The Process/Workflow of Designing Game Characters

Recommended Posts

Posted (edited)
Hey All,

I just joined today and I've been poking around at a bunch of different threads picking up whatever tidbits of knowledge I can, but I figured the fastest way to get direct answers is to just post my questions. And if these questions have been answer before, you can feel free to yell at me and I'll search harder next time.

I'm very new to this whole game design thing, especially on the art side since my formal background is as a software engineer - I picked up game stuff as a hobby just a month or so ago and I've been messing around with Unreal Engine 4 and Unity. I thought it'd be a fun and a cool way to spend my free time, but I've quickly come to the realization that I'm diving into a gigantic universe with basically endless information to learn... it is fun, but it's also all kinds of stressful when you're a perfectionist like me. I'm crying just thinking about how much there is to do and how little I know to design the game of my dreams.

But I guess the first step is to always just learn as much as I can, so here goes my list of questions:
• From what I've seen poking around, I hear that some people design characters via sculpting with ZBrush from a base mesh or some other programs and then convert their high-poly meshes to low-poly meshes for use in-game. That sounds really efficient to me for designing unique characters - is this a common workflow in industry?
• How do games achieve the illusion of having a ton characters/people (NPCs)? I understand that characters have "skins" that are interchangeable. Is this simple a swapping of textures on the same model? Is it an actual change of the entire polygonal mesh?
• In theory could you apply the same animations to multiple skeleton rigs of varying sizes? I'm under the (possibly false) impression that you can have different skeleton rig sizes as long as the joints are the same, right?
• One of the big concepts I'm trying to achieve in my game is hyperrealism in terms of movement for characters. Is it honestly possible to animate all of the movements using stuff like Poser or is it really better off mo'capped, like I've seen in industry game studios?
• With brutal honesty, how much can one achieve as a solo developer/designer? Is it just a matter of time, or are some goals completely out of reach for a single individual?

Thanks for your time guys. I'm looking forward to building stuff over the next several years.

Edited by mikeyvxt

Share on other sites
Posted (edited)

Hey All,

I just joined today and I've been poking around at a bunch of different threads picking up whatever tidbits of knowledge I can, but I figured the fastest way to get direct answers is to just post my questions. And if these questions have been answer before, you can feel free to yell at me and I'll search harder next time.

I'm very new to this whole game design thing, especially on the art side since my formal background is as a software engineer - I picked up game stuff as a hobby just a month or so ago and I've been messing around with Unreal Engine 4 and Unity. I thought it'd be a fun and a cool way to spend my free time, but I've quickly come to the realization that I'm diving into a gigantic universe with basically endless information to learn... it is fun, but it's also all kinds of stressful when you're a perfectionist like me. I'm crying just thinking about how much there is to do and how little I know to design the game of my dreams.

But I guess the first step is to always just learn as much as I can, so here goes my list of questions:
• From what I've seen poking around, I hear that some people design characters via sculpting with ZBrush from a base mesh or some other programs and then convert their high-poly meshes to low-poly meshes for use in-game. That sounds really efficient to me for designing unique characters - is this a common workflow in industry?
• How do games achieve the illusion of having a ton characters/people (NPCs)? I understand that characters have "skins" that are interchangeable. Is this simple a swapping of textures on the same model? Is it an actual change of the entire polygonal mesh?
• In theory could you apply the same animations to multiple skeleton rigs of varying sizes? I'm under the (possibly false) impression that you can have different skeleton rig sizes as long as the joints are the same, right?
• One of the big concepts I'm trying to achieve in my game is hyperrealism in terms of movement for characters. Is it honestly possible to animate all of the movements using stuff like Poser or is it really better off mo'capped, like I've seen in industry game studios?
• With brutal honesty, how much can one achieve as a solo developer/designer? Is it just a matter of time, or are some goals completely out of reach for a single individual?

Thanks for your time guys. I'm looking forward to building stuff over the next several years.

1) ZBrush is pretty much an industry standard. From personal expierience I'd say there are alternatives just as good, but cheaper, if you want to get into that kind of thing. There is also Blender as a free alternative, not as good, but usable.

But first things first.

ZBrush is MOSTLY used for sculpting high-poly models that then get baked onto low-poly retopo meshes. ZBrush usually is used with a base mesh created in a boxmodelling 3D App like Maya, Max or Blender. There are tools and workflows to create the base mesh in ZBrush itself, but the oldschool way of creating the base somewhere else is still often found. Also, oftentimes base meshes are kept and reused from project to project.

