken state because it did not prevent you from trying something funny/unforeseen, just load the latest save and play on.
Re: (Score:2)
contrived so only the intended interaction can occur (like this example),
Predictability in the physics engine only comes into play if the only possible physics interaction is the desired one (see "contrived").
That's simply not true, as evidenced by the many alternative solutions used by players in the linked thread. I'd wager that the same is true for many, if not all, of the physics puzzles in the game.
You're trying to imply something by the word 'contrived', but it's not clear what. All puzzles are contrived. That's what makes them puzzles. The designer needed to have at least one specific solution in mind. That doesn't mean that only the solutions the designer considered are possible or that they're the o
Re: (Score:2)
Realistic physics bring some problems that can break the game, or make it unenjoyable. That's particularly true when the player has access to powerful weapons that can reshape the landscape. How can you have an important building be part of the plot when the player can demolish it? You end up with invulnerable NPCs standing in forest clearings with nothing breakable nearby.
Realistic physics just aren't all that fun either. After a few spectacular demolitions, it becomes apparent that building stuff in the r
Re: (Score:2)
Realistic physics bring some problems that can break the game, or make it unenjoyable. That's particularly true when the player has access to powerful weapons that can reshape the landscape. How can you have an important building be part of the plot when the player can demolish it?
Red Faction had a pretty simple approach to this: If a building was vital for some story mission, it would be rebuild by the enemy after you destroyed it.
The physics of heat (Score:3)
Re: (Score:2)
MST3K Mantra time. Link can run around carrying a dozen or more polearms, bows, and shields, hundreds of arrows, scores of meals, and enough plant and animal parts to fill the flora and fauna wing of a museum several times over. The physics of heat is hardly the only thing in the game that's not being taken too literally. :-)
Wrong authority (Score:2)
This technical producer only knows scripted physics. Half Life 2 had physics of this caliber. What am I reading here?
Nintendo-washing the game history (Score:2)
The difference is that this time, Nintendo did it.
Therefore Nintendo invented it and any prior occurrence is instantly wiped out).
(see, e.g., Nintendo "saving" the video game industry).
Re: (Score:2, Insightful)
>Half Life 2 had physics of this caliber
Having looked at HL2 physics demos, I sort of doubt that. HL2 had clumsy ragdolls and classic mechanics for solid objects. Did HL2 have linkages and tensioning as is shown here with the Zelda game?
Re: (Score:2)
Look at the bridge and train wreck collapse at the beginnin of HL2 Ep. 2
Re: (Score:2)
Re: (Score:2)
Look at the bridge as it hits the floor while it unfurls. Its because of the feed back from the friction of the floor that has people impressed. The force of one of those plates dropping obviously feeds force back though the entire bridge system then as the bridge is pulled taunt, you see it wabble in a natural way. What this means is the force vector (including mass, friction, gravity, misplaced jet engine, etc.) is RECACULATED each frame, and updated across the entire connected structure. Halflife2 ju
Re: (Score:2)
What this means is the force vector (including mass, friction, gravity, misplaced jet engine, etc.) is RECACULATED each frame,
That is far less impressive that you might think. The math is simple and not terribly expensive. Also, there aren't a lot of things in the scene shown in the video, they're all rigid body, simple shapes, and moving at reasonable speeds. As for the calculation happening each frame, I would hope they'd have the sense to use a fixed time step.
and updated across the entire connected structure.
Not just the connected structure, but over all of the constraints, multiple times. This is normal.
make it look so god damn natural
That's the easy part, believe it or not. It's not like special care
Reminds me of Tangled (Score:3)
"The most complicated part of game development is when different systems and features start touching each other," said Shayna Moon
Reading about the animation development for the Disney Rapunzel movie Tangled the story developers talked to the animation team and was told that hair was difficult to animate convincingly but that they could do it as long as nothing touched the hair - which as the story developers said "was the whole story". They did get the hair to interact with various objects and environments in the end, but getting different models to interact was a problem.
Neat but... not feeling the 'wow' (Score:4, Insightful)
Okay, that's nicely executed, but I don't see how this is at all revolutionary or worth talking about to this degree? Hell you can even see the bridge segments clipping through the floor at the end of the clip.
Is this a marketing campaign? Or are people just so far up their own asses being Zelda fans that anything remotely impressive is automatically the best thing that's ever been?
=Smidge=
Re: (Score:1)
Re: (Score:2)
This isn't engineering software. It is just a game and that is why it's impressive. Games are not engineered for this kind of thing they try hard to meet tight deadlines without resources dedicated to anything but avoiding crashes and breakage. The number of scripters for many games is not large and serious system programmers will be just a few; all the real work is already done in the massive game engine/libraries with decades behind it.
