How to perform mocap sessions correctly with technology limitations?

Jonathan85

Hello, i have acces to some inertial mocap "suits" and i will be recording quite a lot of anims i need for projects. But as probably every mocap solution, even this has its limits, for example when you clap hands/palms they intersect/cross into each other etc. There are simply some limitations. And to add, im not an actor, but still i will have to act the majority if not all the mocap animations.

My question for those more experienced with animations and mainly mocap is: When "acting a scene" should i act as realistically with my movements as i can, and leave the "fixing" of it to postprocessing of the anims in some 3D package? Or should i take into account the technology limitations and try to compensate for them straight during the recording?

So for example if i know my hands during claping go through each other should i not clap in reality and just move hands close together expecting that in the "virtual world" they will touch correctly in the animation?

Because trying to compensate for technology limitations will hinder my attemp to act.

So whats the standard procedure in mocap? Act as naturally as you can and fix it later in 3D soft, or try to compensate for technology limits right during the recording with limiting, exaggerating etc. your movements?

antweiler

I have worked similar like you with our XSens system. I would do the motion in a natural way and find ways to fix them later. Since we are not actors, doing natural motions is challenging enough. I find natural timing and dynamic motion more important than details, that need to be adressed anyways.

It also helps a lot, when you record your takes on video (and know later which vid belongs to which take)

However, if you have the system permanently available, you can also give the compensating a try, why not?

Jonathan85

Yeah thats what i thought also, just try to act as naturally as you can (which is challenging enough as you said when one is not an actor), and fix it later.
Yeah i thought about the video also.

BTW did you do multiple takes of the same motion? To choose to best one (or even blend between the best ones) later on? If one has the time, i think its better, did you do it as well?

antweiler

We had to do several takes in a rush outside the office in an industrial environment. Yes, we did several motions in one take, easier to record, view and compare. But you will need assistence, someone who helps write down the take and file names in a list and hits the record button

Jonathan85

No i mean repeat the same motion lets say 3 times to choose the best one afterward, the one with the best acting and with the least ammount of technical issues. I gues its a good idea, i dont even know why im askign this, but why not :-).

antweiler

Of course, as long as you have enough time and disk space

Mark Dygert

Whenever possible I strongly encourage people to hire actors who have some experience with this type of acting. I know, "but they cost money". Well yeah, most things worth doing well require skill and those skills aren't free.

This is what happens when you don't get the right people or try to do it yourself (knight at 0.13s). https://youtu.be/OifvTaAQ678?t=13s

If you can avoid putting on the suit and acting it out please do. Actors understand fully what it means to give themselves to the scene and will do whatever necessary to deliver a quality experience, which really pays off when people are tired and need to push themselves to keep the quality up.  

Animators are a sedentary, squishy folk and are often scared of pain and physical exertion. This shows through in the acting, which is fine if you need to animate that type of character.  But if not, you'll see people guard themselves when they take a fall. An actor won't, they'll fall over and over again like you threw them off a cliff.

Non-actors can be overly aware of their surroundings and have trouble committing, leading to many many takes and none of them being worth it.

I'm not saying they can't pull it off, but it will help if they've ever done any physical acting before. Actors know how to walk like a bad ass and respond to a hit or throw themselves off of something, they love doing it. Animators often love watching it, and like making it happen in 3D, but normally can't pull it off. 

slipsius

Seconding Mark. Actors are the way to go. There are many that specialize in mocap capturing. 

Though, with that said, to answer your questions... Yes, just act as natural as possible while still keeping in mind the character you are trying to capture. Exaggeration helps, but at the same time, things can be exaggerated after the fact. 

In the professional shoots that I've been too, things tend to be acted out a few times, and when they doing something that they like, the director of the shoot will tell the mocap guys (the ones that run the equipment) to keep that take. Not every take is going to be kept. A lot are thrown away. All the data is generally kept somewhere, but the takes you keep are the ones that get sent to mocap cleanup so that the cleanup guys dont waste their time on acting choices you`ll never use. 

If you have an at home solution, then ya, act it out, review it, if you like it, keep it and move on. 

sprunghunt

Jonathan85 said:

Hello, i have acces to some inertial mocap "suits" and i will be recording quite a lot of anims i need for projects. But as probably every mocap solution, even this has its limits, for example when you clap hands/palms they intersect/cross into each other etc. There are simply some limitations. And to add, im not an actor, but still i will have to act the majority if not all the mocap animations.

My question for those more experienced with animations and mainly mocap is: When "acting a scene" should i act as realistically with my movements as i can, and leave the "fixing" of it to postprocessing of the anims in some 3D package? Or should i take into account the technology limitations and try to compensate for them straight during the recording?

So for example if i know my hands during claping go through each other should i not clap in reality and just move hands close together expecting that in the "virtual world" they will touch correctly in the animation?

Because trying to compensate for technology limitations will hinder my attemp to act.

So whats the standard procedure in mocap? Act as naturally as you can and fix it later in 3D soft, or try to compensate for technology limits right during the recording with limiting, exaggerating etc. your movements?

The problems with different skeleton shapes - like hands going through each other - can be fixed by using retargeting in software like motion builder. You should be able to set it up to do it in realtime and not afterwards.