Efficient - Define that word before using it for describing the AAA studios art pipeline of choice. Which ZBrush mostly is. MOST Indie studios will NOT have ZBrush in their pipeline, at least not for high-poly scrulpting, BECAUSE high-poly sculpting is not a very efficient way to create 3D models... especially when you are creating a simple low-poly look, like many Indie games do (which do it just as often because they lack the resources to create beautiful AAA style high-poly art for a full game as they do it as an artistic statement).

Now, if you like high-poly 3D art, I am the last person trying to tell not to touch it with a ten foot pole. Because creating high-poly art can be a ton of fun, especially after you have spent many years to get a certain degree of mastery on the tools and develop your artistic skills.

But just be warned that you will have a rough time trying to create ALL the art for a full game, no matter how small that game is, in true AAA high-poly fashion, and you will need a certain degree of artistic skill, with simpler art styles often letting you get away with less artistic skill.

And that is before getting into how retopology, rigging and animating very detailled models is even more of a pain, and demanding a whole new skillset to be developed again.

2) Most of the more detailled "skins" nowadays are different 3D models that can be mapped to the same rig and use the same animations as the original 3D model (which in turn means the proportions have to match more or less, which of course is also a necessity for gameplay reasons).

If you can share rigs and animations between models you can save a ton of time, and you can go a step further and create modular models (which is how some character designers work, apart from morph targets / blend shape / shape key systems that actually morph the mesh between different end states).

Most of the time NPCs in crowds are much simpler 3D models than NPCs you find standing around on their own, which again are simpler than important story NPCs, which will never be of the sam quality as your Player characters. Just look at different characters in your favorite RPG and see how much less gubbinz an unimportant NPC often has on the model compared to Geralt or whoever your Player character is... which makes sense storywise of course, but is also simplyfing modelling and animating that NPC compared to your player character.

Mostly the illusion is making clever use of the limited resources you can spend on your NPCs... and if we are talking about AAA games, even limited resources for that game being a ton of money :)

3) Not the most knowledgeable person when it comes to animation retargetting, but AFAIK that is the term you are looking for. You are trying to retarget an animation to a different skeleton.

4) Hyperrealism is Hyperexpensive... and we are talking about tens of millions of USD here, not just some grands. But sure, go for it.

Mocap is a good solution to achieve the same result a more skilled animator could do by hand. It can also be a faster way of achieving the same results given your Mocap setup is good, and everyone is well trained in it. Just be aware that most low cost mocap alternatives are... suboptimal, to say the least. The like of iPi and similar systems working with multiple cameras can be a bitch to setup, need a lot of room, and can need a lot of postprocessing or reshooting as the result often is riddled with noise.

