[Maya] How to go from A-stance to T-stance (for HumanIK rigging)
polycounter lvl 6
UPDATED: this was the original question, see my latest question at the end of the thread.
Hi all,
I have got a few burning questions about Maya Quick Rig tool (introduced in 2016):
- While it's good and all for a quick setup, I don't see a path to manually add hand skeleton or adjust existing joints. AT BEST I can play with the "guide" (orange widget) during the step-by-step setup, but they are just indicators, never the placement of actual joints.
- So does it mean real animator never use this? They always start with a default HIK skeleton and go from there?
My main concern right now is to adjust my base mesh, but as suggested in another thread, in order to adjust base mesh I need to rig it properly to see where do those deformation went wrong... hence these questions.
Hi all,
I have got a few burning questions about Maya Quick Rig tool (introduced in 2016):
- While it's good and all for a quick setup, I don't see a path to manually add hand skeleton or adjust existing joints. AT BEST I can play with the "guide" (orange widget) during the step-by-step setup, but they are just indicators, never the placement of actual joints.
- So does it mean real animator never use this? They always start with a default HIK skeleton and go from there?
My main concern right now is to adjust my base mesh, but as suggested in another thread, in order to adjust base mesh I need to rig it properly to see where do those deformation went wrong... hence these questions.
Replies
It does have its limitations of course and is similar to Mixamo's auto rig service but, for any kind of decent rig with efficient controls for a comfortable animation workflow you will definitely have to look at so,ething else. Luckily
you're using Maya and are pretty much spoilt for choice. Have you considered the recently added animation and rigging toolkit? It is now available
as a free plugin for UE4 and links natively to Maya.
Through the awesome youtube comment section (a phrase no one ever said), I finally got the official answer: no, as of Maya 2017, we still can't add fingers after using Quick Rig....
Oooooor is it?
With a plot twist, the official account mentioned that one can build a hand skeleton manually and attach it to the quick rig skeleton.
But here is the catch, AFAIK, I can't keep the Control Rig when attaching a Partial Skeleton. So I am basically working with a bare HIK skeleton, albeit with some helps on joint placement from Quick Rig tool...
http://www.animationstudios.com.au/advanced-skeleton
- Do most character artist model in T-stance, or A-stance?
I did mine in A-stance but now feel I kinda curse myself because Maya asks for T-stance when adding a HIK skeleton.
Quick Rig tool doesn't have this requirement. And I saw BaseMesh wiki page examples mostly used A-stance.
- How do artists / animators deal with this requirement to pose in T-stance?
I want to keep my character in A-stance, and I want to use Quick Rig tool with at least some adjustment to skeleton, but it seems I can't do that.
Let me show what went wrong:
1) I setup the guides for skeleton generation
2) but as you can see, the placement for knee joints are not quite right (even with my adjustment)
3) so I try to only generate a skeleton, that way I CAN move the joints
4) but as soon as I hit Create a control rig, it returns back to T-stance
So my only question at this point is: should I even use Quick Rig tool? Because of the requirement of T-stance of HumanIK, I just have to build the skeleton from scratch?
Imo the biggest problem and joke is that there is no native finger support and if youre not familiar with rigging in maya you dont get them in.
Could anyone confirm that this is the way the game industry works?
Overall you need to strike a fine balance between using clean/solid models (with proper joint placement, good proportions, and so on), and being okay with doing "throwaway" work - because this is your first attempt after all.
As for "is this the way the industry works" : No. Not two pipelines are similar, and this sort of things depends a lot on the kind of game being worked on, the tools being used, and so on. This can also highly depend on the personal preferences of the people involved. For instance modelers with no first-hand animation experience will often tend to push for models to be created in a "cool" relaxed pose because it looks better in screenshots, whereas artists with a more varied experience will understand the value of a model being in a clean "zeroed-out" state.
If anything I'd recommend you to look at the way models are made in japanese games like Final Fantasy and Metal Gear. From what I have noticed models in these series tend to be setup in a way that is very animation-friendly, even if that means that the pose is seemingly stiff.
As suggested earlier, all these issues are somewhat solved when you force yourself to work in a nonlinear way - that is to say, modeling while at the same time having a skeleton in the scene, with a few good poses keyed in. That way you can adjust the model to the skeleton as needed, and vice versa.
This is without a doubt a delicate craft, and it is great that you are being careful with all the steps. But as said, the best way to get a feel for all that is to go all the way through the pipeline to then get a feel of what can be done better. It's tricky because it is highly technical (rigging tools) but also depends on aspect that can only be learned with time (good human anatomy and stylisation).
Question about tools are one thing (using an auto rigger or not, and so on) but at the end of the day one can't work in a vacuum. What's the end goal - making a animated movie ? Putting a character in a game ? And so on. For instance if your goal is to put a character into an engine using an existing base skeleton and animations, there is no need to worry about having a complex control rig at all since the animations would already exist - therefore all you need is a skinned mesh.
