I’m still new with BGE, I wanted to have tips about making a RPG, I’ll be inspired by Final Fantasy 7-8-9.
A world with precalculated background and almost no fights. (Except maybe boss and some occasional fight)
An external world with enemies we encounter randomly.
I’ll take my time, it’s not my job, I do 3D for fun.
How many polygons for characters?
How many polygons for enemies?
Can we switch from scene A to scene B and vice-versa indefinitely? (to have something clear, I’ll create a scene for each locations)
Is it feasible?
I’m alone doing this, maybe it will be a very short game (like a 10 minute game) because it seems like a huge amount of work.
Nobody can answer your questions about polycounts. It will depend on many factors, none of which are obvious before you have a prototype of some sort, you know your target hardware, art direction, have an idea of how many characters appear on screen, etc.
Also don’t use BGE. It’s just never been a very good engine. Pick a different one, something easy to work with, like UE4 or Godot. Heck, even Armory which is in early alpha is probably a better choice than the BGE at this point.
Making a short game by yourself is feasible, but will still take you a long time. Making a 50 hour epic is not.
for scene switching, from what limited exposure i have, seems very stable. i personally use blend file switching. might be a bit slower, but its more flexible.
with use of LOD, you can have very high poly characters. the player shouldnt need any LOD, and by the sound of fixed camera angles, LOD becomes even less important since there wont be much depth.
i would say the bge is more than capable of handling a project like this. bge is most limited by dynamic lighting, which a static rendered scene will have very little of.
I use Blender because i already created a game with it and it’s intuitive. I know nothing about coding (I used codes founded here for a small foosball table game, a script for the text resolution, one for a video, etc).
Unreal Engine is famous, used on pc, xbox one and ps4, but can I really create a game without coding knowledge? Is it 100% free?
I could learn a bit but I don’t want something too complicated.
I’m working on it, I have a character that I need to texture, and 2 scenes I need to finish.
I was thinking about planes for the backgrounds/foreground, so there’s almost no polygons. And to be able to move the character inside the scene I would have add invisible boxes (and a plane for the ground) with collisions.
I would have use lighting just to add shadows of the character.
I’ll maybe use Unreal Engine if it’s not too complicated.
UE4 is not free. It costs a percentage of your gross revenue beyond a certain sum. But you don’t have to pay anything until you find some success at least.
And no, I doubt anyone can make a game without at least basic programming knowledge, but UE4 does include an excellent node-based visual scripting environment.
I would make your characters as low poly as you can. consider that Blender struggles with MakeHuman chars.
in my tests on a decent gaming rig, a single lowest poly MH char along with a handful of mo cap animations dropped the FPS to 30. Thats with nothing but a plane for the char to stand on.
If you’re making a short game just for yourself, yeah, you can get away with Blender. If you’re making a larger game to sell maybe something else will be better
With UE4 you can buy an RPG starter kit not sure if it’s still supported though… And you don’t need to code, UE4 Blueprints (visual scripting included with UE4) are like logic bricks on steroids.
Epic games says you can make a full game with Blueprints. There are tons of youTube videos.
Spend time learning how to do everything without writing any code, just so you don’t have to spend all that time learning how to write code. Well, IMO it’s best to pick up programming, but to each their own.
Regarding polycount, a good rule of thumb is below 10~ thousand tris for a detailed character. That’s what a friend of mine does and his graphics card is ten years old, the dude is a freak. But if you don’t have too much going on in your scene, that’s more or less a good number. If your target audience is more the likes of which upgrades their rigs every once in a while then you can afford a higher polycount, obviously. Texture size and having too many materials can also fuck you up so keep that in mind.
That’s what I’ll do, I’ll stick with Blender. Anyway there are still plenty of things I need to learn with BGE. But there are things I can already do.
Normally I’ll buy a new computer soon.
I’ll try to do something good, I even want something better than actual generation.
Actually my human model have 24.000 triangles.
The problem about polygons could occure during fighting scenes, but I have to create those scenes.
I’ll do a circle with a half sphere sky, add grass (plane with grass), and trees (certainly low poly).
It’s a problem I (normally) won’t have inside the cities because it would be background with only humans/animals in real time 3D.