I have seen a very promising low cost setup using accelerometers and I guess bluetooth units strapped to the limbs to capture the data. Created by some chinese company, sold for <5000$, it was showing promising results for a reasonable cost. But before pondering about that, you need a good rig. And if we are talking about Hyperrealism, you will spend quite some time setting up that rig. After all, what good is the best animation if the Knee is bending at the wrong place, or the model is not deformed right with blend shapes / shape keys when a joint is bent (shape keys are often employed to work around the limitations of simple weight painted deformation of a skinned mesh)? What good is it when the clothing is not also animated nicely (often just slapping a cloth physics setup on it will give you suboptimal results or costs too much performance), of the face looks like a death mask because no matter how many bones you throw in there, you will not get realistic results without shape keys and wrinkle maps? Another thing is that some animations just cannot be done with mocap. When your hero needs to be larger than life, you will either struggle to find an actor that can stretch its limbs enough or do all the backflips and trickery, or you will need to do a lot of manual retouching after mocap anyway. So learning to animat models the manual way IMO is not a bad idea before getting your feet wet with mocap. 5) Depends on your timeframe. In the usual 2-4 year timeframe, with a single dev? Not much if you are aiming for AAA visuals and scope. Especially not when also working a day job or part time (which, unless you already have some games that rake in a lot of money, or live in your parents basement, is a necessity for most devs, even the ones that want to pay their bills from their game development), and not being an industry veteran with many years of expierience. My recommendation would be to get your feet wet, learn the tools of the trade while also getting a reality check as to how long some things take, while working on very, very small projects if you need that to motivate you. You can always tackle a much bigger project some years down the line when you know what you can achieve in what time, and are ready to spend 5-10 years of your life on a slightly larger project (that is still far away from AAA quality), or are ready to go out and find a team to join (just be aware that its hard to find people working on YOUR ideas for free), or spend some of your money on external help (which, given you have enough cash you can burn is the best idea probably... just make sure you are ready to loose all that money, game development seldom is giving you a nice RoI unless you can spend big or spend wisely) EDIT: oh, about the ZBrush alternative I was talking about. I am using 3D Coat mostly, which CAN be just as usable as ZBrush as a sculpting tool, while being much better as a painting and retopo tool and costing half as much as ZBrush. ZBrush does outdo it in some areas in sculpting (mostly some specialist Brushes which can speed up your work if you have learned how to use them), and doesn't need a beefy GPU to run as most things are running on the CPU while 3D Coat pushes a lot of things to the GPU, but in turn you get a dated, obscure UI in ZBrush that will give you nightmares if you do not use ZBrush every day, whereas 3D Coat has a more friendly, more intuitive UI. I ended up spending on a license for both of them, but tend to use 3D Coat over ZBrush most of the time unless its something where the speedup by a specific ZBrush Brush is making up for the headache of getting into the ZBrush UI again. Which to date is mostly sculpting rocky surfaces from scratch. That "adaptive trim" brush in ZBrush is just a godsend for that. Edited by Gian-Reto Share this post Link to post Share on other sites Hey Gian-Reto, Thanks for the in-depth response and the reality check. I'm excited and ambitious, but I'm also fully aware that this is a huge universe I'm diving into. I don't expect to put out A+ games on a yearly basis or anything, and I'm also fully aware of my own limitations and skills as an individual. I'll definitely take your advice and get my feet wet, and I'm not about to let the vastness of the field discourage me from trying. All of this stuff looks awesome. Efficient - Define that word before using it for describing the AAA studios art pipeline of choice. Which ZBrush mostly is. MOST Indie studios will NOT have ZBrush in their pipeline, at least not for high-poly sculpting, BECAUSE high-poly sculpting is not a very efficient way to create 3D models... especially when you are creating a simple low-poly look, like many Indie games do (which do it just as often because they lack the resources to create beautiful AAA style high-poly art for a full game as they do it as an artistic statement). I guess to define "efficient", I suppose I mean the ability to put out the highest quality models/art as fast as possible. For AAA studios with tons of resources, Zbrush High-poly design -> retopologize to low-poly probably works (and it looks good). But you mentioned, Zbrush doesn't really show up in an Indie Studio's pipeline. So what do they do differently? Is it just different software or do they use an entirely different workflow/pipeline? At some point they'll still have to do 3D modeling to throw into the game right? Share this post Link to post Share on other sites Posted (edited) Gian-Reto, said most of what was needed, there is only one thing I want to stress. Making 3D models and Animating, are two completely separate fields. Because mastering any skill takes a live time, most people focus on one skill while only lightly dabbling in a other. What that means is a 3D artist can make realistic characters, while only knowing the basics of animations. Animators can create realistic animations while knowing only the basics of modeling. The time it takes a 3D modeler to make his first realistic human is around 4-6 years after making there first model. This is 4-6 years of extreme 3D workouts, spending more than 12 hours a day modeling, it's at this point that the modeler is considered professional. Even after that time the modeler still has to spend hours modeling, just to retain what they know. There isn't a single 3D modeler that I know who doesn't start the day with a warm up model. You can expect that animation is just as strict to master, that is why it takes a team to make a single character from start to finish. I guess to define "efficient", I suppose I mean the ability to put out the highest quality models/art as fast as possible. For AAA studios with tons of resources, Zbrush High-poly design -> retopologize to low-poly probably works (and it looks good). But you mentioned, Zbrush doesn't really show up in an Indie Studio's pipeline. So what do they do differently? It's all about cost, making a zBrush model and retopologizing is expensive. Doing a character of that level will take over a week for just the mesh. When doing contract work for AAA developers, I charge$2000-$3000 USD for the first mesh, then charge around$800 for every change. The total cost ends around $40 000 -$60 000 USD, depending on how complex the model is.

As the quality of the model scales so does every thing else, a 60 000 triangle model needs motion capture; using simple hand animations will cause the model to look stiff. Once motion capture is used you need to pay an actor, a camera crew, rent some motion capture suits and hire a animator.

