letmaik
Compatibility issue with NumPy 2.0.0
File "/test.py", line 4, in <module>
import rawpy
File "/usr/local/lib/python3.11/site-packages/rawpy/__init__.py", line 5, in <module>
import rawpy._rawpy
File "rawpy/_rawpy.pyx", line 1, in init rawpy._rawpy
ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject
Numpy v2.0.0 rawpy v.0.21.0
Try uninstalling 2.0.0 and installing the latest 1.x version - that fixed it for me:
pip uninstall numpy
pip install numpy==1.26.4
👋 I ran into this issue while working on a rawpy plugin for ImageIO this week, Unfortunately, pinning numpy<2.0.0 is not an option for us, as it would force our downstream libraries into the same pin.
I don't know the intricacies of rawpy's codebase but assuming that it doesn't use any obscure/nitch numpy features this should be solvable by a re-compile and re-release of rawpy for numpy>=2 without any code changes. Is there anything we can do to make this happen? (merging the ImageIO plugin will be blocked until this issue is resolved).
@FirefoxMetzger Releasing a new version built against numpy 2 sounds reasonable. People stuck on numpy 1 will have to pin their rawpy version. Do you want to give this a try in a fork and submit a PR once it's working for you?
I've build rawpy locally against Numpy 2.0 on my Windows machine using python 3.11.
When I run pytest I get the following set of warnings. @letmaik Are these expected or did the switch to numpy v2.0 create them?
================================================================================================= warnings summary ==================================================================================================
rawpy\enhance.py:24
c:\users\sebastian\documents\coding-projects\rawpy\rawpy\enhance.py:24: UserWarning: OpenCV not found, install for faster processing (error: No module named 'cv2')
warnings.warn('OpenCV not found, install for faster processing (error: ' + str(e) + ')')
test/test_basic.py: 16 warnings
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\test\test_basic.py:284: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
if StrictVersion(np.__version__) <= StrictVersion('1.10'):
test/test_basic.py::testFileOpenAndPostProcess
test/test_basic.py::testFileOpenAndPostProcess
test/test_basic.py::testFileOpenAndPostProcess
test/test_basic.py::testFileOpenAndPostProcess
test/test_basic.py::testFoveonFileOpenAndPostProcess
test/test_basic.py::testFoveonFileOpenAndPostProcess
test/test_basic.py::testSRawFileOpenAndPostProcess
test/test_basic.py::testSRawFileOpenAndPostProcess
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\test\test_basic.py:277: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
if im.dtype == np.uint16 and StrictVersion(imageio.__version__) <= StrictVersion('0.5.1'):
test/test_basic.py::testThumbExtractJPEG
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\test\test_basic.py:140: DeprecationWarning: Starting with ImageIO v3 the behavior of this function will switch to that of iio.v3.imread. To keep the current behavior (and make this warning disappear) use `import imageio.v2 as imageio` or call `imageio.v2.imread` directly.
img = imageio.imread(thumb.data)
test/test_basic.py::testBadPixelRepair
test/test_basic.py::testFindBadPixelsNikonD4
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\.venv\Lib\site-packages\skimage\filters\rank\generic.py:332: UserWarning: Bad rank filter performance is expected due to a large number of bins (4096), equivalent to an approximate bitdepth of 12.0.
image, footprint, out, mask, n_bins = _preprocess_input(
test/test_basic.py::testBadPixelRepair
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\.venv\Lib\site-packages\skimage\filters\rank\generic.py:332: UserWarning: Bad rank filter performance is expected due to a large number of bins (4091), equivalent to an approximate bitdepth of 12.0.
image, footprint, out, mask, n_bins = _preprocess_input(
test/test_basic.py::testBadPixelRepair
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\.venv\Lib\site-packages\skimage\filters\rank\generic.py:332: UserWarning: Bad rank filter performance is expected due to a large number of bins (4093), equivalent to an approximate bitdepth of 12.0.
image, footprint, out, mask, n_bins = _preprocess_input(
test/test_basic.py::testBadPixelRepair
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\.venv\Lib\site-packages\skimage\filters\rank\generic.py:332: UserWarning: Bad rank filter performance is expected due to a large number of bins (3936), equivalent to an approximate bitdepth of 11.9.
image, footprint, out, mask, n_bins = _preprocess_input(
test/test_basic.py::testFindBadPixelsNikonD4
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\.venv\Lib\site-packages\skimage\filters\rank\generic.py:332: UserWarning: Bad rank filter performance is expected due to a large number of bins (3900), equivalent to an approximate bitdepth of 11.9.
image, footprint, out, mask, n_bins = _preprocess_input(
test/test_basic.py::testFindBadPixelsNikonD4
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\.venv\Lib\site-packages\skimage\filters\rank\generic.py:332: UserWarning: Bad rank filter performance is expected due to a large number of bins (3892), equivalent to an approximate bitdepth of 11.9.
image, footprint, out, mask, n_bins = _preprocess_input(
test/test_basic.py::testFindBadPixelsNikonD4
C:\Users\Sebastian\Documents\Coding-Projects\rawpy\.venv\Lib\site-packages\skimage\filters\rank\generic.py:332: UserWarning: Bad rank filter performance is expected due to a large number of bins (3922), equivalent to an approximate bitdepth of 11.9.
image, footprint, out, mask, n_bins = _preprocess_input(
-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
========================================================================================= 22 passed, 34 warnings in 30.90s ==========================================================================================