- Sergi Caballer said some things about T-pose and A-pose in this interview, quoted below:
Q: No talk about model topology is complete without discussing the default pose to rig. Do you have a preference? Why?
A: This is always a tricky point. From a modeling point of view, modeling on a TPose is always easier, especially when we have to deal with modeling fingers and hands. Modeling while your piece is aligned with the world axis is always easier, but you can find ways to work with an unattached arm aligned to the world if your goal is to end with an APose character. The same idea could be applied if we’re talking about an easier way to place a character rig guide for joint placement during the rig creation. TPose is always easier, for the same reason.
From the deformation standpoint, APose is probably the most efficient way to get a neutral base for your shoulder deformation, because this is exactly the upper arm’s middle range of motion, especially if you’re only using joints for the deformation. On the other hand, if your pipeline allows you to work with PSDs (pose space deformers), TPose would be my choice because this would be the neutral mid-point for all the corrective shapes for the shoulder/upper arm area.
Another thing to keep in mind is the look or appeal that the character has when we deliver the model to rigging. TPose is always kind of a weird pose because, from a design and proportion standpoint, it is harder to appreciate the shoulders’ width, the upper arm vs. the lower arm proportions, or even the whole arm vs. the body proportions. At the end, as a modeler, you always end up deforming the area at different ranges of motion to see any weaknesses in proportions.
Related to this topic, especially for the shoulder articulation area, I always recommend the work of Brian Tindall (www.hippydrome.com) as a reference for modeling for articulation, to study pivot joint placement and how the corrective shapes help to achieve the nicest deformations, especially with their gifs where you can see the effect of using only the skin cluster vs. adding the corrective shape over the joint deformation.
(I apologise in advance: English is my second language + I am certainly new to modelling + I don't have an art background + I have quite a few questions.)
Do you mean shoulder topology? shoulder posture? shoulder joint placement?
> Overall you need to strike a fine balance between using clean/solid models (with proper joint placement, good proportions, and so on), and being okay with doing "throwaway" work - because this is your first attempt after all.
Do you mean, whichever pose I pick (A or T), I always end up having to adjust the mesh to a different pose (before rigging)?
By "throwaway", do you mean "necessary practice" or "busy work transitioning between modelling and rigging"?
True, but specifically for HumanIK, how do game company handle modelling and rigging?
I assume large companies have dedicated in-house plugins or tools, but how about indie companies? They must made some compromises (consider Inside from Playdead), I am very interested in what they compromise on.
Ultimately I want to get my character from Maya LT to Unity. I don't mind practice, I really don't. But the reason I am so focus on workflow is because I want to ship something, and iterate, and improve on it (some of my side projects).
Yes you are right. And I try my best to collect them... (from official document, to character models ripped from existing games, and smaller budget JRPGs... I won't repeat them all here).
But my question is: how to identify those features that make animation friendly? I needed a rigged character that I can pose with, right?
You did say that, but maybe I misunderstood you: Did you mean a fully rigged reference character? Or did you mean a "skeleton" in terms of HumanIK skeleton?
I am already posing my character into different poses (using keyframe animation trick other mentioned) to check out topology deformations. But I am not exactly sure what you mean by "adjust the model to the skeleton"?
And this question leads back to my need to rig the character properly. Because sometimes the deformation is so bad because I skin-weights it poorly.
So I am in a horrible circle:
I need to improve her topology ->
I need to see her deforms ->
I need to get her rigged ->
I need to figure out if her problem is topology or skin weights ->
but I don't know her topology is good enough...
> But as said, the best way to get a feel for all that is to go all the way through the pipeline to then get a feel of what can be done better. It's tricky because it is highly technical (rigging tools) but also depends on aspect that can only be learned with time (good human anatomy and stylisation).
You are absolutely right, and I follow this same guideline when programming.
But I just realise something: my problem is also I have no definitive yard stick, I would do much better in a fail-and-retry loop if I can say "I just want to get as close to this model as possible".
--------
TL;DR: Thx for all the support, if anyone can answer some of my highlighted questions, that would be great!
> Related to this topic, especially for the shoulder articulation area, I always recommend the work of Brian Tindall (www.hippydrome.com) as a reference for modeling for articulation, to study pivot joint placement and how the corrective shapes help to achieve the nicest deformations, especially with their gifs where you can see the effect of using only the skin cluster vs. adding the corrective shape over the joint deformation.
One thing I still don't get: many times the recommended joints placement for rigging is not at the same place as the actual human skeleton.
I wonder how animators judge when to add joints or move joints?
"Ultimately I want to get my character from Maya LT to Unity."
This is the most important part, and what you should focus on. Unity doesn't care if you use this or that autorigger - as a matter of fact, you absolutely do NOT need a control rig at all, since the only thing that Unity will see is the weighted mesh. Of course you will likely need a advanced control rig if you want to ultimately create your own animations (for IK switching, lookat constraints, and so on), but this is a completely different topic.
