EgoTac-SC: aether_v1

This dataset contains the aether_v1 Zarr recordings. The source consists of 857 .zarr directories (102,413 files, 458,229,180,085 bytes). To make the large upload and download manageable, the recordings are packaged in 45 independent ZIP archives under aether_v1/.

Each archive contains complete .zarr directories and preserves the aether_v1/ path. The ZIP entries use stored mode because the Zarr chunks already use Blosc/Zstd compression. No recordings are split across archives.

aether_v1/manifest.json maps each .zarr directory to its archive and lists the archive size and SHA-256 checksum. The archives contain the aether_v1 folder itself, so extract all downloaded ZIPs into the same parent directory:

python - <<'PY'
from huggingface_hub import snapshot_download
snapshot_download(
    repo_id="wkzhang/EgoTac-SC",
    repo_type="dataset",
    allow_patterns="aether_v1/*.zip",
    local_dir="EgoTac-SC",
)
PY

for archive in EgoTac-SC/aether_v1/*.zip; do
    unzip -n "$archive" -d .
done

The extracted recordings will be in ./aether_v1/. Allow at least 427 GiB for extraction, in addition to the downloaded archives.