- Ensure reproject_and_coadd handles bg-matching with one input image by @svank in #412
- Fixes for mosaic output pixels not covered by inputs by @svank in #413
- Updated docstrings for output_projection and shape_out to indicate that any APE-14 WCS is acceptable by @astrofrog in #407
- Add tests for full range of inputs/outputs in healpix functions by @astrofrog in #408
- Fix Cython warnings by @svank in #418
- Fix devdeps job by @astrofrog in #415
- BLD: pin extension-helpers to 1.* following upstream recommendation by @neutrinoceros in #420
- Added sp-repo-review to pre-commit by @astrofrog in #411
- Bump actions/checkout from 2 to 4 by @dependabot in #422
- Bump stefanzweifel/git-auto-commit-action from 4 to 5 by @dependabot in #423
- Add configuration for nightly wheels by @astrofrog in #417
- Fix pre-commit errors by @astrofrog in #429
- Enable testing of wheels on silicon mac by @astrofrog in #428
- Build against Numpy 2.0.0rc1 or later by @astrofrog in #436
- @neutrinoceros made their first contribution in #420
- @dependabot made their first contribution in #422
Full Changelog: https://github.com/astropy/reproject/compare/v0.13.0...v0.13.1
- Fix TestReprojectAndCoAdd failure on i386 by @olebole in #386
- Fixed a bug that caused reprojected dask arrays to not be computable due to a temporary directory being removed by @astrofrog in #390
- Add ability to specify output array and footprint in reproject_and_coadd by @astrofrog in #387
- Added ability to set
block_size='auto'and fix missing parameters in docstrings by @astrofrog in #392 - Compute footprint in healpix_to_image by @lpsinger in #400
- Fix compatibility with Cython 3.0.2 and update version in pyproject.toml by @astrofrog in #391
- Add tests on Python 3.11 by @dstansby in #401
- Add testing on Python 3.12 by @dstansby in #399
- Python 3.12 testing by @dstansby in #403
- Add testing on Python 3.12 by @astrofrog in #402
- Enable Python 3.12 wheel building by @astrofrog in #405
- @olebole made their first contribution in #386
Full Changelog: https://github.com/astropy/reproject/compare/v0.12.0...v0.13.0
- Fix support for NDData objects with dask .data attributes by @astrofrog in #365
- Fix docs mosaic page rendering by @jdavies-st in #381
- Add despike_jacobian option for adaptive resampling by @svank in #366
- Refactor blocked/parallel reprojection by @astrofrog in #374
- Add 'first' and 'last' moasicking modes to reproject_and_coadd by @svank in #383
- Add modes for nan and inf handling to adaptive algo by @svank in #380
- Added new combine function to compute the minimum and maximum by @fjankowsk in #369
- TST: Update URL for Scientific Python nightlies by @pllim in #368
- Dask support improvements by @astrofrog in #367
- Fix --remote-data tests by @astrofrog in #375
- Update docstring for adaptive defaults by @svank in #378
- Use more points to find image bounds in moasics by @svank in #382
- Skip Python 3.12 wheels by @astrofrog in #385
Full Changelog: https://github.com/astropy/reproject/compare/v0.11.0...v0.12.0
- Fix for HighLevelWCS attribute error by @keflavich in #349
- Fixes for solar frames and non-degree units by @astrofrog in #360
- If shape_out is specified, use this over the array_shape attribute of a WCS object by @astrofrog in #361
- Allow single inputs to find_optimal_celestial_wcs and add ability to specify HDU by @astrofrog in #344
- Add support for specifying output projection as APE 14 WCS with array_shape defined by @astrofrog in #345
- Started adding support for allowing dask arrays as input by @astrofrog in #352
- Mark tests that use remote data by @smaret in #339
- Fix code style by @astrofrog in #340
- Simplify blocked reprojection implementation by using dask and improve efficiency of parallel reprojection by @astrofrog in #314
- Remove code that was required for astropy<4 by @astrofrog in #346
- Add a new 'all' extras for shapely by @astrofrog in #363
- @smaret made their first contribution in #339
Full Changelog: https://github.com/astropy/reproject/compare/v0.10.0...v0.11.0
- Close FITS files after loading by @svank in #330
- Add support for blocked and parallel reprojection in
reproject_interpby @AlistairSymonds in #214 - Add support for efficiently reprojecting multiple images with the same wcs by @svank in #332
- Add support for APE 14 WCSes in find_optimal_celestial_wcs by @astrofrog in #334
- Update package infrastructure by @Cadair in #304
- Changed default filter kernel and boundary mode in
reproject_adaptive, and removedorderargument. by @svank in #291 - Skip wheel tests on manylinux_aarch64 by @astrofrog in #307
- Reformat all Python code using Black by @Cadair in #308
- Use pixel_to_pixel from astropy.wcs.utils by @astrofrog in #315
- Test CI on Python 3.11 beta by @dstansby in #298
- Update pinned version of Cython by @astrofrog in #316
- Speed up test_blocked_against_single by increasing smallest block size by @astrofrog in #319
- Fix weird quotation marks from Black auto-formatting by @svank in #331
- Fix CI by @astrofrog in #333
- @AlistairSymonds made their first contribution in #214
Full Changelog: https://github.com/astropy/reproject/compare/v0.9...v0.10.0
- Drop support for Python 3.7.
