UMA - Unity Multipurpose Avatar on the Asset Store!

Right. But the other issue is "What is UMA supposed to be?" I think we need to look at the free package as a set of tools, a framework that developers can use to make cool stuff. Think of it like .NET. It's not so much a product in itself, as an enabler. Do we need a Pro version of .NET?

I think there's a synergistic relationship between the success of UMA and the success of its 3rd party developers. It's in all of their best interests to make UMA as powerful and flexible as possible. While conflicts of interest may definitely arise, Joen, as an example, isn't the only one calling the shots regarding UMA's core feature set. He is in a good position to exploit any added functionality when developing plugins for it though, so I think there's plenty of incentive to keep doing both. These conflicts of interest happen every time a consortium of different companies get together to form standards for anything. They all hope to profit from their efforts, and this drives and shapes the standards that emerge. It might not be pretty, but it's how the world works.

I see your point here, but it seems this is an age old problem that goes way beyond UMA. Where do you draw the lines though? Every time I get a new version of Windows, 3ds Max, Unity, UMA, or one of its add-on packs, compatibility issues are going to raise their ugly head, not to mention graphic cards and other hardware. So do we get Microsoft, Autodesk, and Unity Technologies to make a Pro Game Maker package that's all been tested and certified to work together on a specific hardware platform? Not likely. And even if they did, what about the people using Linux, or OSX, or Blender? At some point they all say "This is our core product. Our responsibility ends here." I think that's what the UMA Steering Group have done. We can suggest additions and changes, but it's ultimately their call.

 

UMA Pro isn't happening unless someone decides to pick it up and make a Pro version similar to what's been done with the dozens of paid variants of Linux available and then you'll have splinter add-ons, for various 'Pro' versions of UMA, for various 'Free' versions of UMA, ... well you can see where that would go for the efforts of most.

 

Don't be so sure. The comparison to Linux Distributions isn't quite right - a closer approximation is to the Linux Kernel, which is indeed centrally maintained by a closely-knit group. Likewise, the UMA Steering Group's purpose is to, as the name suggests, steer the development of UMA's core.

A UMA Pro version could simply be a high-powered version of that core, which encompasses improvements to core elements (such as the multi-threaded optimizations provided by UMA Power Tools). It would be beneficial to both end-users, and contributing developers (such as UnLogick).

 

Hi there!
Ohhh try out Zbrush QRemesher, it works really well together with Marvelous Designer! =D

 

Hi there! I didn´t provided a tutorial on using your own base meshes exactly because this is a quite complex process, specially because you need to proper handle skinning data
Feel free to contact me directly at skype to get help: one.fernando
Are you using 3dmax? Hard to keep track what each of you guys use.
Cheers!

 

Guys, just to give my position over all the talk about UMA PRO and so on...

Without Joen help and feedback, atlas generation on indie would be 20x slower and there would be no possibility to add extra bones dynamically any time soon; also, we wouldn´t have UMA integrated directly with Mecanim avatar API, and each UMAData would be filled with ugly LateUpdate calls to correct bones... =)
Joen had been literally sacrificing a LOT of his time on UMA those last months and didn´t expected anything in return.

I hope you guys understand that any updates and features we include on UMA core are done on our free time and we don´t earn any % money from general UMA category asset sales.

That being said, I´m asked everyday from a ton of devs why I proposed UMA to be free and open source being sponsored by Unity. UMA wouldn´t be possible if it was a close product, if it was only on my hands. This framework is alive, it evolves and it improves based on each user spending time and sharing experience.
Also, consider that we, as developer, artists, programmers or anything else, are also improving and investing extra time to share our progress to UMA framework and assets themselves.

I sincerely hope UMA can help a lot of people creating games with top quality avatars that we would only see on very expensive games, but I also hope UMA can make it possible for artists to share and sell amazing content that couldn´t otherside be used by different avatars.

Last but not least, have fun with UMA

 

That worked perfectly - thanks!

 

While I fully understand this, and completely appreciate your and Joen's efforts, what it really doesn't give me is much of an idea about where the UMA core is going. Maybe what we see is what we get, and if that's the case then I understand and I applaud your work. But personally, I feel aelrhermoul has hit the nail on the head with his concerns and perhaps even his potential solution. I couldn't have put it any better myself.

So, what I'd really like to know is: Where does UMA go from here? What does the UMA Steering Group have in mind, and is there a roadmap? Personally, I feel these are questions that anyone adopting UMA either as a content creation platform or as a game developer will have.

 

No problem =)
But keep in mind it´s a good strategy to have your own copy of the prefabs being used. Otherside everytime you update UMA pack, those changes will be overwritten.

