Processing an acquisition¶
process reads recorded camera calibration, channel settings, stage positions,
and scan geometry from the raw acquisition. Mirror and stage scans are deskewed;
native planar acquisitions follow the planar processing path.
Choose corrections and output¶
uv run process "/path/to/acquisition" --deconvolve --flatfield-correction --output "/path/to/processed"
Camera calibration runs before optional illumination correction and
deconvolution. --flatfield-correction estimates or reuses illumination fields.
--deconvolve uses GPU RLGC; channel-ordered PSFs can be supplied with repeated
--decon-psf-paths options. Otherwise theoretical PSFs are generated.
The default saved pixels are clipped uint16 values. --save-float32 retains
fractional calibrated intensities. --z-downsample-level controls the integer
reduction of deskewed Z, defaulting to 2. Calibration and interpolation
conventions are described in methods.
Select or skip data¶
--time-range START STOP and --pos-range START STOP select zero-based,
stop-exclusive ranges. With --skip-empty-below VALUE, a channel tile is zeroed
when too little calibrated signal reaches the cutoff. The minimum signal
fraction defaults to 0.01 and can be set with
--skip-empty-min-signal-fraction. This check precedes illumination correction.
Resume a run¶
Resume uses the processing journal beside the outputs and requires compatible
settings. Durably completed tiles are reused. Without --resume, the selected
output run is overwritten. Keep the journal with its image stores; see the
data reference.
Produce a registered mosaic¶
The maximum-Z mosaic created during processing uses nominal stage placement. Run fuse to register tiles and create the registered fused volume and projection. When output is stored separately, pass that output directory to subsequent commands.
The process reference lists options. Experimental scan upsampling and its limitations are described in the deconvolution methods.