- Infrastructure and packaging updates.
- Made many improvements, bug fixes, and significant speed-ups for the adaptive
- resampling algorithm,
reproject_adaptive. These bug fixes may cause - changes to the reprojected images, which are typically negligible.
- Improvements include the addition of a flux-conserving mode, support for a
- Gaussian filter kernel, a menu of boundary-handling modes, and a
center_jacobianflag to trade speed for accuracy with rapidly-varying- transformations.
- Added a
roundtrip_coordsargument toreproject_adaptiveand reproject_interp. By default, all coordinate transformations are run in- both directions to handle some situations where they are ambiguous. This can
- be disabled by setting
roundtrip_coords=Falsewhich may offer a - significant speed increase.
- Improve
find_optimal_celestial_wcsto accept input data descriptions as - just array shapes, not necessarily fully populated arrays. This makes it
- possible to solve for the optimal WCS for a set of images that couldn't fit
- into memory all at once, since the actual data aren't needed for optimal WCS
- determination. [#242]
- Fix implementation of
hdu_weightsinreproject_and_coadd. [#249]
- Fixed compatibility with Astropy 4.1. [#234]
- Updated minimum requirement for SciPy. [#236]
- Made C extension in overlapArea.c thread-safe by removing global
- variables. [#211]
- Made it possible to control whether to output debugging information
- from overlapArea.c by setting DEBUG_OVERLAP_AREA=1 at build-time. [#211]
- Fix compatibility with astropy v4.0.1. [#227]
- Disable parallelization by default in
reproject_exact- this can be - enabled with
parallel=True. [#227] - Fixed a bug with
reproject_exactwithparallel=Falseand return_footprint=False, which caused the footprint to be returned- anyway. [#227]
- The infrastructure of the package has been updated in line with the
- APE 17 roadmap (https://github.com/astropy/astropy-APEs/blob/main/APE17.rst).
- The main changes are that the
python setup.py testand python setup.py build_docscommands will no longer work. The- easiest way to replicate these commands is to install the tox
- (https://tox.readthedocs.io) package and run
tox -e testand tox -e build_docs. It is also possible to run pytest and sphinx- directly. [#228]
- Added support for using any WCS that conforms to the WCS API described
- in the Astropy Proposal for Enhancements 14 (APE 14). The
independent_celestial_slices=argument toreproject_interphas- been deprecated since it is no longer needed, as transformations are
- automatically done in the most efficient way possible. [#166]
- Include a warning for high resolution images with
reproject_exact, - since if the pixels are <0.05", precision issues can occur. [#200]
- Added a new
reproject_and_coaddfunction for doing mosaicking of - individual images, and added section in documentation about mosaicking.
- [#186]
- Added a new reproject.adaptive sub-package that implements the DeForest
- (2004) algorithm for reprojection. [#52]
- Fixed a bug that caused 'exact' reprojection results to have numerical
- issues when doing identity transformations. [#190]
- Fixed a bug that caused 'exact' reprojection to fail if one or more of
- the WCSes was oriented such that E and W were flipped. [#188]
- Improve parse_output_projection to make it so that the output projection
- can be specified as a filename. [#150]
- Fixed a bug that caused HEALPix maps in RING order to not be correctly
- interpreted. [#163]
- Make it possible to specify the output array for reprojection using the
output_array=keyword argument. [#115]
- Refactored HEALPix reprojection code to use the astropy-healpix package
- instead of healpy. [#139]
- Added the ability to specify an output array in
reproject_interp, which - permits the use of memory-mapped arrays and therefore provides the capability
- to handle data cubes much larger than memory [#115]
- Fix test 32-bit test failures. [#146]
- Fix an issue with reprojecting images where there are two solutions along
- the line of sight by forcing round-tripping of coordinate conversions [#129]
- Explicitly define default HDU as 0 for normal reprojection and 1 for
- HEALPix reprojection. [#119]
- Added a function to find the optimal WCS for a set of images. [#136, #137]
- Fix a regression that caused certain all-sky images (e.g. the Mellinger Milky
- Way Panorama, http://www.milkywaysky.com) to be reprojected to all NaNs when
- the output WCS was in Mollweide coordinates. [#124]
- Include missing license file in tarball.
- Updated documentation to remove warnings about early versions.
- Allow users to pass a
field=option toreproject_from_healpix - to access different fields in a HEALPIX file. [#86]
- Significant improvements to performance when the input data is a large
- memory-mapped array. [#105]
- Significant refactoring of interpolating reprojection to improve support for
- n-dimensional arrays, optionally including two celestial axes (in which
- case the coordinate transformation is taken into account). [#96, #102]
- Fixed a bug that caused reprojection by interpolation to be truncated for
- rectangular output images.
- Initial Release.