Universal Modular Entity Construction System for UDK

[TechLord]

Universal Modular Entity Construction Hierarchical Sets or Specifications (UMECHS) is a hierarchical template-driven assembly system for fabricating a variety of 3D entities from modular interchangeable Part sets. Inspired by customization systems found in Massive Multiplayer Online Games an other genres, UMECHS aims to provide a standard assembly system that supports real-time customization for all entities categories: character & creature head/body, vehicle/crafts/machines, melee weapons, projectile weapons, architecture, structures, furniture, plants, hybrids. With UMECHS, one can assemble a library of pre-fabricated Entities and dis-assemble Entities for destruction/dismemberment sequences. The UMECHS Unrealscript Module is being developed to support all of these features and will be freely released to the UDK community, supplied with at least two Parts Pack per entity category. My goal is to develop and evolve UMECHS into a Full-Service System that contains: Documented Specifications, Templates and Demo Assets Parts Pack, UDK Module Unrealscript/Kismet, Knowledgebase/Repository, and Marketplace.

Inspirations:

Player perspective. I've played several games that allow the player to customize characters, vehicles, robots, weapons, etc. Customizing your stuff in-game can be FUN in itself. Customization is now a huge part in giving games some depth. I've played with the Spore Creature Creator for hours on end, just creating all sorts of odd looking creatures. In some aspects, this is sort of the modern day version of yesteryear's construction toys and I enjoyed TinkerToys, Lincoln Logs, Legos in my youth.

Game Developer perspective. Unique game content can be expensive in both time and money. Luckily for me, Game Devs have an option of purchasing Character Model Packs. Model Packs are great, offering some decent quality for an affordable price. However, I have an issue with using the current Model Pack paradigm as they're not exclusive and they're not modular. So every Dev who purchases the pack is using exactly the same models. Obviously, this is not desired for unique look and feel and modifying textures isn't enough in most cases.

A few years back, I had an ephiphany, realizing that a solution to the delima would be with modular entities. Thus, I set out to devise a standardized modular assembly system and supply it with parts. The system can be used to player avatar systems, pre-assemble complex entities, and share/reuse assets among multiple Game Modules with a greater entity variance. With Parts-Packs, we can populate our game worlds with plenty of unique monsters, items, etc.

Benefits:

There are obvious benefits to use UMECHS in games for code-centric Game Devs like myself, but, what about the Artists? One advantage to creating assets for UMECHS, is that Artists can focus on a smaller specific type of part, refining the details. Some Artist may specialize in producing specific parts (ie: I stink at making a full-bodied 3D models, but, I can make prettiest little 3D foot models you ever seen). With UMECHS, the Artist is free to do so. In my opinion, the reduced scope will result in rapid production of high quality parts.

Second Advantage for Artist is improved scalability on packaging and re-selling Assets for UMECHS. Sell by Part, Parts Packs, Pre Assembled Entity Kits, and Entity Collections. Categorize by Genre and Art-style. I believe this will translate into improved pricing for Game Devs.

Requirements:

1. Support many classes of entities: characters/faces, armor, complex and simple weapons, vehicles, machines, architecture, structures, furniture props, trees.
2. Take in consideration to: design tools, FBX pipeline, and UnrealEd, UE3 automated and procedural animation, physics/collision detection, and rendering optimizations (Level of Detail).
3. Provide Unrealscript interfaces to procedurally assemble and animate entities at real-time: scale and orientate mesh parts, skeletons, and sockets, adjust target morphs, swap textures and shaders, wire up UI interfaces for entity customization, etc.
4. Self-contained module, plug-able into any game framework developed with UDK.

Starting Point:

My first step is to define Specification Standards such as a UMECHS Naming Convention for Skeletons (Bones/Joints), AnimTrees, MorphTargets, SkelControl Nodes, Sockets, Textures, Materials and Shaders. I'm seeking advice of an experienced 3D Artist/Animators in this area. The Name convention would be generic and multipurpose to take in consideration skeleton for facial animations, multi-ped creature, vehicles, animation blend trees, procedural animation, and physics/collision.

Second step, would be defining physical standards of skeletalmesh and static mesh parts, joint positions, morphtargets, sockets, and modular textures, and auxiliary data formats. Once the standards are defined, creating models to use the standard would literally be a snap, and new animations, textures, and shaders sets could be added/removed easily. The Standard will be documented and published for public reference.

