I’m new and don’t understand how to create such a sphere, I’ve already tried adding a collision separately, creating an object using an actor, and much more.
I’ve been struggling with this issue for a few days now.
Thank you all.
INFO: A sphere that doesn’t have any collisibles inside, but has them at the edges so that the player is inside the sphere.
So then it isn’t a sphere, its a hull in the shape of a sphere?
You can’t do that using engine stuff.
A sphere is full.
You would need to model a hollow, and then designate rather complex convex collision hulls split in at a minimum quarters to get something going.
The final output is probably going to be quite horrible in terms of collision - unless its a dyson sphere which being gigantic ends up split in mostly flat individual portions…
1 Like
Use modeling tool and make two spheres, use Boolean tool to hollow out one sphere with another. Use plane cut tool to make opening (or even another Boolean with a different custom mesh). In collision settings for that hollowed sphere mesh - set use complex as simple for collision.
1 Like
You can’t use complex as simple. Obviously…
Whats with this fourms and giving out horrible advice?
And that also extends to the idea of using the in-engine geometry thing.
Use Blender instead. Same ideas. 100% more reliable.
1 Like
Modeling tools generate non-engine versions of a mesh.
You can use complex collision as simple for static objects.
1 Like
You should never use complex as simple.
Literally Never.
Its basically only there for the other things you do with the engine that arent games - archviz, movies etc.
Modeling tools in engine will also generate non engine objects.
The point of the engine sphere is that it is a preset avaliable to all. And that’s the only point to it.
Once you alter it, guess what. It won’t be avaliable to all. It’s a custom object in your game folder.
If you alter something in engine you also have to go through the process of figuring out the UV maps.
That process is going to be a ton easier - without even knowing what uv map are like the avarage user - in a full DCC.
Not that the engine modeling tools aren’t “mature” but they are what they are - trash when compared to a full DCC of any kind…
Is it worth trying?
Probably not.
Takes you about 20minute to cook it up in blender after looking up a tutorial and it gives you all the elements you need - collision included…
The fact you need 20 more minutes to import it to the engine because the epic blockheads broke the fbx import system is a different story;
While its how it is at the moment, it should get fixed eventually.
While its not. Turn off the new stuff, which is good advice for anyone, noob or experienced alike.
1 Like
If you create (there is a tool for that too) sphere in modeling tool it is not going to alter engine version. If you create (drop) engine version of sphere and model it - yes, that will change the engine version mesh. Modeling tool also has a UV adjusting thing. Anyway - making custom meshes with Unreal modeling tools can be as good as Blender if you know what you’re doing. In Blender I can make a cube ! Also, good luck importing it to Unreal 
.
1 Like
Sure, but if you want to work in games/industry, don’t you think you should learn how to:
A) Do things properly
B) Use various DCCs?
And
C) import stuff into the engine?
And for the purposes of the forum - it’s probably good to have both options for someone to read/evaluate, I’d agree.
The part that irks me is the complex as simple becuase its going to actually cause people to wreck projects down the line…
1 Like
That depends on what kind of project it is and how good the programmer is in optimizing the project. If it is a single player game with few ai running around you are safe to use complex collision. No idea what DCC is.
1 Like
It isn’t perfectly safe. You need to go an learn some more it seems.
Complex as simple is - again - only for non-essential non phisics collisions.
In other words, when you use complex as simple you allow the player to easily pass through objects or in words easier to grasp “cheat”.
That - amongst abother billion reasons - is why you don’t use that option in videogames ever unless you know precisely what you are doing.
1 Like
If you enable CCD on a mesh it will not easily go trough walls.
1 Like
With complex as simple it still can. Doesn’t much matter what settings you have on the physics engine. You could even be calculating physics fully every millisecond (which is obviously impossible) and you would still get issues.
Regadless of Chaos or PhysX btw.
The issue is that the collision is not a primitive or a Convex Hull.
So no. All that enabling CCD is going to do for the problem is the equivalent of diddly squat…
1 Like
Yes, but it looks cool when objects collide exactly as the mesh. And that’s the most important thing.
1 Like
Its not.
And it won’t - which is more to the point, because who really care about what it looks like when it can completely fail to function?
There will definitely be cases where the whole hull is not even collided with depending on speed of the hitting object, the thicnkess of the 2, the level of ccd, etc.
On a primitive that’s near impossible. And yes it could happen on convex hulls to but far less likely that with complex as simple (if not outright impossible because it is handled entierly different more like primitives).
Also, its probably worth noting that the engine automated hull generation can do a decent job at giving you complex hulls that match your geometry.
Doesn’t beat manual editing, but it is fairly good (and been a part of the engine for some 20 years i think).
1 Like
Well, it`s not impossible to make it work in a contained area. Adjustments to the physics setting, so on… Good programmer finds a way.
1 Like
You don’t understand - maybe you have never worked with it.
There is nothing predictible about a physics engine. Ever.
Its got literally nothing to do with being good (look at the invidia guys) or being bad (look at epic with Chaos). You inevitably will have issues due to the sheer nature of physics (what we understand of them, and the complexity of the calculations).
The only reason its Ok on cinematics is that you can do takes until you finally manage to get the effect you want. And then you save it and re-use it.
Similar to archviz. You wont be throwing projectiles at it, etc.
The moment you need it to function (say its a big rock in your landscape) then you cannot rely on complex as simple.
That’s all there is to it…
1 Like
I`m working on that right now. Using complex collision. Moving objects around. It works. Even on a ryzen 5 with 4GB radeon rx470. Sure, you need to adjust things and make sure you play like a beta tester. So far - only once have I lost a game progression object to physics.
1 Like
That “only once” is already enough to mean 9billion times through the lifetime of the game.
And when you change hardware to something less beefy that “only once” is likely going to become only 10 times in 10 minutes of gameplay.
The difference that convex hulls make on that is lukely to be 0 times and regadless of hardware.
Imagine moving to something like a mobile game with near 0 budget for physic calcs…
1 Like
I said ryzen 5 with 4GB rx470. Anything less beefy than that is a potato.
1 Like
You mean like a Console or a Switch?
Which is what people generally tend to make video-games for?
Also the GFX in use has little to do with the physics, just to make that clear.
Study up on what goes over CPU and what is over GFX.
Also what goes over GFX but gets put back onto the CPU while at it…
1 Like