A microscope's-eye journey through scale. Slide across the physical size (µm, log scale) of real bioimaging datasets shared via the SSBD:database, from single molecules to whole tissues.
Two views, sharing the same data:
- Grid view — a slider with square thumbnail panels arranged in a grid, nearest to the current scale.
- Vertical 3D view — a vertical scale bar; moving it slides the current image/metadata away in depth (bursting toward the viewer or sinking into the screen) while the next one arrives.
data.json— one entry per dataset: SSBD ID, dataset name, title, license, contact, physical crop length (µm), pixel dimensions, and thumbnail filename.thumbnails/— a square, center-cropped PNG thumbnail per dataset, rendered from the lowest usable resolution level of each dataset's OME-Zarr array (composited from active channels using their omero color/window metadata).
The physical scale used to rank datasets comes from the SSBD metadata REST APIs (v1, v2, v3) — specifically each dataset's X/Y scale (µm/pixel). Datasets the API has no scale metadata for are excluded rather than falling back to a less-authoritative estimate.
Each dataset's license (as recorded in the SSBD:database) is shown on its card; check it before reusing an image.
scripts/ holds the pipeline used to build data.json and thumbnails/ from
the openssbd/ssbd-bff CSV listing:
ssbd_omezarr_area.py— lists OME-Zarr datasets from a bff CSV export by physical image area (fetches each dataset's.zattrsfor pixel size).generate_thumbnails.py— renders and center-crops one square thumbnail per SSBD project from its OME-Zarr array.fetch_ssbd_api_scale.py— looks up each dataset's X/Y/Z/T scale from the SSBD v1/v2/v3 REST APIs.merge_api_scale.py— recomputes the physical crop length using the API-sourced pixel size.
These require zarr, fsspec, numpy, and Pillow, and network access to
dmss3gw.riken.jp and ssbd.riken.jp.
This is a static site — any static file server works, e.g.:
python3 -m http.server 8000
then open http://localhost:8000/.