 

I can only speak for myself, but I would say where UMA goes from here depends on UMA Community. You´re now ~5k users, I don´t expect what I say to become a rule =)
I´ll spend most of my free time providing integration of the super high quality 3d scans from http://www.triplegangers.com/ to UMA, take a look on their commercial prices to understand the amount of time and money I´m investing to provide those for UMA

 

You have excellent male female avatars available to you via UMA, perhaps you should ask yourself where to you want to go with UMA. That was the intent, not to have Joen, Fernando, and some Steering Group sitting here taking requests. They don't have those resources.

Perhaps, some good code and good art work will be adapted and integrated by the Steering Group if permission is granted by the originator but the interest is to have a community do this independently or to join the Skype group and figure out how to contribute. And nothing is stopping anyone from contributing Assets to the Asset Store independently or so...I have interests in contributing myself but haven't got far.

 

So can we use ZBrush QRemesher to increase your base mesh details? Can you provide a video guide of showing how to use QRemesher to increase the detail of your base meshes e.g. athena so we can have more mesh detail inside UMA. The process of taking it into UMA needs to be part of the video to show the workflow properly.

 

As I´ve said before, it´s more a matter of correctly setting skinning data than actually increasing poly count to get a high poly version of base meshes
I especially like the potential of working with Marvelous Designer + Zbrush Qremesher while it´s not possible exporting cloth meshes as quads.
Here´s a good video explaining the process:

 

Please consider how much work writing a bone baking tool really is. If you try to google for it the only code snippets you'll get is code that uses a skeleton to transform from magical values you get from your 3D tools into a final mesh for rendering. I had to sit down and figure out all the math that is at the core of any 3D tool with practically no information to go on. To call this monetizing on speed of rendering is far fledged.

The reason I decided to make a threaded generator was because the bone baking was too slow! Profiling showed that it took more than 100 ms causing a loss of performance that the beta group quite frankly told me was unacceptable compared to the 15 ms of the Default Mesh Combiner and 7 ms of the Legacy Mesh Combiner.

If you use the Legacy Mesh Combiners and use Unity Pro render textures having a threaded generator would actually slow you down! You might get an insignificant performance boost if you made the Default Mesh Combiner work in a thread but nothing worth the complexity of adding a worker thread!

You may not intend this as a personal attack, but I do feel attacked by this statement.
So now I make the decision to use my time writing this post, time I could have spent improving UMA? Seriously is all my time suddenly UMA's property and every action I do for myself is considered stolen from UMA?

Months ago I extended UMA's original serialization to go from text strings on the UMAData in the inspector to using assets (still with the same text strings inside), and I did it in such a way that you can write your own formats and they will work seamlessly with the UMA Dynamic Avatar. This is a typical example of the kind of contributions I've been doing. Adding structure and making UMA extendable.

When someone recently mentioned sending recipes over network for his mmo I remembered all the talks we had in the beta group about the topic and sat down and did a recipes pack. I extended the UMA framework to make the Libraries work on hashes instead of strings at the same time. This means a performance benefits to all and it made it easier for me to eliminate the costly strings from the serialization.

In honesty is this crucial to using UMA? no.
Did I write in my product descriptions that you desperately need my packs? of course (that's marketing 101)
Is this monetising? yes (and that should be ok).
Would you even have thought twice about it if I wasn't a UMA contributor?
Do you really believe that because I contribute I should be at a disadvantage?

The UMA core is not about writing optimal networking code, or the leanest way to store recipes!
So let's make some cool characters and have some fun!

 

UnLogick, if I have offended you - I truly apologize. My intention was only to bring some constructive criticism to the table, and raise what I think are legitimate concerns. Your work is obviously invaluable, highly appreciated (more than I express through mere words on a message board) and a blessing to the community. We really wouldn't be talking about a non-free Pro product if that weren't the case!

Let me reiterate once again the context of my remark: the idea here is not that all that you do should be free or for UMA, quite the contrary. The whole point of this discourse is to consolidate core improvements to UMA which you would like to monetize in an official UMA Pro pack that we can count on being updated in lock-step and collaboration with UMA Free.

With that said, I don't wish to drag this issue any longer. I think all ideas, views, and perspectives have been expressed, and I'll leave it up to you ladies and gentlemen of the UMA Steering group to make the relevant decisions. I only hope that you see my argument as not an attack on the amazing work done at UMA, but rather high praise - as it's clear several of us at the very least would be more than happy to pay for a Pro version product.

