imglyb aims at connecting two worlds that have been seperated for too long:
imglyb uses jpype to access numpy arrays and expose
them to ImgLib2 through
imglib2-imglyb.
This means shared memory between numpy and ImgLib2, i.e. any ImgLib2
algorithm can run on numpy arrays without creating copies of the data!
For example, Python users can now make use of the
BigDataViewer extension to visualize dense volumetric
data.
If you are interested in using imglyb, have a look at the examples folder
and extend the examples as needed!
Note:
NEP 18 has
the potential to improve numpy - imglib interoperability, especially when
converting imglib2 data structures to numpy.
imglyb has been tested on Linux, macOS, and Windows.
The following tools are required:
- Python 3
- Java 8 or 11 JDK (JRE is not enough)
- Apache Maven
If you use conda, these will be installed for you.
conda install -c conda-forge imglybFirst, install the prerequisites above. Then run:
pip install imglybIt is recommended to do this from inside a virtualenv or conda environment, rather than system-wide.
First, install the prerequisites above. Then run:
git clone https://github.com/imglib/imglyb
cd imglyb
pip install -e .It is recommended to do this from inside a virtualenv or conda environment, rather than system-wide.
It is suggested to follow and extend the examples in the examples folder
according to your needs.
Or, for a higher-level way to use imglyb, check out
pyimagej.
AWT and Cocoa do not get along perfectly. In general, the Cocoa event loop
needs to be started before the JVM is loaded. (Thanks to @tpietzsch for
figuring this out!) This requires some macOS specific code, written using
PyObjC, to properly start up and shut down the Cocoa application and start
the Java/Python code within it.
The OSXAWTwrapper.py script included in the imglyb library provides an
example of Cocoa code and can be used to run the imglyb examples. Two
packages from PyObjC are required for this wrapper (pyobjc-core and
pyobjc-framework-cocoa), and they should be installed with imglyb
on macOS.
When running the wrapper, one can either provide the name of the target module
(as if using python -m) or the full path to the target script. So using the
module name, the command to run the "butterfly" script in imglyb-examples
looks like this:
python imglyb/OSXAWTwrapper.py imglyb-examples.butterflyRunning OSXAWTwrapper.py via python -m does not work at this time.
