JRCVNC2018F -> JRCVNC2018U_HR image-mode warp via H5 displacement field — jrcvnc2018f_to_jrcvnc2018u_hr

Bridges a JRCVNC2018F-aligned NRRD onto the canonical NeuronBridge VNC grid (JRC2018_VNC_UNISEX_HR, 573 x 1119 x 219 @ 0.461 / 0.461 / 0.7 um) in a single RenderTransformed call.

The H5 (JRCVNC2018U_JRCVNC2018F.h5, ships with flybrains's download_jrc_vnc_transforms()) stores a dfield/invdfield pair sampled on the JRCVNC2018U grid. The dfield is the forward (source -> dest) push from JRCVNC2018F to JRCVNC2018U, matching the brain-side <DEST>_<SRC>.h5 naming convention used by nat.jrcbrains; for image resampling we want OUTPUT (U) -> SOURCE (F), so we pass -i to invert it. Asking for the HR output grid directly works because JRCVNC2018U and JRCVNC2018U_HR share the same physical coordinate system — the JAR interpolates the dfield onto the requested grid.

jrcvnc2018f_to_jrcvnc2018u_hr_h5(
  input,
  output,
  h5_path = NULL,
  threads = 8L,
  verbose = TRUE
)

Arguments

input: source-space NRRD (must already be in JRCVNC2018F physical coordinates — e.g. the output of tif_to_nrrd_channels on a 660 x 1342 x 358 voxel ImageJ "registration" export at 0.4 um isotropic).
output: target NRRD path on the JRCVNC2018U_HR grid.
h5_path: path to JRCVNC2018U_JRCVNC2018F.h5. Defaults to ~/flybrain-data/JRCVNC2018U_JRCVNC2018F.h5 (where flybrains.download_jrc_vnc_transforms() writes it).
threads: integer; passed to the JAR's -q flag.
verbose: logical; pass-through to the JAR.

Value

invisibly, the absolute path to the written NRRD.