References: Modular character pawns, Skeletal Mesh Pipeline, Sockets, MorphTargets, SkelControllers

Concerns:

There are concerns in dealing with organic entities, to produce a smooth-bodied character after snapping on limbs? I've dubbed this issue, the Action-Figure effect. I had a vague ideas for smoothing out limb connections (seams and joints) to produce uni-body characters. To tackle these issues, I found inspirations in playing with the Spore Creature Creator. Solution: install slots and target-morphs inside all parts, use a mult-ipart base mesh called the 'body' to attach sub-parts too. Texture blending may also play a role in smoothing seams and attachment.

I believe UMECHS will be beneficial to the entire UDK Community. Any and all advice will be greatly appreciated.

References:

UDK 3D Art Pipelines Naming Conventions
UDK Custom Characters

[slowJusko]

Would be nice to have access to the mesh compositing, but hey, what can you do

Also, I could have used something like this a year ago.

It is an interesting idea, but getting a decent model maker and animator to help out (for free) is lilke getting blood out of a stone

[TechLord]

Hi slowJusko, Thanks for reading, its a long post.

I agree, Mesh composition would be nice, but, flexibility is a fair trade off. Its briefly mentioned in the Modular Character Pawn Gem:

Compositing meshes (Licensees only) - This method composites the mesh to generate a single mesh from multiple meshes.'

Benefits are obvious, as the multi-skel method for MCP's requires an additional draw, but, allows run time changes to the skeletal mesh. I'll actively seek out ways to include optimization in the UMECHS Standards. We get a form of mesh compositing at design-time in UnrealEd. The UMECHS Pipeline adopts the multi-mesh approach to creating the 'body' base mesh in which parts are attach too. This method is ideal for UMECHS, making it possible to LOD individual parts and export them for use in modular entities. There is no performance penalty because the parts are combined upon import into UnrealEd.

It is an interesting idea, but getting a decent model maker and animator to help out (for free) is lilke getting blood out of a stone

I aim to change their perspective. UDK provides everything you need to create amazing games with an affordable license ($99). One problem, NO content! There is a significant pricing imbalance between the Engine and 3D content. I'm a code-centric game developer with little or no budget, not a 3D Artist/Animator, so, I'm at a severe disadvantage. There are multitudes of UDK game developers like me. UMECHS fills in multiple gaps, providing a powerful Content System for Games Developers supplied with Content, and a high-demand Market Place for Artist to LEASE (See: Leasing Beats) Parts , Parts Packs, Pre Assembled Entity Kits, and Entity Collections (Set of Pre Assembled Entities) at affordable pricing. It creates content development interoperability and balances the relationship between the demand and the supply. UMECHS is a game *development* changer!

[darthviper107]

I'm not exactly sure what you're trying to make, UDK of course supports the ability to connect modular pieces for things like characters and meshes so I don't know whether you're talking about creating a system like that since there's one already. Of course the issue is that there's no standard if you wanted to create a library of parts that could be switched out. I understand the library part of it but for games it really seems like there's a limited amount of content that could possibly be useful for a good number of people, so it probably wouldn't be worth creating a bunch of different mesh choices that aren't going to be used much if at all.

Also--you'll find that there are many cases where you can't actually blend a mesh to another mesh, like say a hand to an arm, the issue is the smoothing. You can get rid of the visible seam if you know what the two meshes are, but if the two meshes can be switched out then there's no way to make them match automatically. The way that games avoid this is by putting seems at areas that are already going to have a seam, like at the cuff of a sleeve or something where it's always going to have a line there, but you can't do that for everything. I believe Spore has a method of actually joining geometry much in a way you would use a 3D program, which would be difficult to add and I'm not sure UDK is capable.

[TechLord]

Hi darthviper107

Thanks for reading and replying.

I'm not exactly sure what you're trying to make, UDK of course supports the ability to connect modular pieces for things like characters and meshes so I don't know whether you're talking about creating a system like that since there's one already. Of course the issue is that there's no standard if you wanted to create a library of parts that could be switched out.

We're developing that standard and one of its purposes is to address the matching and blending issues. Interoperability my friend.

I understand the library part of it but for games it really seems like there's a limited amount of content that could possibly be useful for a good number of people, so it probably wouldn't be worth creating a bunch of different mesh choices that aren't going to be used much if at all.

I truly believe the Sporepedia and plethora of RPGs suggest there's is no limit to the Players imagination nor the Game Creators. Most games are limited by their content, lets remove that limitation.

