Download

Physics Based Tutorial Series

I’m making a series of tutorials on the creation of physics based assets.

I have a playlist dedicated to this:
https://www.youtube.com/playlist?lis…bPGPec1VWRWDSO

Here’s video example:

If you have ideas for what to do next, please leave them below.

To stay up to date with new videos on game development, subscribe to our Youtube channel.

Thanks for sharing! :slight_smile:

Thanks for watching!

There’s a new video out in this series. It shows how to make a trebuchet.

Motorcycle please!

That’s a very good idea. It will be challenging, but I think it’s doable.
I’ll see what I can do.

Nice tutorials! I have 2 suggestions, a ballista and a battering ram. I know they can be animated so we don’t need to use physics, but I think the battering ram for example would be quite nice workign with physics, against a door with physics too. Anyway, hope you continue to do content like this, good luck.

Thanks! I thought about making a ballista, I’ll put it on the list.
Ah, yeah, a battering ram should be a lot of fun, and actually fairly easy to implement. Breaking down a door with that should be satisfying.
I’m not planning to stop anytime soon, physics constraints are a ton of fun!

In the mean time, here’s the newest video in the series:

Hey, these are pretty cool! If you are looking for inspiration, my physics/mechanism itch is often scratched with this amazing dude’s videos: https://www.youtube.com/user/thang010146/videos

Thanks! That’s a very nice channel, with lots of ideas.

@Lusiogenic

Two general suggestions if you’re open to them:

  1. Take all your existing tutorials and get them working in Multiplayer: Replication / Physics-rewind.
  2. Make a few of these. Purely experimental, just a bit of fun learning or a break from weapons etc…
  1. To be honest, I don’t know much about multiplayer. I thought that since there are almost no blueprints involved, I just check a checkbox and it’s done, but I’m thinking it’s not that simple. Gaining the ability to rewind definitely makes this attractive. I’ll see what I can do. Suggestions start piling up. :slight_smile:

  2. You know, I was actually thinking of making one of those. It should be fun to make. I’ll put it on the list.

Thanks!

I’m still wondering if franktech was trolling you with 1, haha. Rama has done some work with multiplayer physics- I don’t have a link to his thready handy… also not sure if more recent changes made network-based physics more approachable and “healthy”. Historically though, networked physics is a tough cookie.

I whipped up a proof-of-concept rewind system a year or so ago, mainly just to show someone something visually versus what I was spitballing off my head. All I did was store transformations in an array, then walked backwards through that array for the rewind. Again, it was just PoC, and was dirty and inefficient, but it worked. Pretty sure others have utilized the same method for their experiments. A recent version (not sure if it was 4.19 or 4.20) introduced the replay system, which I’ve yet to tinker with, but that might expose an alternative way of handling rewinds!

[edit]

Manged to find an old video of that rewind system. It’s unlisted, so not sure if it’s accessible. Let me know if there are issues viewing it.

It’s possible :slight_smile:

I’m not sure it’s worth the effort of making a sepparate rewind system. It would basically come down to the same thing as the replicate function and live replay that’s already present.
I think it should work to simply check the “Replicate” checkboxes. I don’t have the time right now to look into it. It will have to wait for later.

Here’s the new video in the series. How to grab an object using a Physics Constraint as opposed to Physics Handle.
It has the advantage of not having to update the location each frame and an easier configuration.

@Derjyn

How could you think I’m even capable of such evil thoughts. :stuck_out_tongue:
I just don’t want to see @Lusiogenic getting bored, that’s all. :smiley:

True… Rama & TheJamsh etc are geniuses, but even their physics-rewind solutions contain limits / don’t cover every case… But going back to tutorials… Lots of Youtube series show how to create great landscapes or characters, but physics-rewind is also an art-form, yet its rarely covered. So before I was thinking: I’d love to see those physics examples carried over to Multiplayer as is, and just see what happens. Because there’s learning in non-successes too. But devs rarely go there! Most of the time tutorials are played safe: neat goal / neat solution. But there’s useful learning in seeing something that doesn’t work, understanding why, and appreciating the level of complexity that’s required to make things work as expected.

