how do game developers create games with graphics so intense that not even the latest tech can achieve those graphics? (Like even the rtx 3090 struggling with CP2077)
In: Technology
Simply put, it’s easy to make a difficult mathematical problem, and difficult to actually solve it.
This is because all computer graphics are basically mathematical calculations done really quickly, to determine what goes onto your screen at what time. By adding more things to draw and making them more difficult to draw, you increase the number of calculations that your computer has to do. Eventually, it cannot complete all those calculations in time to draw 60 frames per second, and that becomes apparent as lag.
Think of it this way. Your professor has months and months to prepare your final exam, but you only have three hours to solve it. Game devs are like your professor. They can take as long as they want to come up with as complex a graphical setting as they intend. Your computer, on the other hand, is like you sitting for the exam. It only has a set amount of time to complete the calculations.
If you had all the time in the world, you conceivably could prepare for and write any exam, no matter how difficult. That’s the idea behind pre-rendered graphics. Movies, for example, run at 24 frames per second, but all the CGI is rendered in advance. And those graphics can be really complicated, but it’s not a problem for them, because they’re not rendering in real-time.
The graphics quality is easily paramatized. Things like screen resolution, texture resolution, the amount of detail in distant objects, the amount of light rays that bounce around a scene etc are all configured using a few numbers.
Part of it is as a sales gimmick – this game is so awesome it won’t run at max on anything available !!!111!!1!11
Part of it is to keep the game exciting longer, the game will get a spike of interest once something comes out that can run it at max.
It can also be free advertising, if the game is intense enough it will be used as a benchmark for hardware reviewers.
It will also generate sales to hardware junkies that wouldn’t usually be interested in that game but will purchase it to brag about how it runs on their supermaxx overclocked water-cooled computer.
It’s easy really, all things considered. Say, raytracing the algorithm was first described as a method for drawing computer game graphics back in the 80’s or so. The problem was, for a long time, there was no hardware capable of using that. After a decade or so it was first used in animation movies(think toy story) where each “frame”, raytraced, took hours to complete.
Often the developers start out by making something that no machine is capable of running and then the real work is in trimming it down enough that computers that people have can run it. The real magic is in making the game actually run at good fps.
Pretend I am the game, and pretend you are the GPU*. My job as the game is to describe to you how I want you to draw a picture.
When the game’s graphics settings are “low”, I show you a map of a 3D world, and I tell you where the camera is and where it’s pointing, and I tell you there are 40 trees and rocks in the world, and where they are, and I tell you each tree has 10 branches and 50 leaves on it, and I tell you where those are, and I give you three photos of tree trunks and 20 photos of leaves to help you draw them. You draw all of this in 1/30th of a second. Since you can draw that quickly, you can do it 30 times in 1 second, so your game can run at 30 frames per second (fps).
When the game’s graphics settings are “high”, I do all that, but I tell you about a lot of little indentations and flaws on every tree and rock you have to draw, and I tell you each tree has 500 more branches and where to draw each one of them, and 1000 more leaves, and I give you 300 more photos of leaves and tell you where to draw them. I also tell you I want you to do math to figure out how much the sun is lighting up every part of every tree and rock, and then to figure out how much the light reflected from those parts lights up other parts, and draw all the light correctly; and I tell you there are 50,000 dust motes in the air, but instead of telling you where they are, I tell you where they were a long time ago and I give you math formulas to figure out where they are now. And I tell you that you have to do math to figure out where all the shadows are from the sun, but also the shadows from reflected light from other trees and rocks, and I make you draw it all. This takes a lot longer to figure out and draw. If it takes you 1/4th of a second to draw all that, you can draw at 4 fps, and you say “screw this” and switch to low graphics like an FPS gamer does.
However great graphics cards become, it will always be possible for games to bring them to their knees by demanding more and more detail and visual effects in every frame.
*Actually you are a combination of DirectX, your card’s graphics drivers, and the GPU, but I wanted to be simpler.
u can edit something without rendering completly. Wenn i made fur in Blender i said there should be like 1000000 Hairs on the final render but just show me 10% of all Hairs. Same goes with particles or polygons. also u can edit the light bounces. When u edit it u led the light only bounce a few times and when u want the final edit u let it bounce more often. hope i could help and have a nice day
They could create graphics that take 50 hours to render each frame of they wanted to. In many ways it’s more difficult to do the opposite, having a game that runs well on a wide range of hardware.
Generally speaking, targeting future hardware isn’t too tricky as the progress of hardware is relatively predictable, at least in terms of trends. Offline renderers – the sort they use to make animated films and such – are fully of features and settings that aren’t even close to being possible in real time yet.
What makes it into games – even things like Raytracing – tend to be hacked about versions of what’s in offline renderers, designed to run in real time. The better the hardware, the closer to the “real thing” they’re able to get, but real-time hardware is so far away from being able to render these things in real time that it’s not like they’re running out of stuff to do.
Developers often have hardware above and beyond what you and I can find. They can also just develop something that runs say 15 FPS on the absolute top end of consumer hardware available today, and as long as they can prove to themselves that it’s purely a hardware bottleneck and not a glitch, well better hardware will come out eventually that’ll run it faster.
Just like in human relations, giving instructions for a very taxing task is much easier than following them. You could easily open a 3d modeling and rendering software and create a scene that would take even the most robust consumer -available hardware many hours to render.
Well the easy thing is to create good grapics that don’t need to run fast.
In the case you can simulate every single particle of light that would hit the screen, using the same physics that you observe in the real world.
The hard thing is to cheat well enough that the audience doesn’t see where you cut corners in that simulation, to enable your pace to run it real time.
And ofc. if there i a better PC game devs, can easily make use of that power to create more realistic graphics.
Latest Answers