[darthviper107]

I'm not sure that's the case, even if you're doing a fantasy style game I don't know that you would want to use assets from some kind of online library. People would at least call you out for using something like that on main characters so it would mostly only be useful for minor things like NPC's and enemies. And that's only if you like the style of the assets that are available.

[philbrown]

I'm not sure that's the case, even if you're doing a fantasy style game I don't know that you would want to use assets from some kind of online library. People would at least call you out for using something like that on main characters

Why?
To me, it would depend on

1. how good they looked
2. how much I paid for it.
(in other words, if I'm paying $$$, but they're getting their content for free, I'd be alittle peeved)

Other than that... people already pay ludicrous bux to see the same actors in many different movies. Why would they get all huffy about seeing the same "virtual actors" reused... especially if it brought down the cost to play?

Along those lines... I think there's far too many "completely new location/scene/xyz!!!" games there. It was worth it while technology and graphics kept improving. but we're kinda reaching the "retina screen" limit. ie: you cant improve the quality of the image much any more.
In my opinion, it would be nice to see 4 quality titles in 1 year, all reusing the same basic assets, rather than see 1 quality title over 2 or 3 years.

I dig story, over "new for newness' sake"

[darthviper107]

A "quality" title would not be reusing assets. In some cases you should certainly reuse stuff, like consider how many times New York City has been in a movie or game, having to redo it multiple times would be a waste. But for something like fantasy, where a big draw is the particular art design it would not be a good idea. As a customer if I noticed reused assets like that I would think you were lazy and cheap and don't care about your own project. Real-world stuff that's supposed to be the same no matter what game it is is fine for an online library, but something that's supposed to be a unique vision of fantasy or sci-fi is not OK to reuse assets.

[TechLord]

Along those lines... I think there's far too many "completely new location/scene/xyz!!!" games there. It was worth it while technology and graphics kept improving. but we're kinda reaching the "retina screen" limit. ie: you cant improve the quality of the image much any more.
In my opinion, it would be nice to see 4 quality titles in 1 year, all reusing the same basic assets, rather than see 1 quality title over 2 or 3 years.

Hi philbrown. Thanks for reading and replying.

I hope we haven't reached a "retina screen" limit as I'm true believer in that there is always room for improvement. Technology continues to improve and we're on the brink of a paradigm shift in how graphical game content is generated. The Spore Creature Creator is a glimpse into the future of procedurally generated modular content. I'm not saying this technology will replace 3D Artists/Animators, but, I do believe we'll see more advanced forms of automated content creation to reduce production cost and increase production speed. Even UnrealEngine3 offers a Procedural Buildings System.

I predict within the next 3 years, we'll see:

• Real-time photo-realistic procedurally generated modular content.
• Computer aided content creation.
• Semi-procedural generated content systems.

UMECHS is a paradigm shift in how content is manually generated for UDK Games. It provides Content interoperability between multiple Artists and Game Devs. The UMECHS Standard provides an Universal Assembly Convention for Modular Asset Creation. The UMECHS Module (unrealscript middleware) provides a Modular Content Management based on the UMECHS Standard to assemble/disassemble, swap parts, procedurally animate, swap and modify textures, and much more in real-time with considerations to networking. Together, UMECHS is a 3-in-1 solution that doesn't require reinvention of the wheel. This is a time cost savings for all parties:

The 3-in-1 Solution:

• Real-time Player/NPC/Entity customization.
• Populating game worlds with catalogs of Pre-fab Creatures and Props.
• Destruction/dismemberment Sequences.

A "quality" title would not be reusing assets.

With UMECHS, Game Developers have the option to use available Parts, create new 'exclusive' Parts, and mix n' match between them. 3D Artist/Animators have the option author, package, and distribute parts with a specific theme and art-style. UMECHS Entity's quality and uniqueness is the sum of its parts (mesh, animation, textures, materials, shader) used to assemble and animate it. A typical bipedal base mesh comprised of a minimum of 7 segments with a minimum of 2 unique sets of segments will provide 49 unique base mesh combinations (7 ^ 2 = 49). This number exponentially increases with the addition of segments, attachments, textures, materials, shaders, and animations.

UMECHS is a Modular Content System and reusable interchangeable Parts is a good thing for a 3D Artist. Modularity makes is possible to resell assets to multiple developers. What isn't a good thing for a 3D Artist, is creating exclusive works-of-art that are trapped in a project that is never released. There is an extremely high percentage of projects that fail to release. Artist are aware of this fact, and favor direct payment for work rendered verses royalties. I understand that principle from a business perspective, however, from artist perspective, I want the world to see my work.

