[Game] Blood, Sweat, Gold -- modular character system

BIGTIMEMASTER
greentooth
Online / Send Message
BIGTIMEMASTER greentooth
This will be the artist devlog for a turn-based action strategy PC game scheduled to launch early access around September/November 2018. 

The game has been in development for more than a year by a sole programmer/developer and is in a fully playable state (some bugs crawling around, of course), but it lacks 3d assets that 1. look current gen, and 2. are optimized. 

The most important thing in this game is characters, as the primary focus for the player is to recruit characters, train them, and keep them alive (perma-death) long enough to build an experienced mercenary fighting force. However, each players experience will be unique, as both the strategy map and characters are procedural generated. So then, the mission is to strike a balance between creating good looking characters that don't seem like cookie cutter pieces, but also get enough unique characters with variations finished and polished to an acceptable degree by launch. Right now the plan is one outfit with texture variations, along with one unique face variation per week.

The player will be able to choose their characters face/body type, hair/beard, and swap head armor and outfits. To meet these demands, I've developed a system based on some archived threads from this site, although being new to 3d art, a good deal of trial and error prototyping has been necessary to work out all the kinks and develop a solid plan. So, after three weeks of prototyping, now I am beginning to build the first characters, and since a modular character system is both a complex undertaking and also something that is no doubt very interesting to indie/hobby dev's, I will record lessons learned and techincal notes here, as well as post the characters. 

Critiques are of course welcome, but keep in mind I am working in turbo-mode here, and so the most useful critiques are generalized ones that I can apply to future models, or put away until after the essential production is done and it's polishing time. 

Notes about the art style : 
1. Realistic
2. 10-11th century (chainmail is the highest form of armor)
3. Emphasis on the desperation of the peasant class involved in the game.
4. Three distinct body types (though all rigged to same skeleton) - "Soft" man, "Hard" man, "Great" man. AKA peasants, experienced fighters, huge motherfuckers.

First prototype character

First prototype outfit, beard, hat

texture variation

second prototype outfit

texture variation


With problems of rigging and in-engine performance and operation figured out, I started the first of three base characters last week. Here is the textured game model:


Replies

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Notes about a modular workflow: 

    Seen here is two minor variations on the same texture. As I work on texturing each character, I keep very meticulous layers inside Substance Painter, and I separate channels completely. For instance, the redness in the second image is a separate layer sitting on top of the splotchy yellow layer in the first image, and that splotchy yellow layer sits over yet another layer. Added altogether you get one look, while subtracting any gives another. If I do combine colors and, say, roughness in one layer, I make sure to annotate that for easy understanding down the road. Most the time for fill layers though I only work with one channel type at a time, so that if I change my mind I can make easier workarounds that way.

    I make sure to save these Substance Painter projects out into a neat hierarchy so that I can easily come back at a later date to build more variations without having to redo any work. How to organize and set up the hierarchies is the hard part, and I really didn't know how to do it until we got well into prototyping enough to figure out precisely how we needed to name and organize for the in-engine scripts to to work and also to meet the games design criteria. This would seem like something you'd have to sit around thinking about, but it's really too complex and confusing if you haven't done it before. You just have to get started, then the answers start to appear. Don't waste time trying to get a complex thing totally straight in your head. Prototype and it starts making sense, piece by piece.


    Eyes, teeth are 256 material. Bodies/outfits are 2k. If from the top-down perspective the 2k level of detail is lost, we will LOD to 1k materials -- or of course if there is performance issues. Hair and beards will belong to 2k atlases, and will consist of meshes for volume with just enough haircards for a fuzzy silhouette from typical game camera distance.
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    The peasant base model. Wanted to do a lot more wrinkles, but that will have to wait for more time later. 
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
     I have little experience with rigging, and these are really good animations so my lousy rigging really shows, but nonetheless I have made some improvements over the course of several characters. I'm hoping to have a week or two at some point to just scour over the animations and polish the rigging as much as I can.



  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    The Legend. 

    For rigging purposes, right now this guy is the same size as the others, but in game the plan is to upscale him just enough to make him very distinguishable on the field. 

    Any MMA fans can probably tell who was my main reference. I chose Fedor because he does not look like a modern weight lifter, but has a very imposing, bear-like presence. The kind of guy with a lot of natural power, even if living in less than ideal conditions. 

