I am building a ‘city’ in Maya. My idea is a prototype with a ‘player’ in the scene and some NPCs (this will depend on ‘n’ factors which I still decide, including performance). Making a comparison, in a square kilometer (real world) transposed to the virtual environment, to a desktop (also thought of in general here - good processing and video hardware) - without taking into account the AI and post-processing effects, which is it possible to do it in terms of polygons or what would be a good measure for ‘buildings’ or ‘blocks’ in this space (1km2) to weigh on performance? (Of course, here everything is in general terms, but I really need to have some idea).
GPU: Generally you will want to look at preventing triangles getting too small (micro triangles - see PSA: A general rant on model LODs. ), rather than focusing on how many triangles there are. This is the main factor for GPU performance.
CPU: Generally you want to merge as many things in the distance as you can so the CPU submits less batches. Hierarchical Level of detail is best.
There aren’t any fixed numbers for Maya because the data has to be transformed for acceptable performance. Any problem is well understood and solvable in this space but knowing exactly what you want for specific hardware is important because the answer does change depending on that.
So perhaps the best way to get started is to make a prototype in Maya that contains similar levels of detail and then test it first.
In the Unity documentation there is something approximate about character, it’s the same question, only directed to static objects. I don’t think about going to an engineering course and then knowing something about it, there must be something simpler than this answer that just confuses more.
Somebody could come here and really sound like they know what they are talking about and tell you something totally wrong. And then you believe it. That is setting up for failure.
Build a quick version of your city using same amount of resources that you’ll require in the real version, then test on your target hardware.
You don’t need to be an engineer but you do have to follow basic tents of scientific method. Confirm what you think you know, test to figure out what you don’t know. If you can’t produce predictable results, you don’t know.
I ll try be more clear: thing about a BOX, how many times (or how many poligons) i can divide this BOX in standart view and play (with standart config, setup) to start have problems (minimal requeriment to use Unity)?
I try never to write on this forum because it’s too toxic, people ask simple questions and some just theorize to touch on any objective answers - why do they do that? I don’t know, mere exhibitionism maybe. But, the question is simple and obviously because I don’t know about…
Thank you, it is already or would be a start, although you are giving an unreal number, not even supported by experience (science is based on that). It’s different, but I won’t debate with those who are ill-willed. The question that started from a hypothesis (every argument that claims to be scientific starts from the hypothesis) is simple, you don’t know how to answer it, simple like that - even give something approximate like, for example: ‘there are moments, in games A, B or C that use about X number in a given ‘scene’’.
Yes, that’s what I was looking for, I have a few more questions, in case you know of course. Polygon face is, at the time of triangulation, divided by two, correct?
My prototype seeks to avoid exaggerated numbers, I will use compositional techniques, empirical (since I don’t have the technical knowledge to do so) if not by experience even (and rare) relatively simple, such as texture and lighting (and this will be another subject that I will think about later, although I’m reading a few things about). In other words, it will be more ‘conceptual’ the idea of the prototype to what I seek with the story, script, involved.
We have games, not ‘universal answer’s’ - we can use this games like exemple or ‘to start a idea’. Ex. ‘on cene of cap 3 from the game Y, the game render 1 trillion poligons at 800 bilions of frames por secound’ (get easy).
It’s like you asked how many tires a vehicle should have. It is a useless, lazy, naive question.
Now people are treating you like a child, giving a sucker so you be quiet and go away.
Instead of throwing a fit you could have read the first answer carefully and realized the intelligent response would be to describe in detail your target platform.
(my english kill me) - i really need a ‘base’ to have a idea. I ll work with some alpha and planes to d some details… (my idea is work with a lot of alpha planes)
It’s simpler. You don’t know how to answer simple questions, you don’t even have any didactics or interest in answering something on this forum, you just want to theorize about how people should or shouldn’t formulate ideas - in other words, censor what you understand as the wrong way to be ask something’. I told you, I have no business with these toxic types.
Im here since 2010, i never have insues here, ll NOT have now… get off.
Since 2010 and if you are still asking the most frequently asked beginner question it is a shame nobody cared enough to teach you anything. Maybe it’s because you don’t want to listen.
You can have one unsubdivided box with the right shader you can bring framerate to single digits. Just rendering this one box with its 24 triangles.
You have multiple resources.
CPU time, GPU time, fillrate, texture read rate.
When you drain any one of them, fps goes down.
Because of that it is not possible to answer “how many times can I subdivide”. As you can have different framerate with the same number of polygons, depending on what you did.
The geometry density is fine, but you can still murder your framerate by adding too many dynamic lights or complex shaders or too many objects.
Yes, of course, thanks. But we can have an idea with games or experiences with games, so to speak, consecrated (published). I don’t think it’s necessary to understand all aspects in depth to do something, understanding that it’s a prototype for learning purposes anyway (that’s why I avoid posting on this forum). I don’t intend to revolutionize game software engineering, but to have some general or general ideas about, for example, what is explained in Unity’s own documentation about ‘character’s’ (they mention approximate numbers of polygons there). It’s the same question here, only about static objects.
That’s not “in depth” understanding, though. It is shallow level basics.
To answer your question it is necessary to know details.
Untextured geometry with a single light? Should be fine, unless you’re procedurally generating it every frame or something.
Ton of dynamic lights? That could end up being slow
Complex shader? This could be slow.
Ultra high density textures? Could be slow.
Unique high res material on every face? It WILL be slow.
There’s more than one point where you can cause the performance to drop. That’s why you need to give extra information, and that’s why people can’t give you concrete answer.
If you make something like “Another world” the engine will definitely run circles around it.
But you ABSOLUTELY can get poor performance on low polygonal scene based on what you’re doing with it. (There’s postprocessing as well).
(get easy).