save meshes in sharded neuroglancer precomputed

[fb0l]

Hi everyone,
I have a local dataset in neuroglancer precomputed format (sharded) which also contains meshes.
Following the tutorial, I'm loading the meshes and I'm applying some transformations to them.
After the transformations have been applied, I'd like to save the updated meshes back as sharded neuroglancer precomputed format, ideally into a new output folder.

Is this possible with cloud-volume?
Do you have experience with this use-case and know how to continue?

Thanks!

[william-silversmith]

Hi fb0l,

Producing a shard is kind of convoluted, but if you do the updates in batches, you could try following the code here and replacing the original shard filename:

https://github.com/seung-lab/igneous/blob/master/igneous/tasks/mesh/multires.py#L356-L358

You can also generate an unsharded volume and convert it to sharded using igneous mesh xfer and then delete the original.

Let me know if you need additional help!

[william-silversmith]

I forgot to mention, you might find it beneficial to use disassemble_shard

cloud-volume/cloudvolume/datasource/precomputed/sharding.py

Line 520 in fb3a8c3

def disassemble_shard(self, shard):

You'll need to use logic in this file in order to read the multi-lod format:

cloud-volume/cloudvolume/datasource/precomputed/mesh/multilod.py

Line 18 in fb3a8c3

def extract_lod_meshes(manifest, lod, lod_binary, vertex_quantization_bits, transform):