    I found a way to get the body hair done very quickly using some of Substances fur generator materials and a few layers with mask. Pretty simple, but probably not obvious to raw beginners who might be more inclined to do everything by hand. Of course, going the slow way gives more custom and higher quality results, but the goal for this project, right now, is to meet a certain quantity, and thus I can't spend two weeks on each character. In total, the texturing took me about 2 hours, which I think is pretty fast! The key is that I basically have a system now, and I just plug in new colors, textures, and make little tweaks. Also, I hold onto the Substance projects, so I can come back later and do something like turn off the body hair, or intensify some of the color noise layers to give a more weather-beaten look, or make him look sweaty, or dirty, etc. All very fast to do as long as you have the layers organized, named appropriately, and you don't cram too many things into a single layer that you wouldn't remember later.

    Next will be a few of low-tier outfits. Peasant garb, bandit rags, stuff like that.



  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Peasant outfit. Next up, short sleeve tunic, baggy trousers, tall soft leather boots. And hopefully a coif. But will be out of town for 2 weeks. :(

    These were made in MD, which I still kind of suck at. But after some hours playing around I'm able to get the results I wanted. By the end of this project, I think I'll be a pretty good seamstress.
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    snip* 

    I shared some workflow advice, but it was actually wrong! I fixed the mistakes that I had made, but I'll wait until the end of this project to do more detailed breakdowns to avoid putting out bad info.
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Big stupid hat. It's just something quick for testing. Let a guy have some fun, ok?

    Getting the modular system to work is a big, complicated task of organization and lots of back and forth between the artist and the programmer. You really have to have strong communication, otherwise you're going to get frustrated with the tiresome back and forth of trying to define a system of organization with painstaking detail, and then design the tools to implement it in engine. 


    Here is what the basic structure looks like. It seems simple enough, but it really depends on a variety of factors, most importantly the game design. Anyway, after I finish this project, I'll do a detailed breakdown and maybe even a step-by-step tutorial to cover all the annoying technicalities when building modular characters. I don't think it would be much use for me to say anything about it right now, as I end up doing a lot of things over -- still learning as I go.

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Stuff that happened while I was on vacation. Lots of polish is needed, but it's nice to see things starting to come together.



  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Finished the chainmail outfit in 11 hours. I did use a purchased chainmail substance though, so that's a major cheat. Whenever possible, I am avoiding sculpting and baking, and doing the texturing as procedural as possible as this gives me the ability to make lots of variations quickly. For some outfits, typically lighter fabrics that should fold and crease more than thick, heavy layers, I still prefer to start with Marvelous Designer, but for the heavy quilted gamebsons, the chainmail, the leather, I find that I can get fine results with texturing alone. 

    As always, this is work in progress. For now, I just get the rigging to a workable state for demo purposes, but most of this clipping will get cleaned up down the line. 
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Managing about 1 outfit per 2-3 days. This is fast enough I don't have time to mess around too much, but I think in the long run it will be better if I can look at all the outfits together as a whole in engine, and then do all the lookdev, rather than spending a full week per outfit, which may make a clearn result, but chances are I am going to want to make quite a few changes once the mob is complete. This way I can be sure each fills a certain niche, and they kind of work towards a cohesive aesthetic. And once I'm certain I've met the minimum number of outfits, then I can leisurely mess around and try to get the most out of each one.



  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Winter outfit. This one is lambskin, but there will be a wool alternative as well. Fur cards are placeholders for testing right now.

    Side note: combat system is getting a major reworking and the prototype is pretty damn intense even with only part of the systems in place. Very tactical turn-based game play. I'm not really into the genre, and yet it's hard to put down once I start playing. Looking forward to sharing some gameplay videos here in a few months.
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Animations aren't mine of course, but some custom work is being done with the timing to give a good feeling of control for the player. 

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    leather lamellar outfit. Everything is designed for maximum flexibility. Once the target number of outfits is finished, I'll be able to come back to the substnace painter projects and quickly generate alternate colors, levels of grunge, etc. The final aesthetic will likely call for much filthier outfits than this, however since we will have repeating uniforms across the battlefield, I want to offer variations of grunge patterns even if only slightly different so that, say, two uniforms with the same colors can still look different with their unique grunge patterns. 