Thanks for the message
I’ll try to convince a friend working on this (he doesn’t use Blender but he is infographist, he has the diploma)
that depends. you can test how many polygons your computer can handle at the same time, that’s the ammount of polys you can show on the screen in every scene. the polycount of your character, scenario, particles, menus, etc, IN TOTAL should not exceed this limit. a map can have more polys if they are hidden, something many games do is use walls to hide geometry, like mountains or buildings.
it also depends on the camera view and distance. a game that’s isometric will have lower character polycount than one that’s first person, a third person one would be something inbeetwen. you can also look at games to see how many polys they use.
in bge, one thing that could slow down performance is the ammount of bones and polys of characters, as they will be moving. scenarios tend to have higher polycounts since they are static and use less proccessing power.
you have to find a balance beetwen the two, or go for a low enough polycount as to get a good framerate.
if the character is one and there are many enemies, enemies will tend to have lower polycount, bigger enemies may have more polys, while smaller ones will have more.
again, you have to find a balance of polycount for the computer you are going to run it in.
don’t use scenes as levels. in bge you use libLoad. you load a map and when you don’t need it anymore you free it and load a different one. that way you keep levels in different files and save a lot of memory, as well as speed up loading of the game.
everything is posible. but games take time and effort to be made. the bigger the game, the more time and effort you need. even a simple game needs a lot of work, but RPGs also need a lot of coding as well, you need at least one programmer (an actual programmer, someone who knows python and can load files with open for dialogs and use dictionaries for inventory and items as well as spawning and all the rpg code), you can’t make an rpg by yourself if you only know 3d, it’s one of the most complex, difficult genres to make and takes a lot of coding.
that’s a very ignorant thing to say. makehuman models are not game models, they are not optimized for realtime, they have useless polys and high polycounts. you will get the same problems if you use them in unreal, unity, or any engine. a game model, even a high poly one, is OPTIMIZED for performance, with only the essential faces that will be seen modeled, with planes with texture for hair and even teeth, and they don’t have things like modeled toes under the boots.
do not use makehuman models for games, mh models are intended for renders, not games.
Actually, I have some low and medium poly MH models working just fine in blender, along with Mo cap animations, edited in BVH hacker for a Daz rig. So it can be done. Getting ready to export to UE4 for testing. Low poly is the ticket. Higher poly drops the FPS.
i didn’t say it could not be done, it SHOULD NOT be done. useless polys add up to the total of redered polys and gpu/memory/bandwidth transfer and storage, it limits what can be done in a game. a poly that is used for an eye could been used for a detail of the silouette, hair, or not used at all to increase fps. a single model with all these flaws could work fine, the problem is when you add multiple copies of it at the same time, for enemies for example. the game starts to run really badly.
look up game models, look at the topology, then compare it with makehuman models. MH models are not game models, they are made for render, they have things like individually modeled hairs, modeled teeth, modeled irises and nails. most of wich are just planes with texture or taken care of by the normal map in game models.
Yeah, you did. And you’re wrong. Many people use them for games.
If you choose the high poly models, then yes you are right. But, You can choose the lowest poly model MH offers.
Low poly eyes, no mouth or teeth or even hair. And then, if you don’t want toes and fingers, delete them in blender. Add your shoes, or mittens, whatever.
Delete faces under the clothes. You can alter them in blender.
My char is working fine in blender, and I have no doubt it will work in UE4.
(If I can ever get the time to test it, that is)
I choose to sacrifice some things, for a decent looking char, or char side kick.
Some people are OK with 19 Always sensors with True pulse, I want a decent char with decent animations. My choice. Of course enemies are usually small, (far away) so I wouldn’t use an MH char for that.
But my advice to the OP was meant to compare a high poly MH model, with a low poly char, any char, not just MH, to see how badly it would run in blender, (or other engine).
(though I didn’t say so directly, my bad)
So I stand by advice, Lowest poly you can stand.
If that makes me ignorant, then I’ll own it. But it seems that I’m in good company.
LOL (Had to throw that last part in for drama. LOL)
In this game I made for my granddaughter “Tooncity”
there are only 4 armatures.
As the char spawn they are given
a property that decides what cloths they wear, hats, glasses etc.
There are no MH char in this game.
I made the characters, they are about 3254 tris give or take.
as the game progresses, more and more people spawn and walk around.
I had up to 70 people all walking around.
The game runs between 50 and 60 FPS.
So there is a benifit to keeping your char low poly.
In this game “Dulce” The zombie thing is a MH char.
26,756 tris. Using the Carnege Mellon Zombie walk
Mo Cap animation on a Daz rig.