 

UnLogick ~ I'm truly sorry that i began this discussion into the way the assets are being established for UMA. I myself respect the dedication and the talent you and fernando have to bring such a wonderful product to unity. I was merely stressing concerns over how the stuff should be packaged not the monetary aspects which you so solely deserve for this work. I think enough has been said on this subject now ! I wish UMA utmost success

 

I've posted my attempt at putting together a bumped specular shader for skyshop for the uma character.

Here is the code snippets i did which gives me this result.

Please more experienced shader guru's correct my shader code if it needs work

Code (csharp):

1.  
2.         "Marmoset Bumped Specular IBM.shader" file and append this and rename it to your liking ...
3.  
4.     CGPROGRAM
5.         #pragma exclude_renderers flash
6.        
7.         sampler2D _MainTex;
8.         sampler2D _BumpMap;
9.         fixed4 _Color;
10.         half _Shininess;
11.         half _SpecInt;
12.        
13.         struct Input {
14.             float2 uv_MainTex;
15.             float2 uv_BumpMap;
16.         };
17.        
18.          void surf (Input IN, inout SurfaceOutput o)
19.          {
20.             fixed4 tex = tex2D(_MainTex, IN.uv_MainTex);
21.             fixed4 bump = tex2D(_BumpMap, IN.uv_BumpMap);
22.             o.Albedo = tex.rgb * _Color.rgb;
23.             o.Gloss = bump.r * _Shininess;
24.             o.Specular = bump.b * _SpecInt;
25.             fixed3 myNormal;
26.             myNormal.xy = bump.wy * 2 - 1;
27.             myNormal.z = sqrt(1 - myNormal.x*myNormal.x - myNormal.y * myNormal.y);
28.             o.Normal = myNormal;
29.         }
30.      
31.          ENDCG
32.  

The skypshop game output i get is this



The UMAGenerator code to change the shader is like this

Code (csharp):

1.  
2.     // Character creation instance entry point
3.     public override void UMAReady()
4.     {
5.         base.UMAReady();
6.  
7.         if(umaData)
8.         {
9.             Debug.Log ("Character Generated");
10.  
11.             // UMA Diffuse (RGB) Overlay (A)
12.             Texture mainTex = umaData.myRenderer.materials[0].mainTexture;
13.  
14.             // UMA Packed Normal Map, Normal (GA), Specular (R), Gloss (B)
15.             Texture bumpTex = umaData.myRenderer.materials[0].GetTexture ("_BumpTex");
16.  
17.             // Automatically copies over the _MainTex
18.             umaData.myRenderer.materials[0].shader  = Shader.Find ("Marmoset/Vertex Color/UMA Bumped Specular IBL");
19.  
20.             umaData.myRenderer.materials[0].SetTexture("_SpecTex",bumpTex);
21.             umaData.myRenderer.materials[0].SetTexture("_BumpMap",bumpTex);
22.         }
23.     }
24.  

Issues

I'm getting red eye or a pinky eye when looking up close to the character.

I noticed that the UMA shaders switch the R and G channels in their shaders why? I though the manual showed them the opposite way around.

Any Ideas to improve?

 

A question about this: Since it would be important to match your base skinning data as closely as possible to retain compatibilty, would it not be possible to "simply" subdivide/alter the base mesh and then transfer the skinning data in the same way it is done for clothes? Whatever errors and changes would likely be few and simple to correct?

Also, I would guess the various overlays would have to match in order for realtime assembly to work (i.e., body, ears etc.) so any increase in detail on one would have to be mirrored on matching body parts. Correct?

 

Thank guys, apologies accepted.

Your feedback is appreciated and we will talk about this in the steering group.

 

Joen you are doing great work. Any chance you could make a simple export of one of your prefab(Power Tools) avatars I still can't get the Unity export to produce all the parts. Thanks.

 

Hi there!
Simply subdividing the mesh and interpolating skinning data would result in bad quality. Most places where we should have curves would result in curves + straight angles.
UV wouldn´t be a problem though I´m managing to get it working quite well even on lower poly version.

 

@Fernando, thank you very much for the offer. I will definitely be in touch as I think the game I am currently working will eventually need to add some other base meshes to the system. Again, thank you!

--Kevin

 

you´re welcome =)

 

Hey! Got your PM.
->I noticed that the UMA shaders switch the R and G channels in their shaders why?
You might be talking about:
o.Gloss = bump.r * _Specular;
o.Specular = bump.b * _Gloss;
I´m not entirely sure why this happens, but Unity surface shader definition for specular and gloss are swaped, here´s a reference of what each of those should result:

 

Well this one is REALLY complete (complete with animation fbx, shader, controller, scripts, etc) you can import this package into a completely empty project and it will work!
http://glscene.com/unity/hugo_complete.unitypackage

This one is what I would normally call complete, but you still need a project with the animation fbx , shader, controller and scripts. So this one will work if you import it into a UMA project.
http://glscene.com/unity/hugo.unitypackage

 

Is there any delegate event on UMAGenerator() such that a game at start-up can be informed that the character is generated. I'm not talking about having to press a button or something and doing a test for this, but more like an event?