  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    The theme of the game is centered on peasants -- filthy, foul, boorish peasants wollerin' in the stinkin mud just like the swine they is -- so I am trying to find ways to get that across with the low-tier outfits. As much as I want to stick with totally non-destructive texturing workflows, in this case I found that a bit of carefully placed alpha's to suggest torn fabric around the seams really helped. However, all the mud, stains, and some subtle noise layers remain so that I can randomize their placement for variations.

    I also have a tendency to really avoid exaggeration in favor of subtlety, however in the screenshots below, this is actually quite a bit closer than the game camera will typically be. So a lot of times, my smaller details get lost, and thus I am learning to overdo some things so that they can actually do their job in selling the outfit's appearance. 

    Some final tweaks to do with this and then rigging. Then... another one.

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    A lighter-weight chainmail outfit. One more to go to finish the minimum amount of outfits we need, then I am going to change gears for a bit and make all of the faces.

    I've been working to trim the polycount down quite a bit. I wasn't paying attention to is much before, but this latest model is about 50% lighter than my first models, which had a ton of unnecessary density. I think this weighs in around 10k, which is about right to not show jaggy edges at our typical camera distances. I know people say tri-count doesn't matter much anymore, but when you multiplying each of these characters 10-30 times on top of a scene that might already have a million tri's (all the lighting and whatnot figured in), and you are trying to keep this running well on lower-end platforms, it at least pays to not be wasteful.
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Alright. This makes 3 outfits per 3 tiers, a total of nine. I'd like to do a few more for each tier, specifically the lowest tier (a few more variations of ragged, pathetic looking peasant clothes), but for now it's time to make a lot more faces. 

    My plan is to work from the three base body ztools, sculpt a new face (leaving below the neck untouched), baking the new face, and then bringing that new head into the existing Substance Painter projects so that I don't have to make the base skin textures each time, but can just change details and tonal values to give each face a couple skin tones and unique details.

    The three basic character archetypes will also guide the face sculpts. Legendary characters will be big, powerful faces. Hardened characters will be lean, mean, and strong looking. And peasant characters will be lean, ugly, and dopey.

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Mustering the mob. 

    I've got about 3 more months until were going to publish the early access game. Right now the character models are finished, and I started to make a handful of texture variations for each and will probably continue with this until I finish as it's kind of an assembly line thing with a lot of organization I got to keep in my head so I don't make mistakes -- maybe a week and half to two weeks. 

    If anybody is following along and has some suggestions/critiques, now is a great time to share them as I am picking at each model and not in such a rush anymore. I'd really appreciate some feedback, no matter how big or small. I probably won't be able to make any major overhauls, but that doesn't mean I can't learn from the advice for future work.

    One thing I am considering is making one or two scabbard models that can work with all models. We don't have the means to have custom scabbards per weapon, but just like one or two that can just paste to the waist for anybody with a sword equipped might add a lot of visual interest to the outfits without a big time investment. 

    Beyond that, some  outfits will get tweaked a little for better silhouettes and optimized for performance. Overall, I feel like assembling the mob in the muddy field here for look dev while I make the texture variations is really helping me bring them together into a more cohesive aesthetic. The plan is for each outfit and each hat to have 3-5 color variations, and 2 levels of dirtiness. 


    Beyond the art, the combat game play is getting more fully fleshed out, and I have to say, with only a quarter of the final content involved in the current test scenarios, it's surprisingly deep and challenging game play. There is much balancing work to do, but I easily spent five hours the other day playing it over and over, and I'm really happy with how challenging but consistent and fair it seems. Of course the key here is that the programmer is highly  competent -- which means once we figure out a concept, it's not much time before we are able to test it and refine it.


    I think the "old rusted" helmet doesn't look right. I'll probably revisit that one later. Perhaps adding some tiny imperfection to the silhouette, along with some large dents in the normal map will help? And tone down the vibrancy of the rust? Or just less rust overall?

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Last few days have been pretty frustrating. In finalizing the rigging for all the meshes, I had to redo some of them more times than I have patience for. I wonder if I am not using a Maya friendly workflow, what with so many meshes in a single scene all rigged to the same skeleton. Problems with the mesh history causing bizarre issues, non-deformer history utterly destroying things, some meshes getting duplicated inexplicably.... a total nightmare. 