It has a low poly mesh (proxy mesh) and a higher poly mesh.
I used an LOD on the high poly.
And as you can see, the game runs at 60 FPS.
The red haired girl on the right (above) is a MH char, about 27,890 tris.
She has several Carnegie Mellon mo cap animations
And as you can see, the game runs at 60 FPS.
The next image (below) is a MH char I’m working on 47,512 tris
I need to delete the faces under the clothes and hair.
She has a carnegie Mellon MoCap Salsa dance animation.
The Blend runs at 60 FPS.
that’s an average low to medium poly model. for what you show it could’ve been 2000 tris per character, and LOD could be as low as 600 tris. but that’s not the important part, look at your scenery, it is below 100 tris with simple textures. there is a ridiculous difference beetwen your characters and the world they’re in, making the world better would have no effect on performance and look better. making the characters even lower poly would make it look exactly the same, but have better performance.
that’s a ridiculously high polycount, it’s something a AAA game would have. it may run smooth because it is a single one, but adding more would make it slower. the world is again VERY low poly, i can count the tris by looking at it: 28 faces/56 tris.
you could make the game look much better if the world was more detailed, add corners to the walls, actual doors, different floors, decorations, light sources, decals. it looks like a 90s game, a 90s game that needs a 2018 computer to run it.
the red haired girl doesn’t look 27000 tris, it looks 3000ish tris. this is part of the “wasted polys” i talked about, the polys that can’t be seen because they are smaller than a pixel, are wasted polys. the world is again very low poly (i count 12 tris), there is no dynamic lighting nor shadows, and only two characters are seen on screen at the same time.
the only reason why your examples run well is because you are using geometry that belongs to the map, for the characters.
but this is not a competition. @CurtYoung asked for how many polygons for characters and enemies. mh models do not help him. he needs to show more than just 2 characters on screen, and have a more detailed world than just 4 walls and a uni-floor. your tooncity example may be the best for this, and you said it doesn’t have mh characters.
yes, testing is important, but you also need to learn modeling and get better at it. you can lookup models from different generations and find one with a polycount that works for the game you are making, on the computer you are using. you don’t learn that by “just” testing.
still not good. a lot of the topology of a mh model uses perfect loops and circles for easy editing. this is not the best for a game model, sometimes a bunch of faces can be replaced by a single tri, eyes can be just a buch of faces with an eye texture, some arms or legs could have less subdivisions, fingers could be simpler. you can put more detail into a part of the body like the head and take it away from other. you can’t do that with a mh model.
it depends on what model you need, a distant model can be less than 1000 polys, for an isometric game. a third person model could need some of the detail of a mh char, and you can try to adapt a mh model for game, but you have to remove a lot of geometry, not just the toes.
as you can see, it is 6800 tris or around 4000 faces, details included. it uses a normal map, but even without it it would still look good. this is what i talk about when i say “game model”.
i get it, making models is hard, making maps takes time and effort, and using a premade mh model may seem an easy solution, but it is not the BEST solution.
what i recommend is, if you want to do something simple, keep it simple. make low poly models and a low poly world, do what indie devs do and keep it stylized. if you have time, put a little more work on it, and make higher poly models, and also a more detailed world. but keep it consistent.
if you make your own models, it doesn’t matter if they are not perfect, they are YOUR models, it is YOUR game, not just a bunch of assets you put together, and you will also get better at it. if you don’t want to model, you can ask someone to do it, or download some lowpoly mesh and reskin it at worst. but you have to put some effort into it.
…because it is too hard to make realistic textures if you are not experienced
Another thing is that you can test yourself = make a character with 5000 tris (that’s too many
if you ask me! … but the machine can handle it very well).
It does not have to look good, CPU does not understand ugliness.
Apply whatever materials you want on it and press play. Have it rotate and move limbs
to emulate realistic conditions. Also, include a script that does 10.000 math operations
per frame.
Then estimate how many characters you will simultaneously show and press Shift+D
to duplicate that character that many times.
Press Play. Moment of truth.
Then test any Transparent objects, for example smoke, fire, explosions, magics, and
generally things that will have a png texture so that you can see through them. Those are
quite consuming, but usually they are flashes and do not persist on the screen.
Keep duplicating until your computer starts to puff smoke haha.
And finally, do not develop a game on a high end computer.
You will be amazed at what works and what does not.