Going deeper : Unity has a dedicated setup for characters, and its most recent iteration is called the "Humanoid" setup, which relies on a very specific skeleton. You need to start from that, and then work your way backwards all the way back to your authoring app. That means downloading the official Unity assets, seeing how they are done, seeing if you can successfully replace them with your own, and so on. Of course using the Unity Humanoid setup is not necessary, but being familiar with it doesn't hurt since it will allow you to use any animations made for it.
Overthinking the authoring side of things will lead you nowhere if you are not taking the time to very clearly establish your target specs (been there, done that). So : download the Unity Standard assets, load up one of the character examples (for instance the Ethan thirdperson setup) and work on overriding that with your own test character. Once you manage to successfully do that, then it will be time for you to think about the best way to set things up in Maya for the best, smoothest possible workflow.
On a side note : this is why people who have solid experience with game modding make for great hires in a team - because they are used to this "goal oriented" kind of reverse engineering.
As for the note on shoulders : this is related to your other thread. Your other thread shows that you understood good shoulder joint placement, but the examples shown in this current thread show bad placement. The hands of your model are also way too tiny - that means that you are currently in a weird space, thinking a lot about complex technical aspects ... yet not taking the time to properly proportion your character based on solid reference images. Gotta do things in order
http://xyproblem.info/
For instance, had we known right from the start that your goal was to put your character in Unity, all the time spent messing around with an automatic rig builder would have been saved - because none of that is necessary to export out a character.
Admittedly, I was wrong in trying to hide my main objective: I was trying to not leave an impression that I only care about making my characters work in games...
One question I still wondering though, if you don't mind me asking:
- If what I have in mind is an adventure game, which means I need some simple cutscenes, would these 2 types of cutscenes differs a lot in their ways of 3D model preparation?
Is that actually what you have in mind, or is that an hypothetical question ?
"which means I need some simple cutscenes..."
There is nothing simple about creating cutscenes, and cutscenes are not a requirement for an adventure game.
"would these 2 types of cutscenes differs a lot in their ways of 3D model preparation?"
You already know the answer to this question
If the question is "what should I use for my first adventure game", then the answer is probably : no cutscenes at all, or, cutscenes relying on still shots and text, or, 2d cutscenes.
- I am making a management sim. It has adventure element that's crucial to gameplay, but relatively small in volume. I want to create cutscenes specifically for some of the storyline.
- I say they are cutscenes, but maybe they are more like "panning/zooming shots". I need to convey the situation/emotion of characters to players subtly, and I find static image and text to be lacking.
- To add, I am not going for 2D point-and-click adventure or visual novel. IF I was, I would not spend time on 3D, or use Unity. There are far superior solutions for these genres.
- Did I realise Life is Strange is significantly better animated than Atelier Sophie? Yes. But I also realise they have the same number of animators in the credits.
- Which means, there are focus and limit I could put on my "cutscenes", that makes it feasible even for a smaller team.
I am in the process of figuring this out.
At the end of the day it is all about very precisely establishing your target. Let's dig deeper : if you were to ask a few different animators about creating cutscenes similar to these, some might start talking about complex facial setups driven by bones, while others would suggest using a simple set of lerped facial blendshapes.
But !
In the grand scheme of things, all this would take a ton of time (starting by looking for an animator to hire, establishing the contract and very precise task brief, and so on), which would add a ton of inertia to your initial design feedback loop. Whereas you could get 95% of the whole look done by yourself in a matter of days, by simply exporting out posed characters to Unity and setting up camera cuts. By doing that you will figure out a TON of unexpected technical details, which in turn would allow you to very precisely establish your task breakdown when the time comes to hire an animator to get the proper animation work done.
In other words : if you are about to spearhead a project, your best bet is to make sure that you control/understand 100% of the technical aspects of it by doing it yourself first. That means that even if you are not creating the final animations yourself, you still need to know in full detail how to deal with animations assets, import them, and so on. This goes back to what I was saying about using throwaway assets that do not take a ton of time to create but still solid enough to be representative of the final specs. You can find dozens of such assets on the Unity store, meaning that for a few bucks you could already be working on your cutscenes tonight as opposed to fighting with Maya rigs
> By doing that you will figure out a TON of unexpected technical details, which in turn would allow you to very precisely establish your task breakdown when the time comes to hire an animator to get the proper animation work done.
I have the same thought. I was more worried about not knowing 3D modelling, so I started with basemesh. (Also I was a bit reluctant to just hand that job to someone else, without me knowing the details.)
True, I aim to do that.
> This goes back to what I was saying about using throwaway assets that do not take a ton of time to create but still solid enough to be representative of the final specs. You can find dozens of such assets on the Unity store, meaning that for a few bucks you could already be working on your cutscenes tonight as opposed to fighting with Maya rigs.
Yeah I should totally do that.
(I have used pre-made environmental assets for game prototypes before. I guess I was unsure how much reuse is possible with character posing, so I didn't do that this time...)
Anyway, I agree it's best for me to:
(1) create an acceptable scene layout and camera angle,
(2) clear a path between Maya LT and Unity, using pre-made dummy characters,
(3) then attempt to shoehorn my own characters onto that scene.