1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
|
.. _release_notes:
=============
Release Notes
=============
2.2.0 (2026-01-06)
==================
- Added support for resampling and co-adding images using squared weights.
This can be helpful when resampling variance arrays for the purpose of
performing propagation of uncertainties. [#163, #199]
- Fixed a bug in drizzle-resampling with Lanczos kernels that could result in
decreased accuracy of resampled image. Improved discrete approximation to
the Lanczos kernel. [#198]
- Fixed a bug in resampling using "blot" with Lanczos interpolation due to
which kernel window for 5th-order kernel was incorrectly set to
the same as for 3rd-order kernel. Improved discrete approximation
to the Lanczos kernel. [#198]
- Argument ``scale`` of the ``Resample.add_image()`` method has been deprecated
and split into two arguments: ``iscale`` that is multiplied with each input
image before resampling and ``pixel_scale_ratio`` used to compute appropriate
resampling kernel size. [#203]
- In ``tblot``, the arguments ``pix_ratio``, ``exptime``, and
``output_pixel_shape`` have been deprecated and will be removed in a future
release. Use ``iscale`` instead of the deprecated ``pix_ratio`` and
``exptime``: ``iscale = exptime / pix_scale**2``. The argument
``output_pixel_shape`` is not needed because the output image shape
can be inferred from ``pixmap``. Also, argument ``misval`` has been renamed
to ``fillval`` (to match the name from ``tdriz``, ``Drizzle.resample``).
- Added parameter ``out_img`` to ``blot_image`` to allow caller to pass their
own output image array. [#203]
- Added support for propagating DQ bitfields from input images to the output
image using bitwise-OR. [#206]
- Added a warning when using "sinc" interpolation in ``tblot()``indicating
that its use is not recommended while the interpolation is being investigated
for possible issues. [#213]
2.1.1 (2025-08-14)
==================
- Fix a bug in the Cython code for ``tdriz()`` due to which the code would
raise an error for sub-second exposure times. [#192]
- Fixed a numerical instability in the new and old "boxer" algorithm in #175
(withdrawn release 2.1.0). [#191]
- Fixed a bug in polygon intersection algorithm that could result in
incorrect output when the input polygons have nearly collinear edges. [#194]
2.1.0 (2025-02-20; withdrawn)
=============================
- Restored faster "boxer" overlap for square kernel, improving resample speed;
optimized and simplified related code. [#175]
2.0.2 (unreleased)
==================
- Fixed a bug in ``resample.Drizzle`` due to which initially, before adding
the first image, ``Drizzle.output_img`` is not filled with ``NaN`` when
``fillval`` is either ``INDEF``, ``NAN``, ``None`` *and* the ``Drizzle``
object was initialized with ``out_img=None``. [#170]
- Fixes a crash when ``Drizzle`` is initialized with ``disable_ctx``
set to ``True``. [#180]
2.0.1 (2025-01-28)
==================
- Update ``utils.calc_pixmap`` code to be ready for upcoming changes in GWCS
due to which inverse WCS transformations will respect bounding box by
allowing the caller of ``utils.calc_pixmap`` to disable the bounding box(es)
on both or one of the input WCS when computing the pixel map. [#164]
2.0.0 (2024-10-23)
==================
- Backward incompatible major re-design of API to make the code I/O agnostic.
Removed FITS-specific code. Backward compatibility was
maintained with JWST and Roman pipelines only. [#134]
- Deprecated module ``util``. New software should not import from this
module as it will be removed in a future release. [#134]
- Bug fix: exposure time was undefined when ``in_units`` were not "cps". [#134]
- BUG FIX: ``cdrizzle.tdriz`` signature was out of date. [#134]
- Removed support for the ``'tophat'`` kernel. [#134]
1.15.3 (2024-08-19)
===================
- Fixed return type of ``PyInit_cdrizzle``. [#150]
- Allow ``context`` to be ``None`` in ``tdriz``. [#151]
1.15.2 (2024-06-17)
===================
- build wheels with Numpy 2.0 release candidate [#149]
1.15.1 (2024-03-05)
===================
- Fixed the warning type for the "gaussian", "lanczos2", and "lanczos3" kernels
(``DeprecationWarning`` to ``Warning``). [#141]
1.15.0 (2024-02-16)
===================
- Dropped Python 3.8. [#128]
- Fixed a bug in the pixmap coordinate inversion routine. [#137]
- Deprecated "tophat" kernel which will be remover in the next release. It is
not working correctly and should not be used. [#140]
- Added warnings that "gaussian", "lanczos2", and "lanczos3" kernels are not
flux conserving. [#140]
1.14.4 (2023-11-15)
===================
- N/A
1.14.3 (2023-10-02)
===================
- Disable logging to fix a segfault on some systems. [#119]
1.14.2 (2023-09-16)
===================
- Addressed test failures on big endian architectures. [#116]
1.14.1 (2023-09-16)
===================
- Duplicate re-release of 1.14.0.
1.14.0 (2023-09-15)
===================
- Fixed a bug in how drizzle would compute overlaps between input images and
the output image. Due to this bug large parts of input image data may be
missing in the resampled data when output image size was set by the
caller to be smaller than size needed to hold *all* image data. [#104]
- Replace buggy polygon intersection algorithm with the Sutherland-Hodgman
polygon-clipping algorithm. [#110]
1.13.7 (2023-02-09)
===================
- Fixed a bug in identification of lines in input images that should be skipped
because they would not map to the output image. This bug may result in large
chunks of input image incorrectly missing from the resampled image. [#89]
1.13.6 (2021-08-05)
===================
- Fixed a bug in how interpolation and pixel mapping was reporting invalid
values. This bug may have resulted in resampled images containing all
zeroes. [#85]
1.13.5 (2021-08-04)
===================
- Pin astropy min version to 5.0.4. [#81]
- Fix a bug in the interpolation algorithm used by the 'square' kernel that
resulted in shifts of the resampled image typically by 0.5 pixels compared
to the location indicated by the WCS. [#83]
1.13.4 (2021-12-23)
===================
- drizzle ignores the weight of input image pixels when the weight of the
corresponding output pixel (onto which input pixel flux is to be dropped)
is zero. [#79]
1.13.3 (2021-06-17)
===================
- Remove Cython as a runtime dependency [#72]
1.13.2 (2021-06-16)
===================
- Specify ``oldest-supported-numpy`` in pyproject.toml so that the built C
code has the widest possible compatibility with runtime versions of
``numpy``. [#60]
- Fix a memory corruption issue in ``interpolate_bilinear()`` in
``cdrizzleblot.c`` which could result in segfault. [#66]
- Fix a context image bug when drizzling more than 32 images into a single
output. [#69]
|
