What are some options for adding customized characters for someone new to it?

Development Rendering
25

@jwatte, I appreciate your insight on the topic and I agree and would be in the market for a product like that. I’m thinking DAZ3D is positioning themselves to fill in this role with Characters and Accessories for UnrealEngine.DAZ Bridge is the first step. What remains is an embeddable module/plugin to assemble and save/load Customized Entities that can be integrated into any game product developed with the engine. This is an area we intend to fill by developing a Universal Constructor to be used in the assembly of multiple entity types in multiple game products.

I call it the HyperBasher, as it inspired on the *Kitbash *approach to assembly. I hope to build a large and composable set of assets and provide templates for others to use to create assets for use in customizing entities. I’m no longer pursuing a single standard in favor of building a online library and catalog of various base meshes and compatible parts/accessories. The HyperBasher works with unrealengine most basic yet powerful feature: Actor hierarchal attachment scene components. Using this feature for arbitrary attachment provides the most flexibility to meet most Entity requirements.

The HyperBasher is modular in design and embeddable into ue4 applications a blueprint package. If one doesn’t need the assembler, the external save/load functions of System can be implemented as an independent module that can be embedded into applications for just importing assemblies. I am proposing the APB Customization, but I think its going to take more than just a standard to pull it off

  • Master Base Models
  • Supplemental Asset Packs with Compatible Parts
  • Online Catalog of Compatible Parts
  • License that allows existing models to be used only as reference.
  • Embeddable Assembler Module
  • Embeddable Load/Save Module
  • Decoupled GUI (for custom GUI wireup)
  • A Foundation of Developers to produce and maintain all of the above
26

Assembling in engine with masterpose component comes with hard limitations.
if you are making something like daz, then you need to offer the option to export a single mesh with 1 UV, rather than several meshes all with separate materials.

27

Pretty sure that could be either just negotiated with EPIC, or worked around – we see helmets and swords and other such items that don’t work “on their own” anyway already. If you have something that works, it’s my experience the folks in North Carolina are quite willing to cooperate!

In the best of worlds, EPIC would throw some of that megagrant money at creating assets for a paper doll character/avatar system for Unreal and make all the base meshes and “critical mass” initial assets part of the free collection :slight_smile: Make 'em better than the Unity asset store ones and keep trying to lure more people in.

Make Human is interesting, although the first time I tried it it was so bad I haven’t dared going back even though people say it’s better now :slight_smile:
Daz has too high poly – their interest is in people rendering static scenes, more or less, or at least doing an offline production.
Mixamo is weird as they play mostly about animation, rather than character pieces. Maybe they’ll get there? Again, when I tried them a long time ago their tool support was pretty buggy; maybe it’s gotten better with age.
(One of my earliest tech roles included developing and maintaining 3ds Max exporters, taking Physique data out to game-ready paper-doll avatars. The industry has gotten somewhat better since then, but surprisingly LITTLE better! DAE was a total clusterbomb, but FBX was probably the best thing to happen in the last 20 years. Which isn’t a very high bar.)

28

Dunno about formats before FBX, as I wasn’t making art for games back then, just cinematic renders of logos and such. Had never even though about rendering outside of a 3d program either, since all of what was needed was covered - albeit mortally slow.

anyway, currently, make human produces a mesh that you can manipulate as you wish numerically. Put in measurements, get an almost accurate mesh.
Unwrapped and weight painted - which is the big part.

Now, my theory is that just like they handle that mesh creation:
A base template is used -probably- on which the vertex are shifted/distributed - Not actually sure, would have to triple check that by creating and comparing min/max length chars.

If this system works (and it does, it produces decent bases) then it could possibly be altered to do the same for a piece of clothing?
import a base list of vertex/model because of UVs, let the program manage, stretch and redistribute based on the pre-set values.

Naturally, with a bit of work, the same could just be done inside of blender. After all, there’s only about 24 measurements or so to handle.

Now, the system obviously works off an indexed set of vertex, or you’d have to adjust stuff manually.

Maybe it would be worth it to mathematically adjust garments instead?

Example,
select the underlying geometry, automatically inflate the object to be above the underlying mesh (which should be simple since you just shift on vertex normals based on the underlying shape).
At which point you need a way to alter the length, while preserving or adjusting the initial weight paint accordingly.