I've been accomplishing this via, however is there another way? or if not a delegate event for OnCharacterCreated(), OnCharacterRemoved() would be nice additions to UMAGenerator() as standard.

// Character creation instance entry point

public override void UMAReady()

{

base.UMAReady();



if(umaData)

{

 

Not on the UMAGenerator, but there is an event on the individual UMAData. Though there is no event fired on Removal. It wouldn't be a big deal to add another event though.

 

Given UMA is an open source solution might be nice to have a little document show FAQ Coding questions for people to get up and started fast. Another steering group question.

 

Looking at UMAData code there is a public event Action<UMAData> OnUpdated;

I can't see the character creation event?

 

Please sign into the still quite empty UMA wiki and create a FAQ page.

http://serendipity.unity3d.com/mediawiki/index.php/Main_Page

When the wiki reaches a more mature stage it will get a more formal url.

 

That is correct, I meant the OnUpdated event. I'll be adding the following update to the UMAData.

Code (csharp):

1.  
2.         [Obsolete("UMAData.OnUpdated is deprecated and scheduled for removal, please use OnCharacterUpdated instead.", false)]
3.         public event Action<UMAData> OnUpdated { add { OnCharacterUpdated += value; } remove { OnCharacterUpdated -= value; } }
4.         public event Action<UMAData> OnCharacterUpdated;
5.         public event Action<UMAData> OnCharacterCreated;
6.         public event Action<UMAData> OnCharacterDestroyed;
7.  

OnCharacterCreated will be triggered only the first time a UMA character is successfully built! If the build fails for one or more reasons you will only receive a call to OnCharacterUpdated where the UMAData.cancelled is true.
OnCharacterDestroyed will be triggered only on UMAData that have been successfully built.
OnCharacterUpdated will be triggered everytime a UMA character is built/rebuilt.
The old event OnUpdated is deprecated and will be deleted at some point in the future.

I can't say exactly when this update will be available on the asset store.

 

Many thanks for looking into this aspect.

 

Hi!
I am not giving up quite yet... ;-)
Would the following workflow really result in bad deformations?

- Your Zbrush sculpt -> ZRemesher to higher resolution than the current base mesh
- transfer bone weights from default base mesh -> new mesh
- (smooth resulting weight map to ensure smooth transition across new polygons)

....?
Since the feature I have in mind (portraits) is relatively low priority I am trying to come up with a pipeline to the desired result that does not require too much time.

Again, thanks for your input!

 

This should work
You´re using 3d max?

 

Excellent!
We are primarily using C4D but will probably use blender for the final skinning transfer step and adjust as that is what you used to get the rig into Unity. That way we minimize the risk of running into problems with anything C4D might do to the rig on export.
We have had some success with rigging and animating -> Mecanim in C4D but I am not trusting it fully quite yet. In general we will likely transition from C4D to something else as C4D - while extremely good for rendered animation and effects - is quite lacking when it comes to game development functionality.

I will give this a whirl when I have some free time for it. Will let you know how it works out!

 

Ok! Let me know if you run on any problems.
Cheers!

 

Btw, here´s my latest changes on the new scan I´m working on =)
I´ll be able to get back to this one at night.

 

This one is what I would normally call complete, but you still need a project with the animation fbx , shader, controller and scripts. So this one will work if you import it into a UMA project.
http://glscene.com/unity/hugo.unitypackage[/QUOTE]
Joen, thanks a lot. Worked great.

 

So, I've created a few basic cloths like shirts and armguards using the base mesh, but when I tried to use another blender file, even after weight painting and posing it and exporting it as correctly as possible, something weird happens when I attempt to create the slot via the Material Builder and my mesh goes from this:

to this:

Any idea what I might do to solve this?

 

These scans. Are they different base meshes supporting same clothing. Or same base mesh different skin textures?

 

I am probably revealing my ignorance here, so apologies for that.
However, I would really appreciate if someone with more experience with UMA would care to shed some more light on the process of importing clothes meshes and new base meshes into UMA.

In particular I am a bit confused by the fact that the UMA rig is in an A-pose whereas Unity wants a T-pose. When importing the rig into Unity it throws a "character not in T-pose" when configuring the avatar. I am assuming the avatar does need to be configured when importing new assets and animations for use with UMA?
Perhaps related to this I noticed a folder in the project called T-poses. Is this something to be used in conjunction with defining the avatar or is it totally unrelated?

