It’s an exciting period of time for the biomechanics community in general, and motion capture in particular. New markerless and inertial sensors are helping scientists collect large scale data sets outside the lab. Going hand-in-hand with this is the continued desire to share motion capture data. Pooling data between collaborating labs can unlock new research questions. And connecting motion capture with other data sources brings context to motion. But maximizing these opportunities need new tools.
What it takes to share data
A recently published pre-print offered the perspective of several international societies, including ISB, ASB, and ESB, on the opportunity and need to share motion capture data. In particular they identify that
data sharing will also play a crucial role towards achieving a level of consistency that can facilitate comparison and outcome interpretation in large population datasets.
There are many facets to this problem including: implementing the FAIR principles to enable computational support for data analysis, getting consent from participants and other legal permissions, physical data infrastructure, and appropriately reporting metadata. Adjacent yet entangled with these considerations is the specific file format that will store this data and meet these requirements.
At the core of our philosophy
HasMotion is well-placed to contribute to these community efforts. For one, we believe that sharing data and reproducing analysis must be at the core of understanding human motion.
Visual3D’s 25-year history began with the US National Institutes of Health’s need to harmonize motion capture studies across labs. It provides researchers with a standardized framework for modelling and analysing human movement. This includes full support of the C3D format, the biomechanics community’s standard file format for 3D and analog data. And a workspace file, the CMZ, that keeps data, modelling decisions, processing histories, and analytical results all in one file.
“Two analysis files for one trial is one file too many.” – Scott Selbie, CEO @ HasMotion
Sift has extended this careful approach to large motion capture data sets, providing researchers with tools to scale their quality assurance, exploration, and analysis of motion capture data. And at scale there is even more pressure to get analysis done right the first time: the penalty for unshareable, unreproducible analysis is continual, costly recalculations.
Moving Forward with HDF5
It’s natural, then, to continue our work towards sharable, reproducible biomechanical analysis by introducing an HDF5-based file format for motion capture analysis. Joining AnyBody and Moveck in providing the motion capture community with access to HDF5-based storage, our format replicates the CMZ file’s historical functionality with an eye to the future. Users of Visual3D v2026.07.1 will be able to save and load HDF5 files as a preview feature, replicating the CMZ file’s functionality while providing easy access to any framework that supports HDF5.
Why did we choose HDF5? The format originates with work at the U.S. National Center for Supercomputing Applications in 1987 and continues today with the support of The HDF Group, a non-profit corporation. Designed for large, multi-modal datasets, the format has been proven over decades of use in fields such as astrophysics, genomics, aerospace engineering, and finance. And as an open source, self-describing file format we felt that it met our users’ needs while also embodying the qualities that we’ve always believed are necessary for robust science.
Looking Ahead
With initial support for writing to and reading from this HDF5-based format now available for preview, we’re looking forward to what comes next. We stand ready to work with our users, industry partners, and international societies to develop the tools that today’s biomechanics researcher needs.