I do almost all of this manually in minutes.
Having a system that just adjusts it for you would actually be both fun to make and a time improvement…

the plus side is, it wouldn’t really need much more than your character base mesh and a clothing asset.

the down side is you have to manage your own geometry/remove the underlying as needed. merge as needed etc.

the extra plus side is you can use the asset into anything, since it has nothing to do with the engine…

Food for though with the new year coming in…

29

This is actually an idea I had a long time ago, but never managed to make happen (because, priorities …)
You could even do it with geometry position and normals, don’t need UV to match up!
A hat can have an “inside” that is marked as squishing hair; a head has an “outside” that is marked as inflating hat insides, …

Interestingly, you can “know” that the areas where the bones go, are generally “insides,” and thus know which direction to inflate meshes into. You’ll still get problems with things like arm pits and such, pieces inflating into each other. Enough math, maybe that can be fixed?

True, but if you want to build an RPG where the player finds a splint mail in a chest, and then wants to put it on, you kind-of have to build the thing (whatever it is) such that it works in-engine!

30

Nha, the principle behind the geometry is the same for anything.
You place a static mesh, or a different version of the skeletal mesh (in case you want the mail parts to dangle when you simulate physics or something), and you add to the player the matching skeletal mesh that’s built for the player.

If what you are saying is “I don’t want 20 assets for each character clothing peace just because my character proportions can be way different”,
Then it goes back to having to have morph targets or a similar system in place anyway - and marrying the 2.

At a base level the morphs on the base mesh can be made to work on any clothing piece.
Transferring them by vertex proximity like weight paint may be better, but all in all the base setup for that would all have to be put together specifically anyway.

From then on, all piece of clothing would also need to have a near identical vertex structure, so you would have to start modeling from the base part anyway.

The two systems could possibly function identically - at least within a specific value range min/max of the morphs.

in blender you can make the base model, set and frozen (applied shapekey) as you want…
you in engine you can import the same base model that uses the same shifting process to adapt to the character.

So long as its based on morphs, it could work and be performant…
unfortunately at least pre-chaos, morphs and cloth don’t go along.

31

Yes!

My complaint is that I can’t buy these kinds of assets ready-made in the marketplace, and where I can buy them, there’s either 1 base mesh and a number of morph targets, or like 3 meshes that can interchange, but no morphs. I’m mainly a programmer, not a modeler. I’m not going to spend all that time building meshes that will look like programmer art anyway.

It is understandable that I can’t buy what I want on the marketplace right now, because nobody with the prerequisite skills is going to put 6+ months into building the right set of meshes and morphs and textures, just to sell them for $19 a pop or $99 for the lot on the marketplace. Not gonna pay the rent for those six months of effort (per gender!)
But if the system was already there, and it was easy to ADD TO IT, then someone might spend a week-end building a new kind of armor, or some nice shoes, or a hairstyle, and sell it for $19. This is why I think it’s possible to “bootstrap” a aystem with enough of a critical mass.

If it does its job, the cloth simualtion will just conform anyway … although if there’s too little of it, you’ll get the “belly button showing” problem. Which might not be what you’re going for.

Capes and loin cloths, though – there’s a reason this is where cloth simulation is used the most!

32

I use character creator, which is a bit meh for licenses and pricing, but overall okay, even great to fast create and add characters, and very important for me, to add different cloth items / gear pieces. I am not aware of a better option at this point in time and space.

33

Doing it yourself from scratch is definitely a better option all around.

It takes maybe 2 weeks to learn how from 0 knowledge - like you have no idea what a normal map is 0 knowledge.
And you have no licensing issues at all, ever.

As you make more assets and get a feel for it - unless you strive for programmer art I guess? - you end up with viable pieces to a set.
After several In Engine pieces - because you have to learn how to paint the cloth too, making a new piece for what you need becomes trivial if even.

A t-shirt from 0 takes about 5 minutes to make (the mesh plus unwrap - painting the texture is a different story).

The most challenging bits are pants, because of the weight paint in the pelvis area which may need to be married to the underlying geometry (thighs for shorts, ankles etc).
Those can take maybe around 2 hours top to model and unwrap correctly.

I’ll also tell you about how long it takes to texture them…
depending on what you want - say a full set of separate items.
each item has its own material. Texturing it up in Quixel Mixer takes about 5 minutes. Maybe 10 if you want to add normals for buttons like denim or something similar that isn’t geometry but faked.

If you have a consolidated set - meaning 1 model - in which you need cloth areas married to the skin - than that takes a bit longer to setup in order to even start baking. Let’s say 2 hour tops or thereabouts.
with a color id map, the texturing process can take about half an hour or so.

Now, all of that assumes 2 things.

  1. you actually used a sewing machine at least once in your life, or you have enough analytical skill and brain power to see where the clothing you wear has seams - and guess how it would look taken apart and put flat (your UVs)
  2. you actually know how to use quixel for custom meshes, which is really more of an art than texturing with it.

let’s add a bonus point:
3)
You have your own scans of more fabric or knowledge on how to shoot scans and normal maps, because while qixel offers a few (mostly of forniture) fabrics, for the most part the fabric stuff sucks.