There’s a wider aspect here as well… The forums are littered with posters armed with a GDD and heaps of self-delusion… Thinking they’ll get a single-player game done with exotic physics… Then bolt-on multiplayer later (got that T-shirt myself). So it’d be nice to see a tutorial series that isn’t successful for a change, but offers useful insight into the limits of single-player vs multiplayer physics. Another example… This thread got little interest either… Yet its another area that could offer huge insights about what works and what doesn’t and why etc. Personally I hardly watch tutorials anymore. Too many are just too safe or a ‘redo’ of someone else’s pickup & inventory system etc. All I’m saying is, maybe branching out isn’t a bad thing…

Haha, totally agree franktech. Though, I was referring to Rama’s networked physics solution. I wasn’t aware he did a rewind system, and also not aware of TheJamsh’s solution. I’ll have to hunt those down!

It would certainly be nice to see the networked physics “cookie” cracked once-and-for-all, I think we are getting closer. Whether it’s Epic or one of us in the community, I have hope that it’s on the horizon. As far as rewind, I think it’s 100% possible right now. Just not a simple 2-click sort of thing. There are too many use-cases, and thus too many solutions for a one-size-fits all. But when it comes to transformations/physics, it’s doable. With a little love, it can even be quite robust and performance friendly. As I’ve shown, one could whip something up in a couple hours. And that was pure blueprint.

I also agree about the same tutorials being regurgitated. I don’t mind when a creator has a different solution to the same problem, but I am seeing more and more direct copies, which is a tad bit annoying. That being said- I think what Lusiogenic is doing is pretty original, and done well to boot. I foresee many people being referred to his tutorials down the road. At this point in the engine’s life, I think it’s smart and much more helpful to specialize in a niche, if one is going to do tutorials. It’s a bit late in the game to come to the table with tutorials that cover everything, without broad strokes and failing to really bring anything “meaty”.

Rama’s past demos were more about Custom-Physics-Syncing and replay systems than rewind, so sorry if I misspoke there. The syncing is also client authoritative, so maybe not the best for competitive multiplayer etc. On the other hand TheJamsh and others have done a lot of work on server-authoritative physics-rewind. However, to get there the CMC had to be reworked too, which is not the stuff of mere mortals. :stuck_out_tongue:

Hope so… I’d be great to have something plug & play. However, there are devs on here who feel everyone should just STFU and dive into source and design their own solutions. That’s kind of harsh. Not everyone is cut out to be a psuedo-engine developer. Plus time invested in re-inventing the wheel, is time lost on other areas of gameplay. Ultimately, a game (esp a 3D game), is a living breathing feature film. You get a great cinematographer and production designer and audio engineer compliments of Epic. But that still leaves heaps of other aesthetics to get right…

Definitely! That’s why I made the suggestions above, especially the first… In the spirit of asking for something you don’t expect to get, because you just never know … :stuck_out_tongue:

Well, I’ve tried my “brilliant” strategy of just flipping the replicate switch, and it kind of works. I’ve managed to make the Wheel with piston work for replay.
But there’s a problem with the FirstPersonCharacter camera. It doesn’t replicate as it should. It’s jerky at best.
Because I have the grab mechanics tied to the camera, it affects how I move things around.

Apparently, I’m not the first to run into this problem, this post suggest that I should update the position and rotation manually:
https://forums.unrealengine.com/development-discussion/c-gameplay-programming/25860-first-person-camera-movement-replication-problem?p=433221#post433221

I’ll try that and see how it goes.
Something to note here is that from what I can observe, the replication only applies when the objects are interacting with each other. If they are falling, for example, and they are not affected by other objects, they will just calculate the position on their client. This is good because it offers fluidity, but it’s bad because the objects might get out of sync.

I’ll have to test using 2 clients at the same time and see how it works.

Here’s the latest video in this series:

How to make a Wheel with Suspension for Vehicle