Cookies help us deliver our services. By using our services, you agree to our use of cookies. Learn more

crowdAI is shutting down - please read our blog post for more information

NIPS 2017: Learning to Run

Reinforcement learning environments with musculoskeletal models


Video recording

Posted by al-runners over 4 years ago


I would like to record a video (capture frames) automatically. Is there a way to do that programmatically from Python? Alternatively, is there a possibility to control the viewer directly?

Cheers, flexman

Posted by Lukasz_  over 4 years ago |  Quote

Unfortunately, there is no possibility at the moment. For generating gifs, we actually modified SimBody engine https://www.crowdai.org/challenges/nips-2017-learning-to-run/leaderboards This should work in your case as well, but it won’t be possible from python level.

Best, Łukasz


Posted by spMohanty  over 4 years ago |  Quote

Hi @flexman,

While opensim doesn’t yet support programmatic video generation out of the box. The approach we use is to use a modified version of simbody to generate a visualisation on every run.

The modified version of simbody can be found at : https://github.com/spMohanty/simbody

You will have to build it from source, and replace the simbody-visualiser in your conda-env with the one generated by building this repo.

But note, this is a very hacky way of doing this. Opening a feature request in the official simbody github Page’s issue tracker might lead to a better solution.