[darthviper107]

If you see a character that reuses a face it doesn't matter what the rest of the body looks like, the player will remember that they've seen that face in another game, or the armor, or whatever it is. Just like in many MMO games where it's disappointing to see people using the same head or armor as you, and that happens even with games that have a huge amount of customization content.

[TechLord]

If you see a character that reuses a face it doesn't matter what the rest of the body looks like, the player will remember that they've seen that face in another game, or the armor, or whatever it is. Just like in many MMO games where it's disappointing to see people using the same head or armor as you, and that happens even with games that have a huge amount of customization content.

I can agree and thats a crucial point. Then we must ensure UMECHS supports plenty of low level detail customization options for face, armor, and other entity classes and parts that players focus on. This includes facial animation. Dragon Age did a real good job providing plenty of facial detail customization. Supporting such detail in Weapons is also of particular interest to me. Your insight is valued.

Firearm Entity Class

Firearm Modelled Parts (simulated)

Firearm Weapon (TBD)

• Reference: Loosely based on 3P RocketLauncher Skeletal Mesh. LOD suitable for First-Person Perspective
• Base Mesh: Reduced to Single Core component, Splitting sections into Accessories and HotSwaps: See Modular Firearm

Vehicle (TBD)

• Reference: Loosely based on Scorpion Skeletal Mesh/Rig
• Base Mesh split into sections to separate Chassis, Suspension System, Wheels
• Static Attachments
  • Hood, Front Left/Right Fender, Rear Left/Right Fender, Front/Bump Bumper, Left/Right Side Skirt, Spoiler
  
  
• Dynamic Attachments
  • Trans-axles, Wheels
  
  
• Textures: Decals, Emissive Light Decals, Body Graphics

[mcfarrel]

A "quality" title would not be reusing assets. In some cases you should certainly reuse stuff, like consider how many times New York City has been in a movie or game, having to redo it multiple times would be a waste. But for something like fantasy, where a big draw is the particular art design it would not be a good idea. As a customer if I noticed reused assets like that I would think you were lazy and cheap and don't care about your own project. Real-world stuff that's supposed to be the same no matter what game it is is fine for an online library, but something that's supposed to be a unique vision of fantasy or sci-fi is not OK to reuse assets.

If some guys here bother to look at delev proces of variety of commercial games they might have found some "prefab making". So as a part of one mod or game it is absolutely normal and even the pros do this.
But on the other hand this feature might be questionalble in some situations (like above mentioned).

If this feature should be part of one game/mod it is OK to use prefabs to make variability of Your models (changing clothes each game, so the AI actors looks more living), but using widely across the wide variety of games this might be questinable.

Anyway the idea of creating convention and logic for interchangable parts of models is OK, but problematic part might come deciding where to use this and what exactly use...

[TechLord]

Anyway the idea of creating convention and logic for interchangable parts of models is OK, but problematic part might come deciding where to use this and what exactly use...

Use it for all entities (with exceptions to terrain and simple single mesh objects). UMECHS can offer flexibility for content in a game that doesn't require player customization, large assortment of creatures/props, or destruction/dismemberment sequences. The UMECHS Module (unrealscript middleware) provides mechanisms to assemble/disassemble, swap parts, procedurally animate, swap and modify textures, and much more in real-time.

TheGameCreators, Garage Games, Unity, many other Game Development Communities understand the demand for content and meet this need with Character /Content Model Packs. UMECHS can be the Modular Content Pack Solution for UDK. It redefines the concept of Content Packs, being designed specifically with the intention of providing unique characters for UDK Games.

[mcfarrel]

Use it for all entities...

That is correct. I have in mind there are differeincies between using same set of model parts creating modularity of objects in one mod/game and in several games (using same game parts). And more likely to be legal issue (or cuase legal issues) then technical issues...

[TechLord]

That is correct. I have in mind there are differeincies between using same set of model parts creating modularity of objects in one mod/game and in several games (using same game parts). And more likely to be legal issue (or cuase legal issues) then technical issues...

Generally, content packs sold to the public don't stipulate what games the content can be used in. Trying to do so will most likely reduce your profit potential. In the future UMECHS Store, our licensing terms are simple, geared toward maximizing the Artist's distribution & profitability, and the End User freedom & flexibility. If an Artists desires different Terms of Use, they obviously would not sell their parts pack through the UMECHS Store.