I am a professional 3D modeler, even in my own games I stick to low quality models. At best one of my own game models will sell around $1200 if it's a full IP agreement. I am only using zBrush as part of the planing, to quick sketch a model. Indie developers have a lot of limitations, just the license cost of zbrush for a small team is more than$8000 USD, that isn't even with the cost of having 3DS Max, Photoshop, Quixel, Simplygon and SpeedTree.

So no indie team will aim for that level of quality.

Edited by Scouting Ninja

Share on other sites

Okay, so let me try to put things into perspective here.

This is a character I modelled and animated last year. I wanted to see how quickly I could create a character if I stayed with the minimal amount of details needed for a certain view. Also my first time rigging and animating a character.

Took me 8 hours for the high poly sculpt (which was about as detailed as you can still see on the gif above, character was planned for a "commandos" style game, where character can be more detailled than in an RTS, but are never seen up close).

Took me about 8 hours for low poly retopo and the textures (well base ones, I added a subsurface texture later which took another 30 minutes to create).

Rigging and creating 2 animations by hand took me about 35 hours. But as said, first animation project, and potentially to complicated with all the stuff hanging off the belt, as well as clothing animated with shape keys / blendshapes.

Character is pretty usable now. I would have to go back and redo part of the rig though because I didn't combine all the meshes before creating the shape keys / blendshapes. And of course a ton of additional animations would be needed, which because of how the rig is setup, are not as easy to do as I would have hoped. I had a ton of troubles with getting the rig to stick to the two handed weapon without the hands moving too much when the arms are moved, so I would have to retouch every animation frame by hand to make sure the hand does not move around on the gun.

Not to bad for a first try, still, how many characters do you need for a game? This would quickly become a fulltime job on its own! And that is just for rather undetailled NPC characters suitable for RTS or maybe a commandos style game, but not an FPS. Add a little bit more details and a more deliberate design phase for a Player character, and you will struggle to finish high poly modelling that in 16 hours. Add the details needed for an FPS, and you can probably quadruple the amount of hours spent on high poly modelling.

Textures were mostly flat colors, using baked maps and dirt maps to enhance them. Also I have quite the pipeline by now to bake out all the needed PBR maps. If you want to go more handcrafted, you can again double and triple the time needed, especially for a more detailled FPS ready sculpt.

My time needed on rigging and animation most probably are so high because of my noobishness... I was looking up stuff as I went along. And the rig wasn't the most straight forward with the twohanded weapon, animated clothing and bags / grenades on the belt. Still, again, if you want to go closer with the camera to the character, you will get away with less things. I first was convinced I needed wrinkle maps for the clothing to look nice. Turns out no, at that distance you can hardly see that the folds and wrinkles on the clothing is pretty static.

When looking at clothing from close up, and wanting hyperrealism, you might want spent time on that. No point in having this nicelly modelled character when the clothing looks stiff and lifeless.

Last thing to say is, I am by no means a pro. I have some artistic skill I think, and spent the last few years doing quite some 3D projects in my free time, but I am a programmer in my day job, so clearly someone who did this for a living could slash the time needed in half, especially when already having a more efficient art pipeline and bits and bobs to kitbash with (only one I used here was the gun, which was made from an existing gun souped up a little bit to fit the character).

As you see, its a combinatorial explosion the higher you move up the quality meter. Even the pros spend months on some of the AAA Characters alone. 3, maybe 6 months is not uncommon for a very detailled model. Sure, with expierience you get faster, and you can improve your pipeline with tools that let you be more productive.

The most important factor though is to limit your scope. Moving down a notch on the quality might slash the time needed to create the character not only in half, but maybe down to a quarter or so. And for certain types of games, that quality might be completly acceptable, or even not distinguishable from a higher quality character. For example thirdperson cameras that do not let you get close to any of the characters can let you get away with far less details.

Art Style is often not chosen just for a certain effect on the player, but also to keep the hardware from being fried by polygon overload, and most important to keep the load on artists lighter.

This is why Indie games usually go for a simpler art style, if they have to deliver the same amount of game with a fifth or tenth of the artists, no amount of pipeline optimization and procedural generation can prevent them from making some hard decisions to lower the visual quality a notch or two.

Create an account

Register a new account

• Forum Statistics

• Total Topics
628662
• Total Posts
2984105

• 10
• 9
• 9
• 10
• 21