Drosophila melanogaster embryogenesis: a 500-timepoint light-sheet timelapse as Gaussian splats

A Drosophila melanogaster embryo imaged through early embryogenesis on a SiMView-type multi-view light-sheet microscope, with nuclei labelled by a histone H2A variant fusion (w; His2Av::mRFP1; +), fitted to 3D Gaussian splats per timepoint and published as a single 4D archive for the Luxar viewer. Contents 500 timepoints spanning 4 h 10 min of development, from cellular blastoderm through gastrulation to the fully extended germband. One stacked 4D node with time as a real axis, so a viewer scrubs through development rather than loading separate files. Spatial coordinates are physical microns (Z 1.93, XY 0.40625); the time axis is in minutes. file size elements level of detail drosophila_embryogenesis_500tp.gsplats.zarr.zip 1.2 GB 128,000,000 flat, 4 ladder rungs 128,000,000 splats across 500 timepoints is exactly 256,000 per frame. Only the timepoint being viewed is resident, so the whole-node total is not what a GPU commits: a single frame sits far below the per-node element limit. The archive is one flat node with a progressive ladder rather than a tiled or multi-level tree. On a time-stacked node the writer already orders storage by the time barrier, so spatial tiling buys nothing, and coarser whole-object levels are never selected when only one frame is ever shown. Relationship to the other Luxar demo records The permissively-licensed Luxar demo record (doi:10.5281/zenodo.21912280) holds a single frame of this same recording as a standalone still, fitted independently. This record holds the whole timelapse. The two are the same specimen on the same instrument; cite whichever you use, and cite the upstream publication in addition to either. Provenance and licence Derived product: Gaussian-splat fits computed with the Luxar gsplat pipeline. These are lossy representations built for visualisation and interactive exploration; they are not a substitute for the source imagery in quantitative analysis. This version applies no post-fit cull: every timepoint was fitted with a budget of 256,000 splats and all of them are kept, which is why the total is exactly 500 × 256,000. Version 1 of this record (10.5281/zenodo.22118695) held an earlier build whose splat counts had been reduced by a fixed-amplitude cutoff at amplitude 33.4046 — intended as a cumulative-amplitude cull at retention 0.960, but the implementation then in use summed in float32 and tripped its threshold early, so it behaved as a fixed cutoff. That reduction is not applied here. drosophila_embryogenesis_500tp.gsplats.zarr.zip (1.2 GB, cc-by-4.0). Source: Drosophila melanogaster embryogenesis, SiMView-type multi-view light-sheet, nuclei labeled w; His2Av::mRFP1; + (Keller lab, HHMI Janelia Research Campus). Attribution: Acquired in the laboratory of Philipp J. Keller at HHMI Janelia Research Campus, where L. A. Royer was then a postdoctoral fellow; the Gaussian-splat fits in this record were computed later at Chan Zuckerberg Biohub San Francisco. Imaging method: Royer et al., Nature Biotechnology 34, 1267-1278 (2016), doi:10.1038/nbt.3708; the imaging itself is unpublished. The imaging itself is unpublished. For the SiMView-type adaptive light-sheet instrument and imaging method, see Royer et al., Nature Biotechnology 34, 1267-1278 (2016), doi:10.1038/nbt.3708, CC BY 4.0 — that paper is the instrument reference, not a description of this recording. Users of this record should cite both this record and that paper.