asdgg

xcvxcvv​

 

All it boils down to is optimization, all the way from art to code.
There are also many different tricks applied to make the world / level they design appear to be so lively and massively populated.
E.g, you don't really "see" millions or thousands of objects on the screen at once, they are only active when they are within the player's view.

You can read about optimization here.

 

They have higher budget, higher technical expertise, more engineers and programmers and also profile the heck out of their games.
https://docs.unity3d.com/Manual/Profiler.html

Learn profiling and algorithmic optimization.

A basic non-gamedev example optimization example.

1. You need to find one item within an array of a million elements. You do that checking one item at a time from the start to the end of the array.
2. It is slow.
3. You try to speed up a function you call to compare item in array with target value.
4. You get 1% speed increase.
5. It is still slow.
6. Then you learn binary search and implement it.
7. The search is now 50000 (fifty thousand) times faster.

It is a silly example, but that's the rough idea of it. Learn unity profiler, look for places that eat too much resources, and optimize them.

 

You might also wanna look into batching and instancing, which allows you to draw a large amount of meshes to screen efficiently.

 

When you see thousands of objects on the screen, the developer is probably using GPU instancing instead of using a separate GameObject for each item. In my 2017 game on Steam (Disputed Space), I have many thousands of lasers flying around in each scene. All of those lasers are drawn using GPU instancing with DrawMeshInstanced. I have one GameObject handling all of the lasers in the scene. Blizzard (and other AAA companies) are doing similar tricks like that to get the performance.

https://docs.unity3d.com/ScriptReference/Graphics.DrawMeshInstanced.html

 

There is also dots too which will optimize game alot

 

Its easy, just hire an army of experienced 3d designers. Btw not all of them are good, actualy it is more like only bliz and lol.

Also there is a big diff between a model with many poligons and a model with mind blowing beautiful hand drawn AAA texture.

 

For one, I doubt that they really have "millions and millions of GameObjects".

Some of this would come from understanding that GameObjects (or any kind of code "object") come with a bunch of overheads, and finding ways to represent things without making every single one an object. Unity's DOTS is basically a standardised approach to this, so you could look into that.

Some of it comes from knowing what computers can do cheaply and really capitalising on that. Having a thorough understanding of what's going on under the hood means that you can get the most out of whatever resources you're using.

Some of this would come from being very good at making it look like there's a lot going on when (from the computer's perspective) there isn't.

And some of it comes from just writing code and algorithms which are efficient.

As I'm sure you can imagine, that's a lot of stuff to learn. I strongly suspect that in the "AAA" teams you refer to they'd have experts in various different areas, rather than relying on one person knowing it all. If you have a solid grasp of the underlying principles you could do all of the above as an individual, but it gets much faster and easier with specifically relevant prior experience.

 

Yeah, dive into DOTS. It is an excellent way to handle lots of things in a scene in a very scalable way. Unity has come up with a standardized approach to use instead of building custom solutions using DrawMeshInstanced.

 

GPU instancing is done via script. With instancing, you don't instance objects. You use a command to instance meshes. Basically you send the GPU an array containing all of the instances instead of dealing with a bunch game objects. Instancing is a way to render a bunch of stuff without needing a game object for each of those things.

 

They also have very tried and tested pipelines that allow them to chug out much more much quicker. I know they have many LOD layers, which I know personally, I don’t take advantage of enough. Largely because my game doesn’t require them and because it’s a pain to bake. Where big studios probably have an automated process that takes in a high poly model and splits out 7-10 LOD models/textures.

 

Yeah, DOTS is really exciting. Technically speaking, all of that can be achieved manually through DrawMeshInstanced, but it does absolutely rock that Unity has a standardized way to do this now though DOTS. If you have a game idea that needs lots of units moving around, DOTS is an excellent solution.