Constraint on two axes (2D physics simulation)

[xcasadio]

Hello.

I could not find in the demos or in the code the possibility to put a constraint on the axes.
In some physics engine you can define a constraint on an axis. For example if you create a scrolling game on X and Y, you can't move on the Z axis.

I looked at the constraints implemented in BepuPhysics.Constraints but I didn't see anything like that.

Does it already exist or do I have to create a new constraint?

[RossNordby]

LinearAxisServo would work (the second connection can be a 'world' kinematic body), but a constraint may not actually be necessary here. Simply setting the position/velocity every timestep may be sufficient, depending on the requirements.

Either way, be advised that this doesn't totally delete the third degree of freedom, and sufficient force could smash two objects into each other such that the shortest path to separation is along the restricted axis. In that case, the colliders won't separate in the 2D plane. To avoid this, you can make sure the objects have sufficient depth that the restricted axis is never a good choice, or detect such collisions based on the contact normal and handle them directly.

[damian-666]

Wow this got me thinking deep, I do 2d physics game as well. but now I'm thinking in voxel space like a some of the more logical quantum theorists , this is because ..I have to use eulerian Fluids in discreet density fields and rigid bodies interaction with either sdf or the box2d way and some chimera method to put those 2 way interactions, an adhoc invention thet is ok. So I'm mixing discreet eulerian desnity grids, and field models. They leak pessure masss, and energy thouhh. But stable ansit's ok for a game. But this question give me an idea similar to what ive been thinking about..and make me thinkni should study more and consider to use ur engine. Similar to what you do.. what if you etrude the 2d bodies in the 3d dimensión by the time step , x velocity. Sort of minkowskian. Now u have a convex hull or some spiral or 3d shape which u can tesselate as if static. But keeping the timelike faces only in time. Woud it then be posible to do ccd using this extrusión with just a static model as you say...of the time slice. Aslo You could rast rays and stuff on the solids 256 at a time raymarchig? Or with ur tree or whatever. .like in cdd but without 4d math, more like lighting in 3d existimg methods. but for real im wondering if j should use ur engine instead of a c# port of box2d. With only 8 core gpu vector raycasts on a tree and thats.slow. still is increadibly fast in net7 i get 300 to 4000 fps woth two creatures slight wind....but its not multiphysics and I don't want to use 100000000 sph particles, I do oceans. I want to do simd raycasts or gpu on 1km space at 6cm resolution So I'm thinking how can I use raster tech for rays on gpu w thin rigid bodies in 2d ? This idea came up in some 70s adn again in recent papers and they often use 6d because some geniuss the navier stokes pressure solver in a padded fourier space becausenits scales well.. So complex, imaginary vectors x 2 d plus one time =6d. But the grid is 1024 x1024 not 128x128... the results are super fast and no upscaling tricks needed. Now i will getting pretyy deep, but i think u might intuit it. and maybe a bit weird but I can show u the papers if u want to be distracted . Susskind,.Peskin, it's not quacks.Whitten, Hawking.. all consider this model it's get pretty deep.into wave particle eqivalence..we gotta chose one and as a sufer i care abour waves . Especially long period waves. Thry hit like mike tyson when they break their functinn and become partices ,the lip. reason is some of the physics on quantum gravity use this model is because e =.MC 2 not cubed, it's holograpic theory. Mass of black hole is also related to radius sqared. Why? It'sAsphere? So why not cubed? So was born the hologram model..it's hard to grasp. Even in 1 d. No one wants randomnes, 5d cosmology, compact dimensions, and shrodinders CAT doesn't. on a more practical note, so far, They do it for weather sims on a 3d globe and for cosmology to know the geodesics so we can maybe travel without blowijgn gas.out thr back of a crude rocket.... might sound mad but what u are doing is what many physicists finally decided they should...does it make sense to you?.. and when you said "smash the particles so that thry go in the restricted axis..which is time, youve just done an e = mc2. If Youve collided mass such that it is veer in to the spatial direction only, nownits lightlike and a huge wave packet of mass that was propagating the lattice in In time (think tsunami scale its the speed of light ).. a tsumia was a wavelenght.of basically the whole ocean, at 2 in high whe it goes up 90 degrees via shallpw water.massivr kdv energy, just Turned 90 degrees, into to a bunch massive dispersive particles. A Floyd is Mas flux. But back to some uranium, it's Turned by collison, and now a shorter light wave and gamma and other wave in space converting most of its mass. it's a flash of light a mini sun, and possiblly Abomb. In x, y spacelile dimensiones, only massless radiation can travel forever in, but mass just fly thought time mostly until its gets near light speed it lags a bit as in time dilation. But collide mass in a particel accelerator or split and atom so it finds spacelike geodesics closer...and they exchange form. As it's sudenly Steer into spacelike , and u either made.some.quarks or gamma or, an aromic bomb. If Mass is just waves packets or solitons that lag behind and travel moslty in time dimensión, then we don't have all the paradoxes and contractions in Gr fields.and quantum reductivism. It's funny making physics games make.me read all these cosmology quantum and and relavity papers and it's in agreement with gr but not a field theory it's an abelian group theory. None are verified but then again nones got a.better gravity than gr and we know it's not explaining many things. Also material scientists use this stuff for graphene which is a similar constant, not pure but enough to eliminate dispersion and supercondict. Anyways tldr but u might get approached by physicists cause even I do and I'm a puppet maker, no phd. And i want to use or try their code but indont wanna use julia ir fortran!! Or matlab. I'm trying to grt them to at try. Net , at leasr least port c and C ++ and java so incan use some.of their code maybe. Any sorry for the huge rant but I want someone to build a flying saucer without using a supercomputing cluster thx. I'm definite going to take a deeper look around ir code because of the topics that come up get more relevant. And ur voxel space thing.. sorry about the typos, I'm so tired from coding tools. if it doesn't make sense but u are curious I can send u a few links to the research

…

On Sat, Feb 18, 2023, 12:11 PM Ross Nordby ***@***.***> wrote: LinearAxisServo would work (the second connection can be a 'world' kinematic body), but a constraint may not actually be necessary here. Simply setting the position/velocity every timestep may be sufficient, depending on the requirements. Either way, be advised that this doesn't totally delete the third degree of freedom, and sufficient force could smash two objects into each other such that the shortest path to separation is along the restricted axis. In that case, the colliders won't separate in the 2D plane. To avoid this, you can make sure the objects have sufficient depth that the restricted axis is never a good choice, or detect such collisions based on the contact normal and handle them directly. — Reply to this email directly, view it on GitHub <#255 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AD74XGIHHVEIULEDJILERPLWYEGGDANCNFSM6AAAAAAVAKECEQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>