This is more like Nintendo putting the work they always did when they
Re: (Score:2)
There's a reason Pac-Man endures, and it's not just the Gen-Xers playing for nostalgia.
Re: (Score:2)
> It is just a game and that is why it's impressive.
This kind of physics was mildly impressive 10 years ago, when it was implemented in Flash games. Fuckin' Half-Life 2 was 20 years ago and did more impressive stuff than this.
Really not seeing it here.
=Smidge=
Re:Neat but... not feeling the 'wow' (Score:4, Informative)
Okay, that's nicely executed, but I don't see how this is at all revolutionary or worth talking about to this degree? Hell you can even see the bridge segments clipping through the floor at the end of the clip.
The bridge segments aren't clipping through the floor at all - they're moving through a shadow. Getting a big chain of connected objects to interact seamlessly like that is really really hard to do and incredibly impressive. It's hard to get more than a few objects to interact well. I counted 16 bridge segments there, plus the device, the mounts at each end, and uneven ground. Most games wouldn't attempt anything close to that.
Is this a marketing campaign? Or are people just so far up their own asses being Zelda fans that anything remotely impressive is automatically the best thing that's ever been?
The tweet's from an experienced game programmer who's worked on several AAA games. Looking at his experience on LinkedIn, he's got a ton of experience programming physics systems in video games. This is someone with a ton of experience with the issue at hand being really impressed by what Zelda did.
Re: (Score:2)
The tweet's from an experienced game programmer who's worked on several AAA games. Looking at his experience on LinkedIn, he's got a ton of experience programming physics systems in video games.
If that's the case, then he really shouldn't be impressed by what was shown in that video. Maybe he's been drinking?
Now wait until someone manages (Score:2)
to build something phallic with that bridge.
Somebody was given time to do it right (Score:3)
There is a problem in industry (Score:5, Insightful)
A lot of industries like to toss out the experienced guys and hire fresh blood.
Sure they can come up with novel ideas but often it would be good when these ideas stand on the shoulders of the old ones.
That's what this article is about.
Physics in games is always amazing (Score:2)
I'm working on a Box2D like 2D physics engine for an obscure language I like to play with ( https://qb64phoenix.com/forum/... [qb64phoenix.com] ). The math involved with just 2D is mind-numbing. 3D math for 3D physics is even more mind blowing. Watching the Zelda clips, and mostly understanding the math involved behind the scenes, has me thinking "Wow" every time. It's very impressive to me.
Re: Physics in games is always amazing (Score:2)
Re: (Score:2)
I agree with the sentiment (that implementing physics engines is not for the faint-hearted),
Having done so more than once, I would disagree. The parent is working on a 2d physics engine, which is easily in reach of the average developer. It certainly sounds scary, but once you dig into it a bit, you'll find it's not so bad. Really. Give it a try. A quick search for "Verlet integration" should be all you need to get started. You should be able to knock something cool together in a day or so.
I would suggest that 3D is not much harder than 2D, as long as you are modelling correctly
This is absolutely right. The math is essentially the same, at least in your integrator. Collisions a
Re: (Score:2)
Assuming you know Box2D here, what's so special about 20 or so boxes connected to each other with hinge joints getting pulled by a wheel joint at the end?
News for nerds (Score:2)
Walkthroughs for newbies.
I've seen similar physics on Half Life 2 (Score:2)
Seriously
Okay, why is this so "amazing" (Score:2)
Honestly, I watched the clip several times and to be blunt, I didn't see anything that was "jaw dropping" or amazing.
Now, I'm not a video game expert or even an aficionado, but I've seen a ton of clips of video game animations over the years and this was nothing terribly special to me, I saw nothing that seemed even kind of exceptional.
What am I am missing?
To me it was kinda underwhelming, nothing really awe-inducing. It just looked like some "okayish" video game graphics. Lots of games seem to have much be
Re: (Score:3)
Arbitary connections, not partially- or pre-programmed physics events from pre-designed assets. And in particular, long chains of linkages undergoing constant collisions (long chains of linkages amplify problems).
Physics is a pain to get right.
Re: (Score:2)
Hmmm, okay, thanks. I think I get what you're saying.
Maybe it's so advanced that (to me) it looks not-so-amazing, but I'm probably (definitely) not the best person to appreciate the subtleties of what's actually going on.
If I'd played the game I'd probably be wowed, so perhaps the fact that I don't get it is actually a credit to the programming involved.
