Going to zarr would allow much better scalability and memory management by using lazy-loading for tractography. At the moment cardiotensor load the entire vector field. The memory limitation can be overcome by using the bin parameter, however for very large high-resolution dataset, heavily binning is not ideal.
Saving as .zarr would also natively have bin version instead of creating them when user request them.
Going to zarr would allow much better scalability and memory management by using lazy-loading for tractography. At the moment cardiotensor load the entire vector field. The memory limitation can be overcome by using the bin parameter, however for very large high-resolution dataset, heavily binning is not ideal.
Saving as .zarr would also natively have bin version instead of creating them when user request them.