Well you're stupidly good looking. And a genius by the looks of things. You won the jackpot by all standards.

Ok, made in Unreal I hate Unity...(the interface is total from '90s).

This is great reminds me of Brackeys who I miss.

@@davedogge2280 Thank you, that's a wonderful compliment as I also miss Brackeys! His videos helped me so much throughout my university course

@@themarlboromandalorian There's a time and a place for every statement. You succeeded at neither the time, nor the place. 👎

It's not performance related, but using public is bad practice in 90% of cases. Using serializedField + Setters and Getters is a much better approach for clean code and decoupling

Or use Dependency Injection and get rid of the 90% of the inspector setups

I had no idea I was using dependency injection until you mentioned this and it triggered my curiosity, so I googled it to see what it was. I am someone who tries to think before coding to see if there is any other way I could do something to minimize code and improve readability. Crazy how the mind comes from alone to these solutions and you don't even know there is always an existing term for that.

@@FlotzOnYou why is using public bad?

@@chrono9503 because it's not the right encapsulation most of the time. Public means the property is exposed to reading and writing by a foreign class, which you don't want most of the time

Thanks I'm glad to hear it. Yes, it was quite a shock! Had to just remove that plant in the end 😅

When you see the kind of optimizations and efficient modelling methods used in AAA games, you realize how much devs strive to minimizes polys and stuff. I've seen wireframe models of things that shocked me, because it did NOT look like it was that low poly in-game!

Just from personal modeling experience and 2 years of a grade school 3d class, I too was blasted back by the polys of the plant 😂

@@MFKitten you'd be amazed at what a good texture/shader combo can do.

@@yagmanx I know I'm super late here but on the subject of that plant (and other models with similar poly count issues): On your decimate modifier at 7:19 you could select the "Planar" option and set its Delimit mode to "UV". This would decimate the model whilst performing calculations favoring the original UV layout more than other options would. Obviously not as good as making a low-poly version yourself, but it's a closer result that can work just fine with many basic props.

Very powerful!

Thank you

Glad it was helpful!

if you write a bit of code you can actually add sorting orders to 3d meshes as well which is incredibly useful and I never understood why it wasn't buitlin.

This is a very good point! Thanks for pointing it out :)

would still be more performant to just use a box collider or set of box colliders, the code that has to run for mesh collision is just more advanced than checking if a point is within the bounds of a cube.

@@bradjones7491 Any documentation on the algos used for that calculation? I'm certainly curious what the Big-O is (you are certainly right, Im just curious to what extent the difference in performance is)

@@0xlapras646 I mean you would just have to look up how to do it from scratch, but checking boundaries of a cube is mathematically significantly simpler than check the bounds of a complex mesh. for example you could easily determine if a point falls within a volume of 1, but a mesh collider has a non standard volume so you would first have to determine the volume before checking the bounds. Mesh colliders are used extremely sparingly in general, and mostly just for simple shapes like ramps and sometimes terrain. As for the extent of the difference I couldn't say for sure but I'd assume it would be fairly significant since almost every AAA game in existence utilizes the multiple simple colliders method. Mesh colliders should only really be used when the alternative is prohibitive, such as with a massive object like a terrain, where aligning individual colliders would take thousands of hours. Basically mesh colliders are time savers, but are overutilized by people who get lazy, and by doing just a bit more manual setup you can often avoid the additional overhead. Another thing to consider is that often times mesh colliders fail at detections where a box collider would not. While it's pretty unordinary and example of this can be achieved by moving an object at incredible speeds, if the wall is a mesh collider it has a significantly higher rate to allow the object to pass through than a cube. and This is because mesh colliders don't really have thickness so if the physics calculation somehow places that object on the other side then the collision will not be detected, whereas with a cube the object would need to reach significantly higher speeds to breach it's entire bounds, and you could make a wall have theoretically infinite thickness to make it basically impossible to clip into. As a side note about the fail rate of physic detection this mostly occurs because physics calculations are done on an independent update loop that has it's own time scale so a significantly large movement in a single update of that tick function could offset the object further than the collision is detecting thereby avoiding the collision detection entirely. This is because most physics systems don't calculate along the traveled vector when determining collision, rather they only check for the position at that frame of the tick update. That is to say, that if an object moves farther than the entire collision's size in less than 1 physics update tick, then that collision will not be detected. That also leads into another slight tangent about unity in particular, where people think that unity's physics engine is very inaccurate, but this is actually mostly caused by people never changing their physic's tick rate under the time settings in unity. By default I believe it is set to a tick rate of 0.10 which is 10 frames per second and is pretty inaccurate, allowing for most "fast" speeds to easily clip through collisions.

Thank you for the advice!

Do you think such architecture can be relevant in this kind of game? I haven't deep dive into the topic yet, but to me it seems like only games with a massive amount of objects can really benefit from it (or in competitive multiplayers where performance is everything, but it sounds like hell to design and maintain)

Haha glad to hear it lovely 💖

This is really interesting to know. I'm so glad you're in a better place now, it can take a long time to know what really makes us happy. Best of luck for your future :)

you should try unity bolt

I'm so glad to hear that these tips could help you! Best of luck for your tool :)

Thank you and good luck with any future games!

@@yagmanx thank you for the reply and the kickass video!

I suggest doing probuilder - export and export all of those meshes to a fbx or obj file. Afaik probuilder for some stupid reason rebuilds your meshes at runtime every single time the game (scene) is loaded.

@@janda2304 Would be good if more game dev tutorials would warn you about it lol. In my first few days of Unity I was told to use probuilder and only 2 years later am I finding out it's a bad idea

Collision detection with triggers would be better, due to unity using an octree internally to do the same thing a lot more optinized

Take it one step at a time. It sounds like you know roughly what you want to do. Maybe just find a weekend to play around in unreal / unity and see how you find the game engine? Try and stick to small, manageable projects or even follow a tutorial and then mix it up to add your own style. Have fun! Wish you the best :)

@@yagmanx I’m unsure if those programs do it but I think if I ever do make a game it’d be pixelated. If you’ve ever heard of Octopath Traveller, that’s the closest thing I’ve seen to the art style I would like. Not exact, but definitely close to it. Not sure if those engines offer that kinda stuff. Either way it’s like 5 years away lol. Thank u stranger for your kind words

So glad this could be useful for you

...and use one material.

aren't tasks internally ran kind of like coroutines? It might be easier to optimise code with tasks but it's not directly faster right?

@@Lordilucas12 Very similar in terms of operation, but less garbage collection with Tasks (and less garbage collection ultimately means faster), and they have more options for returns / awaits. They are a little trickier to use, but they are also a little cleaner to use once you have it setup. It really depends on the use case you are after.

Coroutines CAN have return values, you just have to write your own StartCoroutine alternative which intercepts the return ... Here's a pseudo-code example of how it works: if ( coroutine.MoveNext() && coroutine?.Current is T returnValue ) { return returnValue; }

Only with an object that is using a rigidbody. Honestly, Unity can handle tons of objects with Mesh Colliders with no performance issues. I mean they do have ProBuilder integrated as their default object building program these days and every ProBuilder object you create has a Mesh Collider on it be default. I have a game that has 10's of thousands of Mesh Colliders and I get well over 60fps. It isn't something to worry about. If you're using detailed imported models then no, you don't really want to use Mesh Colliders on them.

I'm very happy to hear this and totally agree! I use Unreal too and learning all of these techniques myself has definitely helped keep my Unreal projects performant