Anyway, thanks.
Re: (Score:1)
Re: (Score:2)
KSP is a great example of how hard it is to handle physics of arbitrarily-assembled shapes, including long chains in arbitrary force conditions.
And to think that Kerbal ground collisions are a lot simpler to deal with, too.
So many people confuse pre-programmed or semi-pre-programmed hacks with genuine physics. Which is a testament to the developers, but it gives people the wrong view of how hard genuine realtime physics of arbitrary shapes is.
Re: (Score:2)
Nothing in that demo is particularly impressive. Other posters have pointed out far more impressive things done with much older tech. Also, the bridge in question is hardly a long chain of things.
And in particular, long chains of linkages undergoing constant collisions (long chains of linkages amplify problems).
You're overthinking the problem. The "linkages" are just constraints and you're only dealing with collisions between boxes and planes. Your code shouldn't care if there are two or two-hundred links in the chain. Hell, it shouldn't even have the concept of a chain. Constraints are constraints. The complex intera
Re: (Score:2)
You say your code shouldn't care, but it does.
It very much does.
Sure the logic runs the same no matter how many objects you have, but the.ODE solver at the core will run out of precision and diverge, or grind to a halt unless you are exceptionally good at writing this kind of thing. And the interaction between the solver and collision detection is, frankly unpleasant from a stability point of view. This kind of thing scales poorly especially when you have many orders of magnitude in stiffness between the el
Re: (Score:2)
You don't need to be "exceptionally good", you just need to avoid introducing needless complexity. A global solver adds a ridiculous amount of needless complexity. Be smart and use an iterative solver. You get to deal with just one constraint at a time in the simplest ways possible. Interactions, where solving one constraint violates another, will get handled on the next pass. Just a few iterations and you'll end up with a 'good enough' solution.
So, no, your code should not care that some things are lin
Re: (Score:2)
You have just invented an untested algorithm on the fly, as well as assuming an awful lot about how I think algorithms of this sort work.
Just a few iterations and you'll end up with a 'good enough' solution.
That will happen, say, 99.5% of the time. The other .5%, it diverges and fragments of the bridge wind up somewhere in the NaN-o-verse, or something sitting place wobbling like crazy in a weird manner. If you heavily constrain the levels, then you can construct them to avoid this sort of problem most of t
Re: (Score:2)
You have just invented an untested algorithm on the fly,
No, I didn't. [cmu.edu] This is pretty standard stuff. [ncl.ac.uk]
as well as assuming an awful lot about how I think algorithms of this sort work.
What basis do I have other than your comments here? After that last bit, I'm starting to wonder you're familiar with the subject at all. I thought you might have had some experience with Euler integration, given your comments about stability, but the previous quote implies that you're somehow completely unaware of common methods. (Just an aside, Verlet integration is significantly more stable and ultimately more accurate that Euler's method.) There are some
What Physics? (Score:1)
You can freeze to death on a mountain, but getting hurtled high above them with out so much as a sniffle?
I haven't come across that particular shrine yet but did you notice that the parts of the bridge that were submerged came up aut of the lava without a liberal coating of solidified magma? Anyways it is all just fantasy make believe and doesn't bear much scrutiny.
It's only new because nintendo? (Score:2)
I think it's a great addition to a console game like Zelda, I have it, it's a fun game! So, props for that. But as for physics, I mean Valve's half life 2 engine had great physics. If you ever played garrysmod, people make insane contraptions linking parts together and snaping objects together to make bridges or other physic based items with multiple models slapped together, even with thrusters and other things.
Nevermind when you step into space games like Space Engineers that has physics, deformation from
Hobbyist projects (Score:2)
So it was built into game to do so (Score:2)
Physics (Score:2)
Game developers apparently need to get out more and play other games.
This is literally just basic physics engine stuff, nothing fancy whatsoever. Hell, I'd be annoyed if this *didn't* work in any decent modern game (by modern, I mean since Havok / PhysX / Half-life 2 at least).
Honestly, it's barely even half a dozen physics objects, and does exactly what things like Bridge Constructor etc. were doing with countless dozens of freeform objects.
This is just nonsense hype, written by someone who hasn't played
Why are they wowed? (Score:2)
I've watched the clip and I don't see it. Physics have been in games for decades and some of the systems are very complex and lead to divergent behaviour. E.g there are games where you demolish landscape and buildings, others where you assemble contraptions and drive around in them etc. Having a bunch of segments linked together comprising a bridge doesn't seem that impressive or especially hard compared to what other games have done.
Lonk is a Fashion Icon (Score:2)