Simulating complex sports and dancing moves

Hi there,

I would like to know if the AnyBody model would be reliable to simulate the mechanics of complex movements such as those practiced in sports and danse.
These movements would be inputed with motion capture, but I never used this input method before. What is the reliability, the advantages and the drawbacks of using motion capture to estimate muscle activity with AnyBody ?

Hi Pierre,

I would say for complex motions you will need to get mocap data. Many users are doing this and it works fine. It might be a bit tricky, but if you have a good mocap/C3D file then it should be fine.
What we have done before: Gait, Stairs, Kneebend, Squad, Running, Sit-to-Stand, badminton, Egression (with BMW get in and out of a car),… I also saw a dancing model without forceplates once.

For only kinematics, there are several free options like; many of those C3D files are without forces though, and you cannot do inverse dynamics.

Hi Amir,

Ok but what about the muscular recruitment strategy? Does energy-related criteria are still reliable for such complex movements that involve speed and strength ?

Thank you.


Hey Pierre,

this is a difficult question and it is depending on several aspects. In general I would say that the muscle recruitment criteria is fine until a certain value of speed, strength and impact during these sportive activities. There is also no specific point of when results are not reliable anymore, it is more a sliding transition. So it depends on what activity you want to analyze. I’m aware of one study of the Aalborg University group, where fast arm-pedaling with high loads was monitored by EMG and showed very good results. As far as I know there are also studies of cycling and rowing which gave good results. Here are some aspects on how to chose suitable sports:

  • how well is the subject trained for this motion? Is it a first time motion or is it something he “optimized already”?

  • will the subject have to react on sudden changes from outside?

  • is there unexpected high impact in the motion? eg. AnyBody is not made for crash-tests…

That is not everything, but might be some guidelines…

Hi Amir,

Thank you for your answer. This will guide our choices for sure.



You said that you already saw a dancing model without force plate. Do you know if this could be implemented in the repository ?

You also said the following:

…the muscle recruitment criteria is fine until a certain value of speed, strength and impact during these sportive activities…

If a quick movement with high loads (not necessarily impacts) is repetitively performed by athletes, thus it means that it is enough physiologically reasonable to be handled well by the recruitment criteria of AnyBody ?

There is also a problem with weak muscle strengths in the AnyBody Model, especially at the shoulder (see my other thread I guess that this has to be fixed before using the model with athletes, which are known to have a bigger trained musculature. Otherwise, we can expect muscle overload, right ?

What do you think ?



the dancing model was from a user that wouldn’t want to share (so far).

About the strength, if you know that you are dealing with an athlete you have to change the strength of the muscles. There is the option of strength scaling in the model or even change the max strength of each individual muscle.

Hi Amir,

As you say, there is always the option of scaling the strength of the muscles if we want to represent an athlete. So far, I did not test it, but I plan to do it for sure.

I think that it is not normal that the current muscle strength (at least at the shoulder) is not able to perform some movements such as anteflexion, without overloading the muscles. This is not representative of a normal person. This issue has been discussed multiple times in conferences dealing with musculoskeletal models. It is not only an issue in AnyBody. At least, this aspect of limited strength of shoulder muscle should be clearly mentioned somewhere in the tutorials or in the definition of the model.