Mind you, I’m not saying the process couldn’t be simplified by having some tools made specifically for it.
nor am I saying you suck for using some sort of tool that lessens your work…

I’m simply saying that once you learn how to do it you are WAY better off - including for the purpose of getting paid gigs…

34

What’s your hourly rate?

35

I don’t do hourly for game stuff. You can PM me if you want more info.

If you want to known what a fair hourly wage on modeling is, GlassDoor gives it as barely above minimum wage. (Kinda makes sense when a monkey could be trained to do it :P)

37

That’s a possibility if I need to make something like it for the game im making.
Right now I have around 250 clothing items setup for a single mesh.
Maybe in the future I could figure a way to make it interchange regardless of underlying geometry.

The main issue witheverything right now is chaos. And chaos cloth.

I want to give everyhting a wide birth until the next engine version at least, when things should actually become more standard. Particularly since I have no intention of adjusting 100+phat assets multiple times.

As far as making a creator goes, as I mentioned, maybe as another Blender addon or - thinking a little futher ahead - as its own specific program. If I put time in it, definitely won’t be exclusive to unreal.

38

Character creator from realilusion and the unreal markeplace is by far the best option.
Forget doing things yourself if you are not an artist, that something is easy to do does not mean that you have the will to do it.
Then you can use freelance for more specific things, like clothes, hair, props.

40

Groom is experimental. So technically not a good idea to publish a char to marketplace with grooms.

Hair cards are tried and true, but they do add a lot, and I mean, a lot of verts. Not so obviously, more on short hairdos than for long hair ones.

That, and no matter what in .25 they all still light up like a light bulb if your lights aren’t setup right.
Even when setup correctly, let’s say that the normal of vertex on it matters more than on grass or similar.

One thing I have yet to try is to unwrap a channel of the groom as separate cards and manually paint a flow map normal texture, however the cards look fairly realistic even with near programmer art. When you balance the light and use real textures (not content example ones) its brilliant.

41

Don’t mind me, just OP eating popcorn and watching the show.

42

For those search words: The only three hits for “character creation” are “anime character creator” by Somndus Studios (not Unreal skeleton, and few separate clothing pieces,) and “Custom Character Creator” by CB Productions (only has “one of each” of base asssets) and “Assembly: Modular Characters” by StudioSyndiCat (which is mainly code, not a set of assets.) Search finds nothing by “realilusion” or “realillusion” (spell corrected.)

Do you have a link?

43

Search for Iclone and/or character creator 3

Its expensive, and perhaps more of a concern, the licenses aren’t good.

44

There is a 30-day trial that lets you export to unreal 10 times.
I think that the license does not give problems if you are not going to create content for others, and you only want characters for your game.
It is true that the store is very confusing and it seems that you have to buy a lot of things (it may be intentionally or the worst sellers in the world).
but if you just want rigid characters without headaches, just you need Character Creator 3 Pipeline only.

I don’t like the hair and neither the clothes. but you can also import clothes in obj and adjust them to the body and rigged.

45

This package has tons of characters that look different and can be quest givers:

The same developer has a few other packs of many-characters; “street wear” and “female” and “swimsuit” variants.

If the art style isn’t what you need, then you’re running into the problem that any game cobbled together from the marketplace won’t have a unified art style. Them’s the breaks.
If you need fantasy theme, you could buy a bunch of the “hero” and “archer” and “swordsman” and even “skeleton” and “vampire” characters and place them around your level.
If you feel really frisky, you can even change their hair and clothing color pretty easily with material swaps.

The next option to get variety is to use something like MakeHuman. You can create characters there, export to blender, and then export to Unreal in turn. Browse around here for some options: Fashion | www.makehumancommunity.org

If that somehow still isn’t good enough, then what is good enough? Maybe what you really want is the kind of modular, pluggable, standard, with many different assets, that I’ve described above. That’d be great, but we don’t have it. Building that kind of system is not realistic for an indie developer without a few hundred thousand dollars to spend on outsourcing, or a year-plus to spend on learning character modeling.

46

Epic gives a lot of stuff for free, I’m really thankful for that, but what I would really like to see them giving us is some kind of plugin or 3rd party tool that allow people like me who have no modelling skills to create at least some basic modular characters, something like Fuse but modular and with engine integration.
And who knows, maybe the community could contribute with clothes and gear that everyone can use.
It doesn’t even need to be free.