    Anyway, such is life. Take a break, play a game for twenty minutes, and the motivation returns. It can be frustrating to move past a problem, however, not fully knowing what caused it in the first place. 

    Got another outfit and helmet done with texture variations. More trouble there with shader issues, and I had to learn more about how lighting inside Unity works. After getting the lighting squared away, and using a reflection probe (which was a mystical magic orb to me until now), I'm starting to feel pretty good about the quality I'm able to achieve in this test scene. Of course, the lighting in the final game isn't up to me, but it's still good to know your engine. 

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    I had been unhappy with the chainmail outfits silhouette and general look. It looked very doughboy-ish. I thought it might even be worth scrapping and redoing, but I found a few ways to salvage it and I think it looks about right now. 

    1. I simply tweaked the vertices defining the silhouette very slightly on the game model. I tucked in a little bit until the pecs, bulged out slightly above the belt, just small stuff like this went a long way into dramatically changing the appearance of the outfit. 

    2. Using the curvature map (extremely useful, always bake one!) as a mask on a fill layer with height information, I can accentuate the existing curves. Add a blur filter on top to even things out, and you can easily make folds look more intense, less intense, sharper, duller, etc. 

    3. Run it through animations. All of the animations we are using never put the model in anything like an A-pose. It's really helpful to continually review the model in action, throughout all of the animations. Initially many of my models had pretty thin shoulders, as I wanted them to look somewhat emaciated, but I found that when animating, this often led to some pretty bad looking arms and shoulders while swinging weapons. So increasing the volume of the shoulders more than I thought necessary actually gives a better look for the final, in-game and animated model.
  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    Alrighty, that's all the outfits finished for now. 

    Originally, the lead developer was pretty against having "bright" colors as the theme of the game is to be grim, but all the reference of clothing from the period (11th century europe) that I could find was pretty colorful. I decided then to stay pretty true to the ref I was looking at because I thought if all outfits were drab colors, they'd all just blend together into a general mush. I did compromise somewhat and tended towards simple colors like green, blue, black, white, and red. I felt that pushing a grim aesthetic could come more from aging the outfits, dirtying them up, and also using lots of natural effects, like fog, rain, and chilly looking atmosphere.

    All in all, I think I am pretty pleased with the general feel of the mob. Will be nice to add some more outfits in if I have the time. I'd like to see more tattered peasant outfits, more leather, and some more interesting silhouettes in general.

    Some outfits I like more than others. I don't feel the leather lamellar fits -- it's too Byzantine -- but whatever, some people may like it and it's probably better to have too much than too little. 

    Next task -- about 20 or more face variations. The goal is for fearsome, revolting, horrifically ugly mugs that will make you glad you were born in this century. But of course, I don't have a week to spend sculpting wrinkles into each face, so the challenge then is to think of the most essential features that can easily be discerned from a distance, and also work in a way that allows me to come back and easily make changes in the future. This ought to be fun.

  • BIGTIMEMASTER
    Online / Send Message
    BIGTIMEMASTER greentooth
    ***gameplay videos coming next week!!!***

    When starting the heads, I realized just how much I might improve the original peasant model. I've learned alot from this project so far-- especially how to make smarter use of UV's -- and so I couldn't stand to not redo some of my earlier work.

    The first new head, then , is a rehash of the original peasant model, but I spent a day to make a new sculpt for the body as well, and also totally redid the UV's. Then I used a full day to get a Substance Painter project set up with an easily modifiable layer stack so that, moving forward, once I sculpt a new head, I can get the textures done quickly. I hope to get a couple unique heads done per day for the next couple of weeks.

    As with the outfits, each character will get a dirty and a filthy version (though I think I really overdid the mud on this guys filthy version), but each character will also get two skin tone variations as well. 

    Original peasant model in the center, which I spent much more time on. Hopefully I can continue making big improvements like this, but I think I'll show the most improvement when I can sit down and spend two weeks to a month just on a single character -- getting everything just perfect. I'm having fun with this project, but I am getting antsy to put all my new skills to use like that.


Sign In or Register to comment.