Thanks in advance for any help!

 

Ok, so as you can tell form the attached image, importing clothes to UMA did not work that well on the first attempt.
Here is what I did:
- Used the male_unified fbx as the source of skinning info
- Succesfully transferred the weights in C4D
- Exported FBX as per the instructions
- Followed the Youtube tutorial on creating a new slot
- Had a laugh when seeing the jacket flailing around on the floor

What I have checked this far is:
- scale and rotation, and bone placement
- I have compared the asset to the MaleJeans one and the imported rig seems to match. Bone positions match etc.
- That the fbx avatar exported from C4D untouched will succesfully import into Unity

One interesting thing too is the effect on some characters' eyes. All of the avatars get the jacket, and a few of those get weird "hollow" eyes. Naturally I snapped the pic when they turned away.... If important I will snap another pic.

Anyway, I would be very thankful for any help to sort this out. Thanks!

 

Ok, I found the issue. It seems that the actual skinned meshes to use in UMA must be rotated before skinning. The mesh coordinate system must be offset so the skinned mesh is imported with a rotation of (xyz) 0,-90.180.

As you can see below, correcting this fixed the problem. A bit hard to see as we just used the torso overlay, but the jacket is there and positioned correctly (although the torso is intersecting it - have to adjust that).

However, please note the eyes in the enclosed pic. If the jacket is included, the eyes on a few of the avatars go weird. I have not been able to find a common denominator for the problem. Most normal characters look ok (a few do not) and all pignosed or elf-eared characters have their eyes hollowed out.
The rig used for the jacket was left untouched in C4D and I cannot understand why the jacket mesh would interfere with the eyes...?

Also, an unrelated question:
- If we want to use a completely different shader and texture for clothing (without relying on overlays etc.) can we? If I did not include an overlaylist UMA threw an error.

Again, thanks in advance!

 

Sorry for spamming the thread.
However, the eyes problem above is solved. The reason was that I (stupidly) overlooked the fact that more slots had been used since the tutorial, so using slot 7 ended up overwriting a part of the face apparently.

I am leaving the posts as they are as hopefully they might help someone else running into the same problems.

Anyway, the question about materials remains. I would prefer if we could separate the clothes entirely from the UMA atlas generation in order to use untouched materials and our own shaders. The tutorial seems to imply that this is possible, but I have yet to figure out how.

 

Also while working with the wings and horns pack I've become aware we have introduced one ore more bugs. The wings no longer work with any of the mesh combiners. It might take me a couple of days to get additional bones working again.

To avoid distractions I'll be less active on the forum chats over the weekend.

Cheers,
Joen Joensen, UnLogick

 

Is it my impression but are the community finding creating content for the UMA character too difficult? Is there anything that can be done to the workflow, or help in tutorial sense etc etc to make this process easier.

Maya, 3DMax tutorial guides to perhaps. How does the creation process of clothing compare to how Second Life do it? Whats easier and can we learn from such a prolific clothes MMO?

 

Does uma work with 4.2.2? 4.3.3 is still fairly bad. I have both versions installed.

 

[video=youtube_share;StZRJwf5PP4]http://youtu.be/StZRJwf5PP4





Import process explained for 3dsmax in 10 mins...

UMA has a lot of great new functionality and in order to take advantage of that new functionality you have to do a few extra things. When you're going to create content for any extension you have to take into account many things to make sure it works. UMA is new, has bugs, and not much clothing content is floating around for reference material. The extensions are also buggy, but this is all really just an expected side effect of being such a new extension. Core changes had to be made and with the asset store approval process the UMA extensions are sometimes much behind which causes a lot of discrepancies.

After you understand the sequence and steps for importing content its not really much more complicated than creating content like you would normally anyway. Good content worth buying takes time to develop, I assume there are some people making this and I've made some mockup slot overly assets for coders to use for testing their scripts. The process isn't the problem, it's cumbersome sometimes and especially at the beginning but it's not difficult once you understand what you can do and what you cannot do.

 

Having separated materials forces extra draw calls, but yes, each different shader will require it´s own texture atlas. take a look on female avatar eyeLashes and you will notice it has it´s own texture separated from the others, being the only slot using the twoSidedTransparent shader.
I would recommend using atlas and reduced drawCalls optimization as much as possible, but you´re free to use as many shaders/textures you need.
Cheers

 

I don't think it's too hard, just as LaneFox says, it's new. For the first few times the process isn't obvious(I have watched every single tutorial I could found before I did anything), but after that it's usually goes as expected. Usually, hehe.
We need to pratice, that's all.