Skip to content

Latest commit

 

History

History
404 lines (194 loc) · 6.34 KB

CHANGELOG.rst

File metadata and controls

404 lines (194 loc) · 6.34 KB
Raw

Change Log

v0.8.2

Fixed:

  • Move the "import PCA" to the top to avoid import error.
  • Add missing required packages in requirements

v0.8.1

Changed:

  • Default pipelines to filtered back projection algorithm rather than grid reconstruction algorithm

Removed:

  • no minus_log step for diffraction tomo

v0.8.0

Changed:

  • Replace lambda function for sorting the sinogram with actual function
  • radiograph pipeline data -> norm_img

v0.7.0

Added:

  • amorphsivity_pipeline for measuring if a PDF is amorphous
  • xpdtools.pipelines.tomo.sort_sinogram for sorting sinograms by their theta values
  • Ability to stack 2D tomographic reconstructions into 3D for pencil beam
  • xpdtools.pipelines.tomo.recon_wrapper function which wraps tomopy.recon to handle 2, 3 and 4D arrays

Changed:

  • xpdotools.pipelines.tomo.tomo_pipeline_theta uses sort_sinogram rather than a lambda for provenance
  • xpdtools.pipelines.tomo pipelines use the recon_wrapper rather than tomopy.recon

Fixed:

  • Don't use pluck since we can pass in just the G(r)

v0.6.0

Added:

  • Principle Component Analysis pipeline and tooling

Changed:

  • Flatten pencil beam tomo output from (1, x, x) to (x, x)

Fixed:

  • Automatically make the calibration directory
  • Tomo pipeline now has np.nan_to_num before and after reconstruction

v0.5.0

Added:

  • Background scale node
  • Background scale kwarg to cli/process_tiff
  • New example pipeline for parallel operation
  • Added standard deviation nodes
  • Max intensity node
  • Position of max intensity node
  • Tomography pipelines for full-field an ctPDF

Changed:

  • Chunk pipelines so the can be used modularly and instantiated
  • Move from streamz to rapidz
  • Only use tqdm on 'mean' method for binned_outlier
  • Standard deviation is now divided by mean so it is more meaningful
  • Support imports for both pyFAI 0.15 and beyond

Fixed:

  • Orch now deffers the actual installation to the travis top level process

v0.4.3

Fixed:

  • Fix bug where is_calibration_img was not being checked properly
  • no mask setting, we don't need kwargs for no mask

v0.4.2

Added:

  • calib_setting dict to the raw pipeline, if calib_setting['setting'] is False then the calibration gui will not be run even for calibration runs

Changed:

  • pyFAI imports for calibration are now inside the functions so we don't run the imports until they are actually needed.

v0.4.1

Changed:

  • Added pool to xpdtools.tools.binned_outlier` and xpdtools.tools.mask_img for an executor pool interface
  • Added tqdm for slow mean masking

Fixed:

  • binned_outlier properly uses existing masks

v0.4.0

Added:

  • Pipeline and tools for flatfield correction measurements

Changed:

  • Added default mask kwargs to mask_kwargs for usability

v0.3.1

Fixed:

  • starmap into img_calibration rather than just map which gives correct alignment between the inputs and expected arguments

v0.3.0

Added:

  • Quantity of interest support

Changed:

  • tth reported in degrees
  • Run release before conda forge so we use the self generated tarballs

Fixed:

  • Don't break API with generate_binner

v0.2.0

Added:

  • pipelines.extra module which holds extra nodes (zscore, median, etc)
  • numba compiled zscore for faster zscore computation

Changed:

  • removed zscore, median, and std from the base pipeline
  • use map rather than for loop for zscore

Removed:

  • xpd_raw_pipeline module

v0.1.9

Changed:

  • Merged xpd and standard pipelines into one pipeline
  • Exposed the mask, fq, and pdf kwargs to the user better. Now the kwarg dicts are from the nodes and can be updated.

Deprecated:

  • xpd pipeline (it is now in the standard pipeline)

Fixed:

  • iq_comp now is combined via a combine_latest rather than a zip

v0.1.8

v0.1.7

Changed:

  • Zscore is now turned into float16 before saving to reduce size on disk

Fixed:

  • Command line interface destroys sinks so it shouldn't blow up memory
  • generate_binner now has max q of the max q

v0.1.6

Added:

  • Quickstart to Readme.md

Changed:

  • Save z score as .tif file
  • binned_outlier now uses input mask (if any) to remove pixels before running the binned outlier algorithm.

Fixed:

  • All integrated values are processed with np.nan_to_num before output.

v0.1.5

Added:

  • Kwarg for flipping the input mask (may be needed for fit2d masks)

Removed:

  • Docs for beamstop mask

Fixed:

  • Polarization works properly
  • Multi image works properly
  • Code health badge
  • Docs for mask_img alpha

v0.1.4

Fixed:

  • removed relative import from CLI

v0.1.3

Added:

  • Test of the CLI (to make sure it writes out files now)
  • Tests of many (although not all) of the tools.
  • Added support for scikit-beam=0.0.12 which lacks som cached data

Changed:

  • Readme now reflects the conda package
  • Travis now has a display

v0.1.2

Added:

  • Dedicated XPD pipeline which has the capacity to only mask the first image in a series.

v0.1.1

Added:

  • Benchmark scripts for speed testing (Note that these run on local files currently)
  • Numba for median masking, giving a speedup

Changed:

  • Most zip_latest nodes have been changed to combine_latest to avoid unwanted buffering.
  • Use BinnedStatistics`D properties for masking, which reduces recomputation

Removed:

  • streamz dep, now the project depends on streamz_ext

v0.1.0

Added:

  • Command Line interface for integration
  • Add rever changelog activity
  • Speed up masking via median based sigma clipping
  • Z score visualization to callback pipeline

Changed:

  • Fixed up main pipeline